PROCEEDINGS
THE EFFECTS OF CORPORATE GOVERNANCE ON
CAPITAL STRUCTURE:
A STUDY OF LISTED MANUFACTURING
COMPANIES IN SRI LANKA
WDH De Mel1#, MMLC Gunathilake1
Faculty of Management,
Social Sciences and Humanities, General Sir John Kotelawala Defence University, Sri Lanka
# danishademel@gmail.com
Abstract – Capital structure decisions are vital for any
organization in order to identify the optimal combination
of debt and equity, which will enable to achieve an
organization’s ultimate objective of shareholder wealth
maximization. Good corporate governance practices
prevailing in the organization would ensure that the
managers are acting for the best interest of the owners of
an organization. The study attempts to identify the effects
of corporate governance on capital structure in firms listed
in the manufacturing sector in Sri Lanka. A sample of 34
manufacturing entities were selected out of 38 companies
based on the availability of data for the period of 2012 to
2016. The study employed corporate governance variables
of board size, proportion of non-executive directors,
director ownership and CEO duality. The capital structure
was measured using Total Debt Ratio (TDR) as the
dependent variable. The study also employed the control
variables of firm size and profitability. The variables were
empirically tested using multiple regression analysis. The
results revealed that proportion of non-executive directors,
CEO duality and director ownership tend to have positive
and significant relationship with capital structure while
the association between board size and capital structure
depicts negative, insignificant relationship.
Key Words - Corporate Governance, Manufacturing
Sector, Listed Companies
884 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
I. INTRODUCTION
Global financial crisis and other corporate scandals across
the globe stresses on the vitality of corporate governance
and financial management for an organization in order to
deal with the effects of unexpected crises and uncertainties.
Separation of ownership and control in the businesses as
being identified by Jenson and Meckling (1976), gives rise
to the agency problem within the organizations which
is the prime reasoning behind the need for effective
corporate governance. In an organizational context,
shareholders are being identified as the principals while
the managers are agents acting on behalf of the principals.
Baysinger and Hoskisson (1990), insisted that managers
might be satisfiers rather than striving to maximize returns
such that they tend to play it safe with the aim of survival
without paying much attention on enhancing the value of
the firm. On contrary, the objective of shareholders is to
maximize their wealth giving rise to the agency conflicts
between the said parties.
The countries that have implemented sound corporate
gocvernance practices generally expereinced a vigorous
growth of corporate sector, and grasp more ability in
attarcting capital to lubricate the economy. Ehikioya
(2009), suggested that a well-defined and a functioning
corporate governance system would help a firm to attract
investments, raise funds and strengthen the foundation
for the firm. Good governance focuses not solely on
shareholder wealth maximization but also on reducing the
cost of fund (Sheikh & Wang, 2012).
 ISBN number: 978-955-0301-41-6
Successful selection and use of capital is one of the key
elements of the firms’ financial strategy. The literature
in capital structure began with the seminal work by
Modigliani and Miller (1958), on the irrelevance of capital
structure. Since then, capital structure continues to be a
topic of interest in finance. Modigilani and Miller (1958),
pointed out the fact that in a perfect market, in the absence
of corporate taxes and bankruptcy cost, capital structure is
irrelevant for the value of the firm.
Capital structure decisions are crucial for any business
organization such that the decision is important because
of the need to maximize returns to various organizational
constituencies and also because of the impact such a
decision has on an organization’s ability to deal with its
competitive environment (Aboh, 2007).
Managers are held responsible in selecting the optimum
way of financing for the best interest of the shareholders.
Debt creation enables the managers to effectively bond
their promise to pay out future cash flows. When firms
take strategic decisions via issuing debt, firms should be
responsible to pay interest from the future cash flows.
Meantime increased leverage also involves costs in terms of
bankruptcy cost as well as the agency cost of debt will also
rise (Ajanthan. 2013). Thus, it reveals the importance of
good governance in order to reduce such costs associated
with debt capital.
Claessens, Djankov, Fan and Lang (2002), explained that
by having better corporate governance, firms can benefit
via greater access to external finance. As suggested by Liao,
Mukherjee and Wang (2012), good governance enables the
companies to utilize information in an effective manner
which would lead to a reduced cost of capital and thereby
to make effective and faster capital structure decisions.
Research Issue
Enron, World com, Asian financial crisis and series of
other corporate scandals in United States of America
and elsewhere, reminded the importance of sound
corporate governance mechanisms over past few decades.
Nevertheless, in the recent past, certain Sri Lankan
companies too were victims of the catastrophes of various
corporate governance issues such as, Golden Key scandal.
Thus, corporate governance can be identified as a timely
issue persisting in the country.
The nexus between corporate governance and capital
structure has not been comprehensively explored in
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
developing countries specially in the Sri Lankan context.
Extant literature on corporate governance is heavily
focused on corporate governance and firm performance
even in Sri Lankan context (Achchuthan & Kajananthan,
2013; Velnampy, 2013; Ujunwa, 2012; Ehikioya, 2009)
Therefore, this study attempts to fill the gap in the
literature by taking into consideration the various
corporate governance measures and their effect on the
capital structure decisions of the organizations in the
manufacturing sector, with relevance to the firms listed in
Colombo Stock Exchange (CSE). Therefore, the research
questions are identified as;
1)What are the measures of corporate governance in
Manufacturing Sector of Sri Lanka?
2)What is the impact of measures of corporate governance
on capital structure in listed manufacturing companies in
Sri Lanka?
The research objective is to explore the effects of corporate
governance on corporate capital structure with relevance
to manufacturing entities listed in Sri Lanka.
II. LITERATURE REVIEW
The need for corporate governance within the organizations
stems from agency theory which is a contract under
which the principal engages another person (the agent)
to perform some service on his behalf, which involves
delegating some decision-making authority to the agent
(Jenson & Meckling, 1976). In order for the shareholders
to ensure that managers are working to the best interest
of the shareholders, they have to incur certain costs called
“agency cost” (Vakilifard, Gerayli, Yanesari & Ma’atoofi,
2011). Jenson and Meckling (1976), defined such costs as
the sum of, the monitoring expenditure by the principal,
the bonding expenditure by the agent and the residual
loss. Ujunwa (2012), stated that corporate governance
practices are designed to resolve issues associated with
agency problems and thereby to protect overall interest
of stockholders. Accordingly, adequate monitoring or
mechanisms need to be used in order to reduce the conflict
of interest between shareholders and management, among
shareholders as well as between debt holders and firms
(Fama & Jensen, 1983).
Study of capital structure is given utmost importance
under financial literature. Modigliani and Miller (1958),
argued that under perfect capital markets in the absence of
10TH INTERNATIONAL RESEARCH CONFERENCE 885
PROCEEDINGS
taxes and transaction cost, a firm’s market value and cost
of capital remain invariant to the capital structure changes.
The value of the firm depends on the earnings and risk of
its assets rather than the way in which assets have been
financed. Accordingly, even though debt appears to be
cheaper than equity, when firm increases the debt level,
equity will become riskier and cost of equity rises as a result.
Modigliani and Miller (1963), relaxed the constraints and
introduced corporate tax into the model concluding the
fact that increase of debt level can increase the value of
the firm due to deductibility of interest charges for tax
computation. Among other capital structure theories,
pecking order theory assumes that firms generally prefer
internal finance to external finance (Myers, 2001). This is
based on the assumption that information asymmetries are
relevant only for external financing, which indicates that
managers know more about their companies’ prospects,
risks and values than outside investors. He further states
that if external funds are required for capital investment,
firms will issue the safest security first that is debt before
equity.
Corporate governance has been identified in prior studies
(Berger, Ofek & Yermack., 1997; Wen et al., 2002; Sheikh
& Wang, 2012) to inflence the capital strcuture decisions
of the firms. In Sri lnakan context, Kajananthan, 2012;
Achchuthan, Kajananthan and Sivathasan, 2013; De Mel
and Dassanayake, 2015, have found significant relationship
between coprorate governance and capital structure in the
firms listed in Colomo Stock Exchnage. Claessens et al
(2002), suggests that good corporate governance practices
enable the organizations to have better access to financing
and reduce overall cost of debt by winning the trust of
investors.
Board size has been identified as an important determinant
of corporate governance in various literature (Wen et
al., 2002; Berger et al.,1997) such that an effective board
is essential to the success of an organization. As stated
by Pfeffer (1972), CEO cannot dominate a bigger board
due to the fact that collective strength of its members is
higher and can resist the irrational decisions of a CEO.
On flip side, Amason and Sapienza (1997), argues that
large boards are more prone to conflict among directors
and further according to Lipton and Lorsch (1992),
large boards could be less effective due to the fact that
some directors may free-ride on the the efforts of others.
They also stated that when a board consist of more than
ten members it becomes more inconvenient for them to
express their ideas and opinions.
886 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Board of directors comprises of executive directors
and non-executive directors (NED). Weir, Laing, and
McKnight (2002), defines executive directors as full-time
employees of an organization with clearly defined role and
responsibilities while NED do not serve as the employees
of the company. The composition of the board structure
is an important mechanism because the presence of NED
represents a means of monitoring the actions of executive
directors and of ensuring that they act in accordance with
protecting shareholder interest (Fama, 1980).
when the managers’ ownership of the firm’s stocks increases,
it would help alleviate agency conflict between managers
and owners (Jenson and Meckling,1976). This is because,
a manager who possess large proportion of company
shares, has more incentives to maximize job performance
and thereby to enhance company performance. Further
according to Aboh (2007), directors too should hold
some amount of financial risk as shareholders, which
would give them an incentive to act in the best interests
of shareholders.
CEO duality is another corporate governance variable. It is
a corporate leadership structure, that merger the positions
of chairman of the board of directors and CEO. If CEO
duality exists, “this signals the absence of separation of
decision management and decision control” (Fama &
Jensen, 1983, p. 314). They further stated that since the
chairman has the greatest influence over the actions of
the board, the separation of decision management and
decision control is compromised when the chairman of
the board is also the CEO of the firm. Therefore, two-
tier leadership structure can more effectively control
the agency problems associated with the separation of
ownership and control.
III. METHODOLOGY
The conceptual framework developed for the study
is depicted in Figure 1. It incorporates the corporate
governance variables intended to use in the study in terms
of, board size, non-executive directors (NED), director
ownership, director remuneration and Chief Executive
Officer (CEO) duality. Further capital structure will be
determined in terms of Total Debt Ratio. Study would
also employ the control variables namely, firm size and
profitability.
 ISBN number: 978-955-0301-41-6
Figure 1: Conceptual Framework
Source: Author
In examining the corporate governance on capital strcture
in Sri Lankan listed manufacturing companies, the study
employs panel data methdology. The general form of panel
data model can be illustrated as follows;
Where i denotes the cross-section dimension and t denotes
the time dimension. Yit is the firm i’s capital structure at
time t. Further, α is the intercept of the function while β
represents the corresponding coefficient. Xit contains the
set of explanatory variables in the estimation model. μit is
the disturbance term.
The study employs the following model.
Where,
TDRit = Total debt ratio of firm i at time t
LTDRit = Long term debt ratio of firm i at time t
BS it = Board size of firm i at time t
NEDit = Non-executive directors of firm i at time t
DOWNit = Directors ownership of firm i at time t
CEODUit = CEO duality of firm i at time t
ROAit = Return on assets of firm i at time t
FSIZEit = Size of firm i at time t
β0 = Intercept
β1 – β8 = Coefficients of the explanatory variables
µit = Error term
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
As adopted in many of the prior researches in this area,
this study also employed pooled Ordinary Least Squares
(OLS) method in order to estimate the coefficient. (Sheikh
& Wang, 2012; Abor, 2007; Bodhagi and Ahmdapour,
2010; Masnoon and Rauf, 2010; Abor, 2007; Wen et al.)
Table 1: Definition of variables
Source: Author
In examining the relationship between corporate
governance and capital structure, the organizations that
have been listed on the Colombo Stock Exchange (CSE)
over the period of 5 years spanning from 2012 to 2016
were taken into consideration. The study focuses on
manufacturing sector. Out of 38 listed manufacturing
entities in Sri Lanka, 34 companies were selected based on
the availability of data for years under consideration. The
required data for the study were gathered from secondary
sources. Accordingly, the data and information required
were collected from the annual reports of the listed entities
in the selected sample.
IV. FINDINGS AND DISCUSSION
As depicted in Table 2, the significance value is less than
the significance level, thus there is enough evidence to
state that the model is significant at 5% significant level.
10TH INTERNATIONAL RESEARCH CONFERENCE 887
PROCEEDINGS ISBN number: 978-955-0301-41-6

Therefore, it shows that corporate governance practices Table 5: Multiple regression analysis
contribute significantly to capital structure (F=4.989;
P<0.05). Moreover, as depicted in Table 3, the model
predicts 12.5 % of the variation found. Model Standardized t Sig.
1 Beta Coefficients
Table 2: ANOVA (Constant) 3.815 .000
BS -.082 -1.041 .299
NED .183 2.434 .016
DOWN .227 2.986 .003
CEODU .229 2.967 .003
ROA .082 1.009 .315
FSIZE -.267 -3.077 .002
Source: Author
Source: Author According to the regression results, board size depicts
inverse but insignificant relationship with Total Debt
Table 3: Model Summary Ratio. Other corporate governance variables employed
in the study in terms of, proportion of non-executive
directors, director ownership, CEO duality show positive
and significant relationship with the capital structure.
Among the control variables used in the study, profitability
measured in terms of Return on Assets (ROA), tends
to have positive but insignificant association with Total
In order to determine the presence of multicollinearity Debt Ratio. Firm size shows a negative and significant
among the independent variables used in the research, relationship with the capital structure.
calculation of Tolerance Test and Variance Inflation Factor .
(VIF) was being used. The results are presented in Table 4. Larger boards are expected to have representation of
Accordingly, none of the tolerance level is less than or people with diverse backgrounds and thus anticipated to
equal to 1, and also VIF values are less than 10 Accordingly bring knowledge, wider perspective and intellect to the
it demonstrates that there does not exist the problem of board (Mollah, Farooque and Karim, 2012). Berger et al.,
multi-collinearity. (1997), states that firms with larger board membership
Table 4 have low leverage or debt ratio. Their argument portrays
that, a large board would lead to strong pressure in the
corporate board room to pursue lower leverage as a way of
enhancing firm performance. Findings of Brenni (2014),
also consistent with this, stating that board size has
significant negative relationship with leverage, in the study
based on listed real estate companies in United Kingdom.
Wiwattanakantang (1999), Vakilifard, et al., (2011),
Bodhagi and Ahmadpour (2010) also found a negative
relationship between board size and leverage. As per the
study in Sri Lankan listed manufacturing entities, larger
the board size, lower will be the use of leverage. In prior
studies conduectd in Sri Lankan context, De Mel and
Dassanayke (2015), reveal similar results and moreover,
Source: Author
Kajananthan (2012), Achchuthan et al. (2013), have stated
The results of multiple regression analysis are depicted in that capital structure is not affected significantly by the
Table 5. number of members on the board.

888 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
The relationship between proportion of NED on the board
and capital structure reports positive significant results.
This regression result is consistent with prior studies of
Aboh (2007), Jensen (1986) and Bereger et al. (1997), and
Sheikh and Wang (2012). Aboh (2007), stated that high
proportion of outside directors is believed to be associated
with high leverage positions due to the fact that existence
of NED could lead to better management decisions
helping firms in attracting better resources. This is because
external board members may have sound knowledge and
useful information on financing facilities. The Sri Lankan
Code of Best Practice on Corporate Governance (2013),
mentions that the board should include NED of sufficient
caliber and number, for their views to carry significant
weight in the Board’s decisions. The Board should include
at least two Non-Executive Directors or such number of
NED equivalent to one third of total number of directors,
whichever is higher (Principal A.5.1). This has been
satisfied by listed manufacturing entities in Sri Lanka.
Regression results show an inverse relationship between
director onwership of shres and capital structure in listed
manufacturing entities in Sri Lanka. Friend and Lang
(1988), established a negative relationship between capital
structure and CEO shareholding. Such relationship is
also confirmed by Fosberg (2004), and Sheikh and Wang
(2012). According to Sheikh and Wang (2012), the negative
relationship between managerial ownership and capital
structure indicates that increased managerial ownership
aligns the interests of managers with the interests of
outside shareholders and reduces the role of debt as a tool
to mitigate the agency problems.
There exist a positive and significant relationship between
CEO duality and capital structure. As per the Code of Best
Practice on Corporate Governance (2013), a decision to
combine the posts of chairman and CEO in one person
should be justified and highlighted in the annual report
(principle A.2.1). Fosberg (2004), found a positive but
insignificant relationship between CEO duality and
amount of debt being used. This was further confirmed
in subsequent literature by, Aboh (2007) and Bokpin and
Arco (2009), based on the studies relating to Ghanaian
listed entities.
According to Aboh (2007), such positive relationship
is important in avoiding conflict between CEO and the
board chairman, where the two personalities are different.
Hence that would enable the CEO in a one-tier system to
pursue an effective debt strategy based on the advice of the
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
board. Moreover, Vakilifard et al. (2011), also suggested a
significant and positive relationship between CEO duality
and capital structure.
Among the control variables used in the study, profitablity
measured in terms of Return on Assets (ROA), shows a
positive but insgnificant relationship with Total Debt Ratio
(TDR). Brenni (2014), and Bodaghi and Ahmadpour
(2010), found a positive relationship with profitability
and leverage which is supported by trade-off theory
disagreement with the pecking order theory where highly
profitable firms will first utilize retained earnings therefore
suggesting lower external financing needs. For profitable
manufacturing firms, obtaining better negotiations for
debt financing would be easier in order to gain tax shield
benefits. As far as firm size is concerned, the larger the
firm size lower will be the Total Debt Ratio with respect to
listed manufacturing entities in Sri Lanka.
V. Conclusion
Capital structure decision is one of the crucial decisions
taken in an organization. The study attempted to identify
the effects of corporate governance on capital structure of
manufacturing entities listed in Colombo Stock Exchange.
Empirical results indicate that there exists a significant
relationship between corporate governance and capital
structure in listed manufacturing companies in Sri Lanka.
The corporate governance variable of, board size, reported
a negative, insignificant relationship with capital structure,
measured in terms of Total Debt Ratio. Other corporate
governance variables used in the study, proportion of
non-executive directors, CEO Duality, director ownership
showed positive, significant association with Total Debt
Ratio. Study also employed the control variables of firm
size and profitability. Profitability measured in terms of
Return on Assets (ROA) reported a positive, insignificant
association with Total Debt Ratio while firm size is
negatively correlated with capital structure.
The empirical evidence documented in the study tends
to have various implications for directors and managers,
decisions makers, regulators and policymakers, investors
and researchers. The findings suggest that good governance
is vital when firms making decisions on the method of
financing and reminds the importance of adhering to
corporate governance guidelines.
Further, investors will be facilitated in terms of identifying
entities that have established good governance practices.
10TH INTERNATIONAL RESEARCH CONFERENCE 889
PROCEEDINGS
Accordingly, investors may undertake their investment
decisions with confidence.
Future research can be carried out with taking ownership
structure into consideration in order to determine its
impact on capital structure decisions of Sri Lankan
listed firms. Ownership structure is also a vital aspect of
corporate governance practices such that prior researches
too have incorporated ownership patterns (Shiekh &
Wang, 2012; Bodaghi & Ahmadpour, 2010). However, in
Sri Lankan context, obtaining information with this regard
requires substantial analysis due to ultimate ownership.
Hence future research is opened with respect to detailed
analysis of ownership structure in order to determine its
impact on capital structure decisions of Sri Lankan listed
firms.
The scope of the study was limited to corporate governance
practices and their effects on capital structure merely to
public quoted entities in the manufacturing sector of CSE.
Nonetheless, evaluation of corporate governance practices
with respect to other sectors can be analyzed in further
research.
REFERENCE
Aboh, J. (2007). Corporate governance and financing
decisions of Ghanaian listed firms. 7(1), 83 - 92.
Achchuthan, S., & Kajananthan, R. (2013). Corporate
Governance practices and Firm Performance: Evidence
from Sri Lanka. European Journal of Business and
Management, 5(1), 19-26.
Achchuthan., S., Kajananthan., R., & N.Sivathaasan.
(2013). Corporate Governance Practices and Capital
Structure: A Case in Sri Lanka. International Journal of
Business and Management, 8(31), 114-125.
Ajanthan, A. (2013). Impact of Corporate Governance
Practices on Firm Capital Structure and Profitability: A
Study of Selected Hotels and Restaurant Companies in
Sri Lanka. Research Journal of Finance and Accounting,
4(10).
Amason, A. C., & Sapienza, H. J. (1997). The effects of
top management team size and interaction norms on
cognitive and effective conflict. Journal of Management,
23(4), 495-516.
890 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Baysinger, B., & Hoskisson, R. (1990). The Composition
of Board of Directors and Strategic Control: Effects on
Corporate Strategy. Academy of Management Revies,
71(4), 65-443.
Berger, P., & Ofek, E. Y. (1997). Managerial entrenchment
and capital structure decisions. Journal of Finance, 52(4),
1411-38.
Bodaghi, A., & Ahmadpour, A. (2010). The Effect of
Corporate Governance and Ownership Structure on
Capital Structure of Iranian Listed Companies. 89-96.
Bokpin, G., & Arco, A. (2009). Ownership structure,
corporate governance and capital strcuture decisions
of firms: Empirical evidence frim Ghana. Studies in
Economics and Finance, 26(4), 246-256.
Brenni, P. (2014). Corporate Governance and Capital
Structure Decisions of UK Listed Real Estate Companies.
Research Journal of Finance and Accounting, 5(1), 42-54.
Claessens, S., Djankov, S., Fan, J. P., & Lang, L. H. (2002).
Distangling the Incentives and Entrenchment Effects of
Large Shareholders. The Journal of Finance, 57(6).
De Mel, W. D., & Dassanayake, D. M. (2015). Efects of
corporate governance on capital structure: Evidence from
listed companies in Sri Lanka. The 10th International
Research Conference on Management and Finance, (pp.
228 - 242).
Ehikioya, B. I. (2009). Corporate governance structure
and firm. 9(3), 231-243.
Fama, E. F. (1980). Agency problems and the theory of the
firms. The Journal of Political Economy, 88(2), 288-307.
Fama, E., & Jensen, M. (1983). The separation of ownership
and control. Journal of Law and Economics, 26(2), 301-
325.
Fosberg, R. (2004). Agency problems and debt financing:
Leadership structure effects. Corporate Governance, 4(1),
31-38.
Friend, I., & Lang, L. (1988). An empirical test of the
impact of managerial self-interest on corporate capital
structure. The Journal of Finance, 43(2), 271-281.
Hsiao, C. (1986). Analysis of Panel Data.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Hussainey, K., & Aljifri, K. (2012). Corporate governance
mechanisms and capital structure in UAE. Journal of
Applied Accounting Research, 13(2), 145-160.
Jenson, M., & Meckling, W. (1976). Theory of the Firm:
Managerial Behavior, Agency Costs and Ownership
Structure. Journal of Financial Economics, 3(4), 305-360.
Kajananthan, R. (2012). Effect of corporate governance
on capital structure: Case of the Sri Lankan listed
manufacturing companies. International Refereed
Research Journal, 3(4), 63-71.
Liao, L., Mukuhrjee, T., & Wang, W. (2012). Does
Corporate Governance impact on Capital Structure
Adjustments.
Lipton, M., & Lorsch, J. (1992). A Modest Proposal for
Improved Corporate Governance. The Business Lawyer,
48, 59-77.
Masnoon, M., & Rauf, M. (2013). Impact of corporate
governance on capital strcuture: A study of KSE listed
firms. Global Management Journal for Academic &
Corporate Studies, 3(1),
Modigliani, F., & Miller, M. (1958). The cost of capital,
corporation finance and the theory of finance. American
Economic Review, 48(1), 261-97.
Ujunwa, A. (2012). Board characteristics and the financial
performance of Nigerian quoted firms. 12(5), 656-674.
Vakilifard, H., Gerayli, M., Yanesari, A., & Ma’atoofi,
A. (2011). Effect of Corporate Governance on Capital
Structure: Case of the Iranian Listed Firms. European
Journal of Economics, Finance And Administrative
Sciences, 165-172.
Velnampy, T. (2013). Corporate Governance and Firm
Performance: A Study of Sri Lankan Manufacturing
Companies. Journal of Economics and Sustainable
Development, 4(3), 228-235.
Weir, C., Laing, D., & McKnight, P. J. (2002). Internal and
external governance mechanisms: Their impact on the
performance of large UK public companies. Journal of
Business Finance and Accounting, 29(5 & 6), 579-609.
Wen, Y., Rwegasiri, K., & Bilderbeek, J. (2002). Corporate
governance and capital structure decisions of Chinese
listed firms. Corporate Governnace: An International
Review, 10(2), 75-83.
Wiwattanakantang, Y. (1999). An empirical study on the
determinants of the capital structure of Thai firms. Pacific-
Basin Finance Journal, 7, 371-403.

Mollah, S., Farooque, O., & Karim, W. (2012). Ownership

structure, corporate governance and firm performance.
29(4), 301-219.

Myers, S. (2001). Capital Structure. The Journal of

Economic Perspectives, 15(2), 81-102.

Pfeffer, J. (1972). Size and composition of board of directors:

The organization and its environment. Administrative
Science Quarterly, 21, 218-228.

Sheikh, N. A., & Wang, Z. (2012). Effects of corporate

governance on capital structure: empirical evidence from
Pakistan. 12(5), 629-641.

(1992). The Cadbury Report : Report of the Committee on

the Financial Aspects of CorporateGovernance.

The Institute of Chartered Accountants of Sri Lanka;

Securities and Exchange Commission of Sri Lanka. (2013).
Code of Best Practice on Corporate Governance 2013.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 891
PROCEEDINGS ISBN number: 978-955-0301-41-6

DEVELOPING AN ELECTRONIC
ACCIDENT REPORTING SYSTEM TO
SRI LANKAN APPAREL INDUSTRY

PMNP Wijerathne1, and NHC Manjula1

1
Department of Building Economics, University of Moratuwa, Sri Lanka
# newphoinex@gmail.com

Abstract - Under-reporting of occupational accidents Keywords - Occupational Safety and Health,Accident

is becoming a major concern where management of Under-reporting, Electronic Accident Reporting System
occupational safety and health is considered. Drawbacks
of the current accident reporting system have been
identified as the main cause of accident under-reporting in
I. INTRODUCTION
the industrial sector. Thus, this study is aimed to develop
Occupational Safety and Health (OSH) is one of the major
an electronic accident reporting system and a mobile app
concerns of any organization that seeks to sustain. Apparel
to overcome the prevailing issues of the manual system
industry is one of the mass producing industries in Sri
and thereby reduce the under-reporting of occupational
Lanka (Kapuge & Smith, 2007). It is also a sector with
accidents in the apparel industry. Four objectives were
higher percentage of OSH issues (Madurawala, 2013).
formed to study the accidents/incidents,the current
According to the Factories Ordinance (1942), in Sri Lanka,
accident reporting system in the Sri Lankan apparel
all the industrial accidents are essential to be reported to
industry, the requirement of an electronic accident
the Department of Labour. About 15-25% of accidents has
reporting system to the Sri Lankan apparel industry and
been recognized to be under-reported in manufacturing
to develop an electronic accident reporting system to the
sector(Beaumont, 2007). This under-reporting can be
Sri Lankan apparel industry. The study was structured in
problematic when it comes to conducting in-depth
several steps. In-depth knowledge was gained regarding
studieson OSH (Psarros, Skjong&Eide, 2009). The
the research stream which was sorted upon the degree
reason for the accidents/incidents to be under reported
of relevance to the study. Semi-structured interviews
is identified as the complexity of the current accident
were conducted with industry experts to identify the
reporting system (Probst, 2013).
requirement of an electronic accident reporting system,
issues of the current system and needed improvements. The
Findings of the interviews revealed that under-reporting
of accidents mostly happen due to the inefficiency of the II. ACCIDENT REPORTING
manual accident reporting system currently used in the SYSTEMS
industry. The existing system was found to be discouraging
and rigid. Suggestions were provided by the professionals An accident reporting system ensures all workplace
and OSH experts to improve the current system. Once the parties are aware of how to report a work-related injury to
system and the app were developed, they were validated by the authorities and to the appropriate persons designated
an expert survey and a test run respectively. by the employer. Reporting of Injuries, Diseases and
Dangerous Occurrences Regulations 1995 (RIDDOR)

892 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
requires the reporting of work-related accidents, diseases
and dangerous occurrences. The information enables the
enforcing authorities to identify where and how risks arise
and to investigate serious accidents.There are two types of
accident reporting systems in general industry practice.
They are manual accident reporting system and electronic
accident reporting system.
A.Manual accident reporting system
According to New Jersey Department of Education
(2008), if accident need to report, injured person
should associate with a vocational – technical education
program, courses, career orientation course, or structured
learning experience and with the treatment provided by
a licensed physician. Further, considering the reporting
process, when accident reporting, it requires manually
filled accident form with following details as country,
district, school, incident/structured learning experiences ,
description of accident, date and time of accident, narrative
description of accident, corrective action taken and report
completed by. Moreover, Department of Health (2013)
stated that manual accident reporting requirements as, it
is need to respond to the most senior member of staff and
they require to complete the accident form by authorised
officer or SRS program manager by submitting details
of who was involved, how, where and when the accident
occurred, who was injured and what action is being taken
in response to the accident.Consequently, manual accident
reporting system consists of unique forms which need to
be filled manually by authorized person and process of
manual reporting is complex in nature.
B. Electronic accident reporting system
According to Safety, Health and Environment Unit
(2012), through the electronic accident reporting system
accidentcan be reported instantly. Moreover Reporting of
Injuries, Diseases and Dangerous Occurrences Regulation
(RIDDOR Regulation) (2013), mentioned accident
records such as, broken bones or dislocation, amputation,
injury from electric shock or serious burns, loss of sight
or chemical burn to the eye, crush injuries affecting the
head or torso, injury as a result of confined space work
that results in hypothermia and loss of consciousness
due to head injuries, chemicals, toxins, infected materials
or asphyxia can send through online media. University
Collage Landon maintain online incident reporting system
and it is system that have complete detail reporting and as
well as the PDF report of the entered details (SafetyNET
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
online incident reporting, 2011). Moreover, University
week (2004) mentioned that university employees can
access to the system through University Employee
Health and Safety Department web site. Simple Accident
Reporting App (SARA) altremis App Venture Ltd was
developed as iPhone app to report incidents, near –
misses and accident. SARA mainly focused on immediate
reporting without paper work and save employees from
accidents (Altermis Ltd, 2012).According to Department
of Health and Safety, employees, students, visitors and
contractors within the premises should involve to the
accident reporting to Department of Health and Safety
(Head of Health and Safety, 2014). Walsh and Antony
(2007), mentioned that electronic accident reporting was
most suitable report accident comparing quality and cost
of accident reporting systems.
III. SRI LANKAN PRACTICE
OF ACCIDENT REPORTING
Considering of accident reporting system in Sri Lankan
context still it is manual reporting system. Accident
reporting it is govern under Factories Ordinance No.45
of 1942 and under that General Register should be kept
in every factory as mentioned in Section 92 of Factories
Ordinance No.45 of 1942 (Factories Ordinance No.45 of
1942, 1942). According to Factories (Amendment) Law
No 12 of 1976 Sec.61 it is mentioned that written notice
is to be sent to the District Factory Inspector Engineer
by using Form 10 and such an occurrence it should be
inform by telephone immediately and it should be confirm
by transmitting a Police Radio message to the D.F.I.E.
(Factories Ordinance No.45 of 1942, 1942).
Accident reporting process in Sri Lanka
10TH INTERNATIONAL RESEARCH CONFERENCE 893
PROCEEDINGS ISBN number: 978-955-0301-41-6

Generally, Sri Lankan apparel industry follows Factories V. DEVELOPMENT OF

Ordinance requirements. According to District Factories
Inspector Engineers (DFIE) they inspect whether industry ELECTRONIC ACCIDENT
follow the Factories Ordinance requirements of accident
reporting. In industry side they are legally govern to report
REPORTING SYSTEM
accidents and apart from it organizations has individual
According to the experts, reporting system should be
norms to collect accident records. According to industry
user friendly to report accidents and it should feature
experts, there are huge financial and non-financial cost of
user friendly command buttons, easy to collect accident
accidents it was significant reason to take more attention
records, instantly notification facility and it should direct
on accident details. Further to DFIEs, large amount of
communication with industry. Industry require the
paper work, less of staff, complexity of current system and
system with facilities of user friendly functional features,
human errors are leading difficulties of current accident
quick data entering facility, instantly notification facility
reporting system. Large amount of paper work, high paper
and easily documents sending facility.
waste, difficulties of storing documents, complexity of
current accident reporting system were misled to accident
When considering mobile app it should facilitate to manage
under reporting.
even lower level employees in industry and to verify the
feasibility of use mobile app in lower level employee a
IV. METHODOLOGY simple questioner was developed. Consider of the findings
of questioner survey large amount of employee aware with
Through the literature review, it was discovered that an the current organization accident reporting system but
under reporting condition has occurred in reporting work drastic of employees were expressed under satisfaction
place accidents and incidents in apparel industry of Sri of the current accident reporting system. Other than
Lanka. The main reason was found to be the complications writing description most of the employee convenient with
which are present in the current accident reporting the functional features of mobile phone and majority of
system. Under 3 phases, data were collected. Pilot study the employee stated that mobile app is appropriate to the
was carried out among the two District Factory Inspector accident reporting.
Engineers (DFIEs) by using semi structured interviews. 3
case studies were conducted having one safety executive, Considering above factors developed accident reporting
an accident record keeper and lower level employees for system naming “eARS” web base system and “eR” mobile
each case. Semi structured interviews were conducted app. “eARS” system were arrange as follow,
with safety executives and accident record keepers, mainly
under the following headings, current accident reporting
system, difficulties of manual accident reporting system
and requirement of electronic accident reporting system
to collect further information. A questionnaire survey
was conducted with lower level employees to identify
manageable level of functional features of mobile app
within lower level employees. Content analysis, cross case
analysis and analysis by using charts used to analyse the
collected data and for content analysis N-Vivo software
was used. For system validation industry experts were
selected User – friendliness of functional features of
developed mobile app was validated by a questionnaire
survey among lower level employees.
Figure 2 Login interface of eARS

894 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6 PROCEEDINGS

When considering mobile app it should facilitate to manage

even lower level employees in industry and to verify the
feasibility of use mobile app in lower level employee a
simple questioner was developed. Consider of the findings
of questioner survey large amount of employee aware with
the current organization accident reporting system but
only few of the employees expressed satisfaction towards
the current accident reporting system. Other than writing
description most of the employee convenient with the
functional features of mobile phone and majority of the
employee stated that mobile app is appropriate to the
accident reporting. The interfaces of eR are as follows.

Figure 3 Facilities of eARS

Figure 7 First interface of eR app

Figure 4 Interface to indicate injured body part

Figure 5 Further attachment

Figure 8 Details of accident

Figure 6 Final interface of the system

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 895
PROCEEDINGS
Figure 9 Details of reporter and photo evidence
According to this study, validation of the system was
conducted in two steps as follows,
Step 01
Fundamental intention of this step is to check the
comprehensiveness of the developed system and app
and there conducted expert survey with the two District
Factory Inspector Engineers.
Step 02
This step was arranged to check the feasibility user
friendliness of the developed app. For the validation
survey there were selected 30 lower level employees in
sewing department, maintenance department and security
department. Considering the opinions, comments and
responses received for the developed “eR” App and the
eARS it was evident that these contribute to mitigate
the existing issues related to accident reporting in the
industry. According to respondents, level of satisfaction of
functional features of “eR” app can be elaborate as follows,
(rate 1 is least and 5 is the highest)
Figure 10 Satisfaction level of functional
features in eR app
896 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
VI. CONCLUSIONS
To create an accident free environment in industry, it is
very essential to have better accident reporting system to
prevent future accidents. Complexity of current accident
reporting process, large amount of paper work are most
significant causes to accident under – reporting. In world
context many advanced methods are used in accident/
incident reporting minimizing the complexity of accident
reporting process. Therefore, this research was carried out
to develop an electronic accident reporting system to Sri
Lankan apparel industry with considering requirements
of Sri Lankan apparel industry. “eARS” web base accident
reporting system and “eR” mobile app to were developed to
Sri Lankan apparel industry while achieving the research
aim and the objectives.
REFERENCES
Beaumont, P. B., 2007. An analysis of the problem of
industrial accidents in Britain. International Journal of
Manpower, 1(1), pp.28-32.
Department of Health and Safety. (2014). Head of Health
and Safety. Department of Health and Safety.
Department of Labour, Sri Lanka. (1942). Factories
Ordinance No. 45 of 1942. Department of Labour.
Madurawela, S., (2013).Why Health and Safety in the Work
Place is an Important Economc issue.Mirror Business.
Malmadana Kapuge, A. and Smith, M., 2007. Management
practices and performance reporting in the Sri Lankan
apparel sector. Managerial Auditing Journal, 22(3),
pp.303-318.
Probst, T. M., & Estrada, A. X., 2010. Accident under-
reporting among employees: Testing the moderating
influence of psychological safety climate and supervisor
enforcement of safety practices. Accident Analysis &
Prevention, 42(5), pp.1438-1444.
Psarros, G., Skjong, R., & Eide, M. S., 2009. The acceptability
of maritime security risk. Journal of Transportation
Security, 2(4), pp.149-163.
UCL Estate and facilities deivision.(2011). SafetyNET.
Estate and Facilities Division.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

ACKNOWLEDGMENT valuable time to make this dissertation a reality. Each and

This dissertation is not a result of a single person’s work
but an outcome of dedication and support of many to
whom I must take a minute to thank to, all the pillar of
guidance, the utmost support, the profound advisers and
every one of them are reminded with heartiest gratitude
and thankful thoughts who showed the path to make this
dream come true.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 897
PROCEEDINGS ISBN number: 978-955-0301-41-6

DISASTER MITIGATION STRATEGIES

IN URBAN DEVELOPMENT:
WITH SPECIAL REFERENCE
TO KADUWELA AREA
PBWS Malshani1#, ID Wattuhewa1
1Department of Management & Finance, Faculty of Management, Social Sciences & Humanities,
Sir John Kotelawala Defence University, Sri Lanka
# wathsala199223@gmail.com

Abstract - Disasters are unavoidable occurrences that Although as the other countries in the world, Sri Lanka
cause damage to the life style of, not only humans but is similarly prone to particular, different kind of natural
also animals. Wherever the country is and whatever the disasters which are non-avoidable, with increasing losses
location of the country is in the world, it is the truth and for life and property in the past few decades.
law that they have to face disasters in many ways, not only
natural disasters but also man-made disasters. Out of Out of natural disasters most urban areas are vulnerable to
natural disasters and man-made disasters, natural disasters the flood disaster and comparing the rural and the urban
are the most important as they cannot be avoidable. areas, urban areas are the highly vulnerable to the flood
This study examines the impact of structural mitigation disasters.
strategies and non-structural mitigation strategies on
the effectiveness of flood mitigation strategies in urban
development. In particular, the study focuses flood
disasters in Sri Lankan urban cities with special reference
to the Kaduwela geographic area. The study employed a
questionnaire and unstructured interviews for the data
collection. The study finds that structural flood mitigation
and non-structural flood mitigation strategies influence
the effectiveness of Flood mitigation strategies in urban
development. The study also discusses policy implications
in mitigating natural disasters.

Keywords - flood, structural mitigation, non-structural

mitigation

I. INTRODUCTION
Sri Lanka is a country which has a very significant and
remarkable location, which is considered less disaster Figure: 1 - number of people affected by different
prone area (hazards and disasters are very rare) comparing disasters from 2010 to 2015
to other countries and Island in the Indian Ocean.

898 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Figure: 2-Areas where people affected by
flood as at 20.05.2016
The above two figures shows that flood is the most
vulnerable disaster and urban areas are affected more
than the rural areas by flood disaster. The current study
addressed the problem, if there are several mitigation
projects to avoid the flood disaster in urban areas, the
risk and the vulnerability does not reduced to the level
that they expected, it is important to study the reasons
behind it as here are number of mitigation strategies are in
currently. The main objective of this study is to identify the
current flood mitigation strategies in Kaduwela area since
current study narrow down to the Kaduwela area.
A .Research Objectives
The primary objective of the research is to identify the
disaster mitigation strategies in Kaduwela to reduce the
risk and the impact of flood disaster.
Secondary objectives are, firstly, analyse the current
disaster mitigation strategies that implemented by the
Disaster Management Centre to reduce the risk of flood
disaster in Kaduwela. Secondly, to find out reasons, why
still Kaduwela gets more damage because of the flood
disaster, if there are disaster mitigation projects and finally
to provide, new and technological awareness programs
and mitigation strategies to overcome the flood disaster.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE
PROCEEDINGS
B. Significance of the Research
This study would be significant in three different aspects
namely, theoretical, empirical and stakeholder significant.
Theoretically Flooding in the cities and the towns is a
current phenomenon caused by increasing occurrence
of heavy rainfall in the short period of time and this is
an incident which happened every year, means that it
must have the better solution than the current solutions,
to minimize the risk and the vulnerability for the flood
disaster. Theoretically getting the idea of those most
important two concepts, flood disaster and the mitigation
strategies will help the enhance the knowledge about this
most familiar disaster and as well as to overcome this
disaster effectively by finding and recognizing the falls of
previous mitigation projects in Kaduwela faces this bitter
experience every year.
Till 2005 there was a Disaster Management Act, the
major document which provided the legal basis as well
as, there are number of mitigation projects like, Sri Lanka
Multi Hazard Disaster Mitigation Project (SLUMHDMP)
in 1997, The Greater Colombo Flood Control and
Environmental Improvement Project, after severe flooding
of the Capital City of Colombo on 4th June 1992 receiving
493mm of rainfall within 12 hours which is equal tone-
eight of the city annual rainfall (Karunaratne 2001) to
reduce the damage and risk of this natural disaster, but
the people who live in Kaduwela still face this major
natural disaster yearly. Empirically the study found out
the reasons why flood mitigation projects and strategies
not successful and what are the suitable and new strategies
that can implement to overcome this disaster.
The stakeholders significance will be immerged from this
study can be discussed as follows, the facts which is find
through this study will be important to the Ministry of
disaster Management specially when the other Authorities
(like ministry of Rehabilitation and reconstruction,
ministry of Urban Development, ministry of Irrigation,
ministry of Defence etc.) are going to implement any
project with the help of Disaster Management centre as
it can provide the information about flood prone areas,
which area damaged by the flood mostly etc. and what are
the mitigation strategies, how they implement etc. When it
comes to people in the Kaduwela, this will help to prevent
and reduce the impact of flood disaster as well as risk and
the vulnerability in this urban area.
It will improve the quality of the life in the Kaduwela
people and make the way to sustainable life.
899
PROCEEDINGS
II.METHODOLOGY
The research was conducted using 100 of people as the
sample which is selected as per the random sampling
method and research strategy taken in this research is
survey method as the data collection was done using
structured questionnaire method, unstructured interview
method and 5 Point Likert Scale method. The research
choice is mixed method as the study include of both
quantitative and qualitative mechanism.
The independent variables which are structured mitigation
strategies (were measured using 5 Point Likert scale
method) and non-structured mitigation strategies (were
measured using 5 Point Likert scale method) evaluated
against the dependent variable which is effectiveness
of mitigation strategies in urban development (were
measured using structured questionnaire and unstructured
interview method).
Figure: 3. Conceptual framework of the research
Once the conceptual framework was developed the
factors which comes under each independent factor
was identified through literature. Under these structural
mitigation strategies, dams, drainage system, levees and
floodwalls, flood barriers and flood gates were identified.
Under non-structural mitigation strategies, preparedness,
government and regulations, policy institutional mandates
and institutional development, flood vulnerability and risk
reduction, flood early warning system, land use planning
and management and flood proofing and building codes.
Under the sample statistics analysis, such as sample
adequacy, sample reliability, goodness of the fitness of
the data set and significance of the data set and sample
900 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
validity were test using SPSS (version 22). The Kaiser-
Meyer-Olkin method was used to test the sapling
adequacy of the study. The reliability of the study was
analysed through Cronbach’s Alpha model which shows
how well the sample data set can generalize and predict
the population information was very well adequate. The
independent variables were compared with the dependent
variable and identified goodness of fit of the model by the
Model Fit. ANOVA analysis is used to test whether there
is statistically significant difference between dependent
variable and independent variables of the sample.
III. RESULTS
Regarding the sample statistics, the sample adequacy
ratio was 0.658 at 0.00 significant levels which is greater
than the acceptable value of 0.5 thus the sample is well
adequate. According to the sample reliability sample data
set can generalize and predict the population information
was very well adequate. The independent variables were
compared with the relationship between the dependent
variable and identified goodness of fit of the model.
The R Square value is 0.721 which interprets that the
independent variables have 71 precents of effect in change
of the dependent variable.
Considering the dependant variable which is the
effectiveness of the mitigation strategies in urban
development and independent variables which are
structural mitigation strategies and non-structural
mitigation strategies, it was revealed that there is an effect
of structural mitigation strategies and non-structural
mitigation strategies over the effectiveness of the
mitigation strategies in urban development. Through the
data analysis it was revealed.
Figure 4: Effectiveness of structural mitigation
strategies
 ISBN number: 978-955-0301-41-6
The above figure shows that more than half of the people
of the sample agreed that structural mitigation strategies
are effective mitigation strategies for urban development.
Figure 5: Effectiveness of drainage systems
Under the structural mitigation strategies, drainage
systems are the most effective structural mitigation strategy
comparing to other structural mitigation strategies which
were mentioned in the methodology. The above figure
clearly shows it.
Considering the non-structural mitigation strategies
nearly half of the people responded as agree that non-
structural mitigation strategies are effective mitigation
strategies for urban development.
Figure 6: Effectiveness of the non-structural
mitigation strategies
Under the non-structural mitigation strategies, land use
planning and flood management is the most effective non-
structural mitigation strategy comparing to other non-
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE
PROCEEDINGS
structural mitigation strategies which were mentioned in
the methodology. The below figure shows it clearly.
Figure 7: Effectiveness of land use planning
and flood management
Comparing structural mitigation strategies and non-
structural mitigation strategies out of 100 people
sample there were 74% of the people who agreed
structural mitigation strategies are effective for urban
development and 59% of the people who agreed non-
structural mitigation strategies are effective for urban
development.
IV.DISCUSSION
This study is about Disaster Mitigation Strategies in
Urban Development with special reference to the
Kaduwela area, regarding the mostly affected flood
disaster. The main purpose of this study is to identify
the effectiveness of the current flood mitigation
strategies in Kaduwela area.
The researcher recognized that there were two independent
variables and one dependent variable regarding to this
study and they were structural mitigation strategies and
non-structural mitigation strategies as independent
variables and dependent variable was effectiveness of flood
mitigation strategies in urban development.
According to the data analysis with reference to the
data collection it was revealed that which areas must be
developed further, using structural mitigation strategies as
well as non-structural mitigation strategies.
Recommendations are the key factor which revealed
the further development of the areas under the areas
of, recommendations for communities, authorities and
admirations as well as further researchers.
901
PROCEEDINGS
A. Recommendations for communities, authorities and
admirations.
An Integrated Post Flood Assessment discovered that
the medium-term economic damages to industries and
commerce are about ten times higher than that of physical
damage. The majority of small-scale entrepreneurs were
affected extremely due to absence of in-house capabilities
and resources to prepare Disaster Management plans,
business continuity plans and recovery plans. Improved
co-ordination and facilitation between banks, industries
and businesses (especially the SMI sector) could be
beneficial. This area may also require introduction of
a number of policy interventions at the level of Central
Bank and Treasury.
Most of the flood disasters occur because of low addressing
of land use planning and flood management. So it is
important to provide a science base for addressing land use
planning if the people are going to build their construction
in flood prone areas. Land use approval processes in local
authorities, availability of data for such decision making
and better coordination between agencies will help to
better land use planning and flood management.
Drainage systems are the most effective structural flood
mitigation strategies for urban development and it is
not economical to expand the existing drainage systems
to accommodate to extreme flood events. Instead, there
should be multiple mechanisms that can be put in place
to improve drainage (clearing, regular maintenance, etc.),
while enhancing the preservation capacity within the
premises (constructed wetlands, ponds, buffer storages,
rain water harvesting, infiltration improvements etc.).
Though the disasters cannot be avoided or prevented, it is
important to be proactive, by providing more accurate and
timely weather forecasts and warnings to general public.
Most of the disasters can identified through the Early
Warning Systems, as it is provide the information about
the disaster before it happen. So it is important to improve
the Flood Early Warning System to Kaduwela areas who
got affected frequently. Most of the people recommended
this as an effective flood mitigation strategy. The Sri
Lankan government should pay much attention to this
with the help of DMC of Sri Lanka.
Most of the disasters happened because of the lack of
knowledge of Rules and Regulations. So the government
and the relevant authorities must organize programs to
improve their knowledge regarding how they behave when
902 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
they build constructions in flood prone areas, what are the
mitigation strategies that they must addressed. It is better
if the authorized people can visit affected area and share
the knowledge with the people as they recommended it
with kind heart, and knowledge sharing programs should
be attractive and effective for the every person who was in
different level of knowledge.
When the researcher visited to most affected areas in
Kaduwela, identified that there was lack of constructions
and barriers for the flood disaster like flood gates, flood
barriers and floodwalls for the relevant areas. So it
is responsibility of the government and the DMC to
construct flood barriers, floodwalls and flood gates.
B. Recommendation for further researches.
The researcher limited the current research for only
one step of the Disaster Management Cycle and that is
mitigation, but there are steps Preparedness, Response
and Recovery for disasters. As well as the current research
limited only for one disaster and that is flood, but there
is considerable number of disasters which needed to be
addressed. The researcher focused only one area through
this current study and that is Kaduwela, but there are
areas which vulnerable for the various disasters most of
the time like Rathnapura (flood), Badulla (landslides),
hambanthota and Monaragala (drought) etc.
V. CONCLUSION
Disaster mitigation is vast area to study and it is important
to study this subject as it helps to the sustainable
development of the county, reducing the vulnerability of
disasters. As a country the government and the related
authorities like Disaster Management Centre, Red Cross
and NGOs should aware of the mitigation strategies for
every disaster. Considering the disasters the most affected
disaster is flood and it is important to study the mitigation
strategies of flood disaster. So this current study focused
on the flood mitigation strategies in urban development
with special reference to the Kaduwela area.
The researcher recognized the current mitigation strategies
for flood disaster as structural mitigation strategies and
non-structural mitigation strategies which were the
independent variables in the conceptual framework. As
well as based on the secondary objectives the researcher
identified why Kaduwla area still got affected by the flood
disaster heavily if there were flood mitigation strategies.
 ISBN number: 978-955-0301-41-6
Through the recommendations of the people the
researcher identified what are the most effective and what
are the useful strategies that were not in the field.
ACKNOWLEDGMENT
The researcher would like to express sincere gratitude to
everyone who supported through the research and special
thanks goes to author’s supervisor Mrs.ID Wattuhewa for
her guidance, effort and help throughout the research. I
express my sincere gratitude to the faculty of management
of Sir John Kotelawala Defence University and the Dean of
the faculty Dr.MM Jayawardana for accepting the research
project and the Head of the Department Dr.Namali
Sirisoma.
As well as I would like to thank all the people who were
supported me by giving the valuable time to fill my
questionnaire and provide answers, Director and other
respective people of Disaster Management centre, Director
and other respective people of Kaduwela Municipal
Council and Grama Niladari who spent their time for give
the information and share the experiences.
REFERENCES
Abhas, J., 2011. Prevention pays but be prepared for the
unexpected. Geneva, s.n
Anon., 2016. Disaster Management Centre of Sri Lanka.
[Online]
Available at: http://www.dmc.gov.lk/index_english.html
Anon., 2016. The UN and Disaster Management.
[Online] Available at: http://www.un-spider.
org/risks-and-disasters/the-un-and-disaster-
management
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE
PROCEEDINGS
Anon., 2016. The UN and Disaster Management.
[Online] Available at:http://www.un-spider.org/risks-and-
disasters/the-un-and-disaster-management
Authorities, G. f. P, 2009, The Planning System and Flood
Risk Management, s.l.: The Office of Public Work.
David, G., 2014. Dams and Flood Mitigation, s.l.: NSW
Government.
Industries, V. S. E. S. a. D. o. E. a. P., 2013. State Flood
Emergency Plan. Victoria: s.n.
Jayawardane, A., 2005. DISASTER MITIGATION
INITIATIVES IN SRI LANKA. University of Moratuwa.
Kim, S., Li, H. & Nam, J., 2015. Overview of Natural
Disasters and their Impacts in Asia and Pacific 1970-2014.
Disaster Risk Reduction Section ICT and Disaster Risk
Reduction Division.
Organization, W. M., 2013. Integrated Flood Management
Tool Series-Flood Forecasting and Early Warning,
Associated Program on Flood Management, s.l.: World
Meteorological Organization.
Sanjukta, B., 2008. DISASTER RISK REDUCTION A
PLANNING APPROACH. Director & HOD Department
Of Urban Planning School Of Planning And Architecture.
Sivakumar, S., 2016. Flood Mitigation Strategies Adopted
in Sri Lanka, Jaffna: International Journal of Scientific &
Engineering Research.
Tim, J., 2011. Toward a New Conceptual Framework for
Teaching About Flood Risk in Introductory Geoscience
Courses. JOURNAL OF GEOSCIENCE EDUCATION,
pp. 5-12.
903
PROCEEDINGS
A STUDY ON THE FACTORS INFLUENCING
THE USE OF e-COMMERCE BY
CUSTOMERS IN COLOMBO
WITH RESPECT TO B2C MARKET
SL Lokuhitige1#, LU Ranwala2
1 Faculty of Management & Social Sciences, CINEC Campus, Malabe.
2 Faculty of Management & Social Sciences, CINEC Campus, Malabe.
# shakilalakshan41@gmail.com
Abstract - This study explores the factors influencing the use of
e-commerce by the customers in Colombo district with regard to
the B2C market. The conceptual framework was designed based
on the Theory of Planned Behaviour (TPB) and 24 potentially
influential factors were explored throughout the study. Primary
data were collected by means of a questionnaire given in internet
based and printed formats where the respondents were received
from both users and non-users of online purchasing in the
Colombo district of Sri Lanka. 354 valid questionnaires were
used in the analysis and the Cronbach’s Alpha was 0.719 in the
reliability test. A proper and well explained descriptive analysis of
the variables followed by the KMO and Bartlett’s test, Hypothesis
testing, Factor analysis and an advanced model fitting was
carried out. Analysis revealed that there is a very positive opinion
towards online purchasing and people do have the willingness
towards it. In view of the above and the model developed, it
was found that attitudinal factors have a profound effect while
subjective norm and perceived behavioural control shows a
superficial effect on the online purchasing intention. Based on
the inferences from research findings, recommendations and
strategic and managerial suggestions were also made.
Keywords - Online purchase intention, Theory of Planned
Behaviour, Decomposed Theory of Planned Behaviour,
e-commerce
I. INTRODUCTION
With the advancement of technology, people have been
able to fulfil most of their activities just away from a click
of a button. Any field of work or industry you name,
technology has played, is playing and, will play its part
904 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
in order to facilitate the corresponding field or industry a
user friendly, simplified and a value adding service in every
aspect possible. Agriculture, Health & Safety, Finance,
Education, Security, Construction, Communication and,
Supply Chain can be identified as some of the industries
significantly expedited with technology.
In order to sustain, people have to fulfil their needs and wants. To
achieve these, they will have to find the sources of supply of these
requirements. With the passing of time and the development of
trade, different suppliers and manufacturers immerged where
people could accomplish their purchasing requirement of goods
and services from them. In a typical scenario, a customer will
reach a retail outlet or a store proximate and purchase the
product/s he/she wants. The aforesaid technology has made
a turning point in purchasing where at present a lot of online
platforms have come to stage which has made these activities far
more easy, simple and efficient in the meantime. With the help
of these ecommerce platforms, the customers have been able to
do their purchasing online via internet without purchasing in-
person. This has made drastic changes in supply chain where
at some point it could be perceived as an advantage whereas at
some point it is not. Electronic Commerce also referred to
e-commerce is one aspect that has emerged with the aid of
technology where it can be identified as a business module
or as a part of a large business entity which empowers a
firm or an individual to carryout business activities over
an electronic network; internet. Simply, e-commerce is the
use of internet to carry out business or rather commercial
activities such as online purchasing and subsequent
functions. The Operation of ecommerce can be identified
in major market segments/business model namely,
Business to Business (B2B), Business to Consumer (B2C),
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Consumer to Consumer (C2C) and, Consumer to Business
(C2B). Business to consumer (B2C) is a business or transactions
conducted directly between a company and consumers who are
the end-users of its products or services. However, the rise of
the internet has created an entire new B2C business channel in
the form of e-commerce or selling goods and services over the
internet.
II. METHODOLOGY AND
EXPERIMENTAL DESIGN
A collection of literature are refereed in order to identify
the experimental design for the study. Reliability test,
descriptive statistics, chi-square test for independences,
Factor analysis, KMO & Bartlett’s test and Regression
Linear Models are used for the analytical purposes.
in order to explain and predict consumer online shopping
attitude, intention and behaviour. Chang M. K., (2005) had
observed six studies of attitude toward online shopping
and all of these studies has shown that attitude toward
online shopping shows a significant positive impact on
online shopping intention and behaviour.
C. Decomposed Theory of Planned Behaviour
Taylor, (1995) introduced the idea that TPB beliefs
can be decomposed into multidimensional constructs
where attitude, subjective norm and perceived control
behavioural beliefs are decomposed as illustrated below.

A. Technology Acceptance Model

As per Davis, (1989) a theoretical model was developed

to explain and predict the user behaviour of information
technology where the researcher has built a relationship
between two main determinants of technology acceptance
identified as, Perceived Usefulness (PU) and perceived
Ease of Use (EOU).

1) Perceived Usefulness (PU)

As defined by Davis, (1989) PU is “the degree to which

a person believes that using a particular system would
enhance his or her job performance”. This can be simply
identified as the aspects where people tend to use or not
use an application to the extent that they believe that it will
help them to perform their work better. This follows from
the definition of the word useful: “capable of being used
advantageously”.
2) Perceived Ease of Use (PEU)
In contrast with perceived usefulness, perceived ease of
use refers to “the degree to which a person believes that
using a particular system would be free of effort”. This
follows from the definition of the word ease: “freedom of
difficulty or great effort”.
B. Theory of Planned Behaviour
Figure 1. Decomposed Theory of Planned Behaviour
D. Sampling
Simple random sampling method is used as the sampling
method of the research. Simple random sampling is a
randomized process without any favoured treatment
where each element in the population has got an equal
probability of being selected to the sample (Sample of n
from N population).
Simple random sampling is accepted due to this
characteristic, as well as, it is suitable to a population
which is very much larger than the sample.

The researchers Teo, (2001); Vijayasarathy, (2003); Wu, Colombo district is selected based on the fact that Colombo
(2003); Chang, (2008); Laohapensang, (2009) and Chiu, province comprises the highest population (2012) among
(2005) has used the Theory of Planned Behaviour (TPB) all districts (Economic and Social Statistics of Sri Lanka.,

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 905
PROCEEDINGS
2014) and that people in the Colombo district represents a
homogenous group with similar lifestyles where given that,
Department of Census and Statistics states that Colombo
district has the highest population of internet users and a
high literacy rate of 27.6% and 47.1% respectively.
E. Data Collection A descriptive analysis is done on the demographic
A combination of a physical paper-based questionnaire
and an internet-based questionnaire were used in data
collection. Internet-based questionnaire was developed
using Google Forms and mailed to accessible e-mail
addresses. Social network sites were also utilized to distribute
the questionnaire via the internet. The questionnaire was
distributed to 420 in total, concerning the time constraint of
the study. Considering both modes, a total of 362 responses
were received out of which eight respondents were rejected
due to partial completion of the questionnaire. Hence the
number of complete respondents were 354 which accounted
for a complete response rate of 84.28%.
F. Conceptual Framework
The conceptual framework is built primarily on the
theories of planned behaviour and the decomposed theory
of planned behaviour. Accordingly, the independent
variables which are Attitude, Subjective Norm and
Perceived Behaviour Control will be further divided with
reference to the decomposed theory of
Figure 2. Conceptual Framework
planned behaviour along with the demographic factors.
The dependent variable is identified as the Online
Purchase Intention (OPI).
906 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
III. ANALYSIS
A Cronbach Alpha value of 0.719 for 35 items was achieved
and it interprets the reliability of the questionnaire used in
data collection.
variables including age, gender, education, sector of
employment, income level, etc. It is identified through
the feedback that there exists a high overall online
purchasing intention of 81% or 286 respondents
out of 354 respondent sample whereas, 19% or 68
respondents have a low overall online purchasing
intention. This implies the overall willingness of the
respondents to engage in online purchasing.
A. Bivariate Analysis
The bivariate analysis has delivered a variation of measures
in a two-way table as interpreted by an example below.
Table 1. Online purchasing
intention vs. Save time
Out of the respondents who have a low overall online
purchasing intention, 63.2% have agreed to that online
purchasing saves time while 20.6% and 16.2% hold neutral
and strong agree upon the fact.
On the other hand, it is evident that, 59.1% of the
respondents who claims to be having a high online
purchase intention have agree on the fact that online
purchasing saves time. Therefore collectively, a total of
86.4% of the respondents with a high overall willingness to
purchase online have positive views towards the fact that
online purchasing saves time.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

B. Chi-Square Test 1) KMO and Bartlett’s Test:

Table 3. KMO and Bartlett’s Test:
With the results of cross tabulation (bivariate analysis),
hypothesis testing has been carried out using Chi-
square test to check the relationship between the online
purchasing intention and other variables concerned.
The Hypothesis testing is as follows,

H0: Online purchasing intention is independent from the

ith variable
H1: Online purchasing intention is dependent on the ith
variable As KMO test statistic is 0.679 which is greater than 0.6, it
can be concluded that sample size is adequate for a factor
Table 2. Chi-Square Test for the ith variable analysis to be proceed. The two hypothesis which are
tested in the KMO and Bartlett’s test are as follows.

Ho: There does not exist any correlations among the

variables
H1: There exists correlations among the variables

As p-value of the Bartlett’s test is 0.000, null hypothesis

is rejected. Hence, it can be concluded that, correlation
matrix is not an identity matrix which further supports
the strength of the relationship among variables used in
factor analysis.

2) Total Variance Explained:

Figure 3. Total Variance Explained

As per the Figure 3 shown above, we can observe the

initial Eigen values which are greater than one. It can
be observed that the first eight components carry Eigen
values which are greater than one where we can select these
C. Factor Analysis eight components. In addition to that, the first component

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 907
PROCEEDINGS ISBN number: 978-955-0301-41-6

identified above, accounts for 24.957% of the variance

while the second, third and the fourth components
accounts for 9.484%, 9.028% and 8.000% respectively. The
eight factors identified above has explained 72% of the
total variance explained collectively.

3) Scree Plot:

The scree plot helps to determine how many factors to

be retain. It can be seen that the curve begins to flatten
between the component eight and component nine.
Therefore it further elaborates that only eight components
can been retained.
Figure 4. Scree plot for the variables

Table 4. Rotated Component Matrix of variable

908 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

D. Rotated Component Matrix below mentioned variables under five percent level.

The main purpose of factor rotation is to minimize number • Age
of factors on which the variables under investigation have • Gender
high loadings. Meaningful factors can be obtained by • Education
rotating factors. Specially, Varimax Rotation method is • Employment
used in this study. • Income
• How often purchased online
As shown in the Rotated Component Matrix table shown • What is mostly purchased online
in Table 4.30 above, rotation has reduce the number of
factors on which the variables under investigation have 3) Advanced Model Fitting
high loadings which makes the interpretation of the
analysis easier. According to the rotated factor loadings, Omnibus Test for Factor 01 is tested and a significance
24 variables can be categorized for extracting eight factors. value of 0.000 is obtained which is less than 0.05. Hence
The eight factors are identified as below. the current model outperforms the null model.

Factor 01 – Influence and Convenience The parameter estimates of “Influence and Convenience”
Factor 02 – Ability and Advantageousness were identified and the below mentioned fitted model was
Factor 03 – Delivery and Influence built up in conclusion.
Factor 04 – Cheap
Factor 05 – Reliability
Factor 06 – Preference
Factor 07 – Risk Freeness
Factor 08 - Availability

Reliability of the factors are tested for and the first two
factors are found to be reliable with 0.800 and 0.758 values
of Cronbach Alpha.

E. Factor 01 – Influence and Convenience

Under factor 01, which is “Influence and reliability”, the The variables used above in the model are identified as
following key areas are taken in to discussion. bellow.

• Reliability Table 5. Interpretation of the variables

• Correlation
• Advanced Model Fitting

1) Reliability: The reliability of the variables affecting to

influence and convenience (Factor 01) were checked and
a Cronbach’s alpha value of 0.800 was obtained. Hence
this factor is used for further analysis.

2) Correlation:

H0: Influence and convenience (Factor 01) is independent

from the ith variable
H1: Influence and convenience (Factor 01) is dependent
on the ith variable

Influence and convenience (Factor 01) is dependent on the

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 909
PROCEEDINGS
As per the above developed model, Compared to a person
whose age is above 55 years,
• A person in the age category of 46 – 55 years makes
the person 1.304 times more likely to be influenced and
convenient in using online shopping.
• A person who is in the age category of 36 – 45 years
makes that person 1.346 times more likely to be
influenced and convenient on using online shopping.
• A person in the age category of 26 - 35 is 3.069 times
more likely to be influenced and convenient towards
the use of online shopping.
• A person in the age category of 18 – 25 years makes
the person 2.033 times more likely to be influenced and
convenient in using online shopping.
• A person in the age category below 18 is 0.255 times
more likely to be influenced and convenient to engage
in online shopping than of the person above 55 years of
age.
Similarly, the parameter estimates of “Ability and
Advantageousness” were identified and the subsequent
fitted model was built.
IV. DISCUSSION AND
CONCLUSION
1) Discussion on the demographic profile and
preferences:
It is observed that 81% of the respondents have a high
overall willingness towards the use of e-commerce or
rather engage in online purchasing where only 19% has
a low overall willingness. 354 valid questionnaires were
used in the analysis with a Cronbach’s Alpha value of
0.719 in the reliability test. This suggests that the internal
uniformity of the research instrument is good and has
succeeded the required thresholds. The demographic
attributes in respect of the overall intention for online
purchasing is stated below.
Age can be considered as an important parameter in
studying the willingness for online purchasing. As per the
analysis carried out in chapter four, respondents in the age
group of 18 – 25 accounts for a 47.2% of the respondents.
It was observed that 86.7% of the respondents in this age
910 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
category have a high online purchasing intention. The
second highest age group below the age of 18 and 100%
of the respondents below the age group of 18 have a high
online purchasing intention. Next comes the age categories
above 55, 36 – 45, 26 – 35 and 45 – 55 where the overall
online purchasing intentions of these age categories
being high were identified to be 15.2%, 80.8%, 76.0% and
53.8%. Hence it implies that population below the age of
45 is more likely to have a high overall online purchasing
willingness/intention.
When the gender is taken in to consideration, out of
the total male respondents, 78.1% have a high online
purchasing intention while for females, it is 85.4% of
the total female respondents. Therefore it shows a slight
tendency of the females to have a higher online purchasing
intention than the males.
The level of education can be also be considered here,
where 100% of the respondents having done the O/Ls
have a high online purchasing intention and so were the
93.5% of the respondents with the A/L qualification. It was
evident that 78.2% of the respondents with a degree level
education qualification, possess a high online purchasing
intention. Respondents with masters and doctorates are
less likely to be engaged in online purchasing.
The area of employment cannot be identified as a
significant factor where unless retired, all the respondents
do have a tendency towards online purchasing.
Once the type of goods most purchased online by the
customers is concerned, it was identified by the study
that expect for books, a majority of the respondents
have an interest in buying footwear/clothing, jewellery/
accessories, electrical appliances and electronic items
online. This implies that the online store service providers
have a great opportunity in a variety of goods.
As per the above findings, discussions and
recommendations can be done in two perspectives which
are from the perspective of the customer and the service
provider.
Service Providers’/ Business Perspective
• Service providers could consider to focus more on
the age category below 45 years of age since this age
category in the sample consists respondents with high
online purchasing intention.
 ISBN number: 978-955-0301-41-6
• There is a sight tendency of the females to be more
influenced towards the use of online purchasing than
men.
• E-commerce or rather online purchasing is more
commonly influenced with the level of income an
individual received. Higher the salary, higher the overall
willingness towards the use of online purchasing.
• A majority of the sample with high online purchasing
intention agree to that online purchasing allows
comparison shopping. Hence, it would be ideal for the
service providers to study the online market and have a
strategic pricing policy to have competitive advantage.
• 39.5% of the respondents with high online purchasing
intention have neutral views on the fitness of online
purchasing to the customer needs. Therefore the service
provider could consider doing a market research and
a study on the customers to improve the fitness of the
online goods available as per the customer requirement.
• Both respondents with high and low online purchasing
intention have a neutral perception on the fact that the
delivery fees of online purchased items are low. Hence,
the online purchasing service providers could consider
taking this as an opportunity to convert this population
to active users of online purchasing.
• It was evident that the sample respondents have a
negative perception towards the reliability of the
delivery once purchased online. Therefore, the service
providers could try to change this perception in the
customers’ mind and convert this population to a target
market.
Customers’ Perspective
• Through the findings of the research, it was observed
that attitudinal variables have a profound effect while
subjective norm and perceived behavioural control do
not show a superficial effect on the online purchasing
intention.
• Attitudinal variables which were found to have
significant impact include the below mentioned,
1. Save time
2. Comparison shopping
3. Ease of getting skilled to engage in online purchasing
4. Fitness for the purchasing need
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE
PROCEEDINGS
5. Reasonable delivery time
6. Ease of learning to engage in online shopping
7. Fitness for the purchasing need
• A majority of the respondents/customers have said
that they mostly purchase clothing and footwear items
online where, as the second highest comes jewelry and
accessories.
ACKNOWLEDGMENT
First, I would like to extend my sincere gratitude to
Ms. Lakshmi Ranwala who was contributory in laying
the underpinning for this research and the continuous
supervision, motivation, appraisal and assessment
provided in putting me on the right path to success.
Without her guidance this research would have not have
been a success. Next, I extend my gratitude to all the
other members of staff of the Department of Logistics and
Transport, Faculty of Humanities and Social Sciences for
their generous support in proceeding with the research.
Further, I am greatly thankful to all my friends at my
workplace; KPMG who helped me in many ways to make
this research a success and a special thanks goes to all the
respondents for taking the time to assist me by completing
the questionnaires with interest. Last but not least, I would
like to give my heartfelt thanks to my family members and
friends for their support and direction throughout the
study.
REFERENCES
Ajzen, I., 1991. The Theory of Planned Behaviour.
Organizational Behaviour and Human Decision Processes.
Volume 50, pp. 179-211.
Bhattacherjee, A., 2000. Acceptance of e-commerce
services: the case of electronic brokerages. Systems, Man
and Cybernetics, Part A: Systems and Humans.
Chang, H. H. &. C. S. W., 2008. The impact of online
store environment cues on purchase intention trust and
perceived risk as a mediator. 32(6), pp. 818 - 841.
Chang, M. K. C. W. &. L. V. S., 2005. Literature derived
reference models for the adoption of online shopping.
Information & Managemen, 42(4), pp. 543 - 559.
Chi-Square Test for Independence, 2014. Stat Trek.
Availableat:http://stattrek.com/chi-s quare-test/
independence.aspx
911
PROCEEDINGS
Chiu, Y. L. C. &. T. L.-L., 2005. Gender differs: assessing
a model of online purcahse intentions in e-tail service.
International Journal of Service Industry Managemen,
16(5), pp. 416 - 435.
Chui, M. H. &. C., 2004. Predicting electronic service
continuance with a decomposed theory of planned
behaviour. Behaviour & Information Technology.
Cronbach, L., 1951. Coefficient alpha and the internal
structure of tests.
CRR, 2015. Center for Retail Research. Available at: http://
www.retailresearch.org/onlineretailing.php
Davis, F., 1989. Perceived usefulness, perceived ease of use
and and user acceptance of information technology. 13(3),
pp. 319-340.
Internet World Stats, 2010. Internet World Stats - Usage
and Population Statistics.
Available: http://www.internetworldstats.com/stats.htm
J.C.Jayasekara, 2009. Digital Library. Available at: http://
dl.lib.mrt.ac.lk/handle/123/961
Jiyeon, S. Jayendra & K., 2012. Factors affecting Indian
consumers’ online buying behavior. Innovative Marketing.
Joey, G., 2004. The theory of planned behavior and Internet
purchasing. Journal of Internet Research.
Keith, M., 2015. Global Ecommerce Sales. Available at:
https://www.remarkety.com/global-ecommerce-sales-
trends-and-statistics-2015.
912 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Laohapensang, O., 2009. Factors influencing Internet
shopping behaviour: a survey of consumers in Thailand.
Journal of Fashion Marketing and Management.
Lin, H., 2007. Predicting consumer intentions to shop
online. An empirical test of competing theories.
Rogers, E., 1995. Diffusion of Innovations, s.l.: Simon and
Schuster.
Taylor, S. &. T. P. A., 1995. Understanding Information
Technology Usage: A Test of Competing Models.
Information Systems Research.
Teo, T., 2001. Demographic and motivation variables
associated with Internet usage activities. Internet Research:
Electronic Networking Applications and Policy.
Todd, S. a. J., 1995. Is there a future of retaining in the
Internet, s.l.: A. Peterson, Electronic Marketing and
Consumers.
Touliatos, J. a. C. N. H., 1988. Research Methods in Human
Ecology/Home Economics, Lowa.
Vijayasarathy, L. R., 2003. Predicting consumer
intentions to use on-line shopping: the case for an
augmented technology acceptance model. Information &
Management.
Wu, S., 2003. The relationship between consumer
characteristics and attitude toward online shopping, s.l.:
Marketing Intelligence & Planning.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

ANALYSIS OF THE IMPACT OF GREEN

PRACTICES ON THE PERFORMANCE OF
LARGE SCALE APPAREL INDUSTRY OF
SRI LANKA
SYD Wanniarachchi1 and DR Ratnajeewa1
1General Sir John Kotelawala Defence University, Sri Lanka
# dilanirr@gmail.com

Abstract - This research has been conducted to analyze and environmental activists. This concept of green
the impact of green practices on the performance of large manufacturing has become a key strategy for companies
scaled apparel industry. People are increasingly concerned in generating competitive advantage and sustainability to
with the social responsibility and business ethics while achieve the organizational goals and objectives.
doing businesses. Suppliers prefer companies that have
good ethical practices where green logistics is one of the Keywords - Green practices, Environmental impact,
major priorities. Consumers are also increasingly more Water usage, Energy usage, Carbon footprint
aware and concerned with global issues, which results
in greater demand for eco-friendly or green goods and
services.
I. INTRODUCTION
Consumers in the current business world, who are
The objective of the study was to analyze the impact of
the final users of clothing made in countries like Sri
green practices in the performance of large scale apparel
Lanka, are becoming more aware of environmental
industry of Sri Lanka. Study covers the production lines
impacts from industries and thus companies are
that are enlisted with the Sri Lankan Board of Investment,
feeling the pressure to be more environmentally
which principally accumulates substantial scale
conscious in their manufacturing (Cai & Li, 2008). At
organizations in the garment sector.
present, most of the fabric companies focus on going
green due to increasing concerns about adverse
The sample is 30 large scale garment factories.
environmental impacts from industries, particularly
Judgemental sampling was used to select the major apparel
global warming.
manufacturers. Interviews as well as questionnaire survey
method were used for data collection. The study discovers
Though a few number of research have been done in the
most of the plants overviewed deliberately actualized no
apparel industry, it is a timely requirement to conduct a
less than one natural management practice which affects
research since there is a research gap on this area as the
positively, for example, water re-use, material re-use and
impact of green practices on the operational performance
ecological reviews, reduction in the carbon footprinted,
Sri Lankan large scale apparel industries has not been
solar power usage. Employees have a more favourable
focused on much.
view on companies that have environmentally friendly
practices.
The major environmental impacts associated with the
production and use of apparel throughout its life cycle
Today businesses focus more on quality management
include wastewater emissions from dyeing, finishing,
to grab customer interest via inculcating rich corporate
washing processes, increase in pollution, solid waste
social responsibilities which is the main concern of
production, significant depletion of resources from
the consumers, suppliers, employees, government
consumption of water, fossil fuels, and raw materials

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 913
PROCEEDINGS ISBN number: 978-955-0301-41-6

(Islam, 2013). Chemicals are also released in wastewater Except for a few organizations the degree of commitment
from processes such as pre-treatments, dyeing, finishing, of the ground level manufacturing has changed acceptably
and laundry. The objective of the study was to analyze (Thilakarathna & De Silva, 2014). The frequency of
the impact of green practices on the performance of large environmentally responsible practices carried out by the
scale apparel industry of Sri Lanka. organizations will momentarily receive an enhancement
from the International Organization of Standardization.
ISO 14000, a registration process for vendors similar
II. LITERATURE REVIEW to the ISO 9000 series, will specifically target ecological
procedures universal (Wu & Dunn, n.d.).
The life span of the businesses is depending upon the
Green industry in the textile sector will integrate different
feasibility towards the environment. This can be observed
methodologies in numerous ways in order to maximize
in terms of which natural resources the production
the level of benefit they incur.
process might need, how resources are used and
recharged, the overall influence of the end production on
This will also generate a very good environment for
the environment and where the product ends up resulting
the efficiency and productivity of employees where the
its disposal (Islam & Rahman, 2014).
industries have taken better mechanisms to generate a
successful and an effective working environment for the
Total value of export earnings in the sector was at US
employees. There are a lot of measures taken namely,
dollars 2,424 million accounting for 52 % of the total
temperature reduction passively in the green factory by
export earnings in 2002. The contribution to the Gross
using cool roof (off white color coated Zinc and aluminum
Domestic Product (GDP) was 5.3% in 2002. This industry
roofing sheet), photovoltaic roofs (solar panels) and green
has delivered more than 330,000 direct employment or
roofs (concrete roof with vegetation on top) (Thilakarathna
5% of country’s total employment in more than 1,060
& De Silva, 2014).
garment factories (Dheerasinghe, 2003) The industry
has enjoyed an enormous growth level over the past four
Sri Lanka suppliers and manufacturers have taken a
decades and is today Sri Lanka’s primary foreign exchange
cooperative effort in order to reduce the carbon footprint
earner accounting to 40% of the total exports and 52% of
to make the local apparel industry more eco-friendly and
industrial products exports (Central Bank of Sri Lanka,
sustainable (Export development board, n.d.) Additionally
2015).
the green factory is powered by carbon-neutral sources,
but most factories get power from the main grid and
The most noteworthy industry in Sri Lanka is the clothing
diesel generators. Further it saves 50% of the potable water
industry. It has been mounting within the last 3 decades
consumed by an equivalent normal factory (Thilakarathna
and has become the number one foreign exchange earner.
& De Silva, 2014). Green factory uses push taps, low flow
It has reduced water consumption by 70% along with 50%
plumbing fixtures and many other things to save water,
of the electricity consumption. With reduced overheads,
rain water harvesting tanks are constructed on top of
fast return on investments and through the practice of lean
toilets which send water down for flushing under gravity
manufacturing these factories have found their business
flow. For purifying the air, to diminish soil erosion and to
sustainability (Export development board, n.d.).
preserve water, green factories have planted native trees.

Thus green factory helps preserve the natural environment

Through building green factories top companies in Sri
and ecosystems healthy for humans, animals and plants by
Lanka have taken ingenuities to make their manufacturing
reducing waste and greenhouse gas emissions, controlling
process sustainable. By doing so they have succeeded with
pollution and treating land, air and water as precious
environment credentials such as certifications and titles as
resources (Thilakarathna & De Silva, 2014) For industries
a part of company’s ethical trading.
with lower margins, such as the clothing industry, green
supply chain management can move towards reduction
Outsized organizations started using green practices
in supply chain related costs. These cost decreases can be
from their ground level manufacturing itself as a
lead into substantial competitive advantages and profit
marketing strategy for their patrons to go beyond
(Eryuruk, n.d.).
their competitors to show the corporate, social and
environmental responsibilities they take as a company.
The sustainable construction movement is now universal

914 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

in scope, with almost 60 national green building councils conceptual framework depicted in Figure 1 was developed
establishing determined performance goals for the built based on previous literature and factors identified.
environment in their countries (Charles, 2012). The
concept of Green building is an all-inclusive start with
the understanding that the constructed environment can
have thoughtful effects, both plus and minus but mostly
positive, on the natural environment, as well as the people
who reside buildings every day (Kriss, 2014). Green factory
has helped the employees to reduce their absenteeism rate
and remaining in the factory.

The average absenteeism rate in green factory and

non-green factory were 3.07% and 4.38% respectively.
Therefore, we can see a clear difference. Green building is
a struggle to strengthen the positive effects and alleviate
the negative effects throughout the entire life cycle of
a building. By adjusting energy performance, reducing
emissions, reducing waste, encountering recycling,
reducing sediment contamination and soil water erosion
green factory achieves environment sustainability.
The factory’s land area is left to nature, covered with
greenery or water and managed as a habitat for plants and
animals to make the environment pleasing and to make
the factory more eco-friendly (Thilakarathna & De Silva,
2014).
Figure 1: Conceptual Framework
During the analysis, the Kaiser–Meyer-Olkin (KMO)
test was used to test the sampling adequacy of the
study. The reliability of the sample was analysed
through Cronbach’s Alpha value which evaluates the
internal consistency within the variables of the study.
Descriptive statistics and Correlation analysis were
used as the main data analysis methods. SPSS was the
software used for the data analysis.
IV. RESULTS

III. METHODOLOGY The collected data was entered into SPSS 22.0 which can
be identified as a statistical software tool.
The impact of green practices on the performance of
the large scaled apparel industry is considered as the Table 1: Test of Reliability using the
dependant variable of this study. Reducing Energy usage, Cronchbach Alpha value
Reducing Carbon usage, Reducing Water usage and
Reducing the environmental impact are the independent
variables considered.

The study was based on primary data collection. To obtain

the primary data, a questionnaire was designed consisting
of the factors obtained through an extensive literature
review. Thereafter a questionnaire survey was conducted.
The questionnaire was distributed among the factories
Table 2: KMO test value
that are practicing green practices. The Likert scale with
five options was used to capture the responses in the
questionnaire. 30 large scaled apparel manufacturers in Sri
Lanka who are engaged in green practices were the sample
considered for the study.

Judgemental sampling was used in order to select the major

apparel manufacturers. Interviews as well as questionnaire
survey method were used for data collection. The

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 915
PROCEEDINGS
As in Table 2 the KMO value obtained is 0.806 which is
greater than the acceptable value of 0.5, therefore this
ensures the sampling adequacy of the study.
The Cronbach’s Alpha value is 0.980, as in Table 1, which
is greater than the acceptable value of 0.7, therefore the
internal consistency is depicted.
Table 3: Correlations- Reduction in energy
usage and organizational performance
** Correlation is significant at the 0.01 level (2-tailed).
As per Table 3 Kendall’s tau_b the reduction of the
energy usage and the impact of green practices on
the organizational performance are positively related
assuming other factors (environmental impact, water
usage reduction, carbon footprint reduction) remains
constant.
Since the correlation coefficient which is 0.808 represents
that the correlation is strong. As per Spearman’s rho
reduction of the energy usage and organizational
performance are positively and strongly correlated
(correlation coefficient which is 0.916) Therefore, there
is a strong positive correlation between reduction
of energy usage and the impact on organizational
performance.
916 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Table 4: Correlation-Reduction in the amount
of carbon footprint and the organizational
performance
**. Correlation is significant at the 0.01 level (2-tailed).
As per As per Table 4 Kendall’s tau_b the reduction
of the carbon footprint amount and the impact of
green practices on the organizational performance
are positively related.
The correlation coefficient which is 0.819 represents
that the variables have a strong positive relationship.
As per Spearman’s rho reduction of the carbon footprint
amount and the impact of green practices on the
organizational performance are positively.
The correlation coefficient is 0.931 and there is a strong
positive correlation between the reduction of carbon
footprint amount and the impact on performance.
 ISBN number: 978-955-0301-41-6
Table 5: Correlations-Reduction in the
amount of water usage and impact on the
organizational performance
**. Correlation is significant at the 0.01 level (2-tailed).
As per Table 5 Kendall’s tau_b the reduction of the
amount of water usage and the impact on organizational
performance are positively related. Since the correlation
coefficient is 0.723 it represents that the correlation is
significant.
As per Spearman’s rho reduction of the amount of water
usage and the impact on organizational performance are
positively related. Since the correlation coefficient is 0.864
the correlation is significant.
There is a strong positive correlation between
reduction of water usage and the impact on
organizational performance.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE
PROCEEDINGS
Table 6: Correlations - Reducing the
environmental impact and the organizational
performance
**. Correlation is significant at the 0.01 level (2-tailed).
As per Table 4.4 Kendall’s tau_b reducing the
environmental impact and the impact of green practices
on the organizational performance are positively related.
Since the correlation coefficient is 0.830 the correlation is
significant.
As per Spearman’s rho, reducing the environmental impact
and the impact on the organizational performance are
positively related. Since the correlation coefficient is 0.936
the correlation is significant. There is a strong positive
correlation between reduction of the environmental
impact and organizational performance.
917
PROCEEDINGS
IV. DISCUSSION AND
CONCLUSION
A. Discussion
Several green practices of the Sri Lankan large scale
apparel industry were analyzed by this study. According
to the previous literature by (Dheeerasinghe, 2003) the
current study selected thirty large scale apparel companies
which are practicing green practices currently. The
organizational performance was the dependant variable
and the independent variables were the reduction in the
environmental impact, reduction in the water usage,
reduction in the energy usage and reduction in the
carbon footprint. Through the analysis it can be stated
that reduction of water usage, reduction of energy
usage, reduction of carbon foot print and reduction
of the environmental impact all have strong positive
relationships with the organizational performance of the
large scale apparel manufacturing industry. Also through
the interviews conducted it was revealed that many of the
large scale apparel manufacturers have already started
recycling suitable products such as plastics by partnering
with the relevant parties. Use of solar panels and push
system taps for reduction of water wastage are some of the
practices most popularly implemented.
In the modern business world, green practices have been
identified as one of the most upcoming trends that have
become a key concern towards sustainability. As per
the core objective of this research, the impact of green
practices on the performance of the apparel industry is
analyzed. It has been identified that Sri Lanka has turned
into a global player in the apparel production and exports
becoming a major supplier for many leading fashion
brands (Thilakarathna & De Silva, 2014). Therefore
implementing more and more of the discussed green
practices will help manufacturers to improve their
reputation as environmental friendly manufacturers
and at the same time improve their organizational
performance. In the long run they will be able
to make an impact on their customers’ mind as
sustainable parties in manufacturing.
B. Conclusion
As the apparel sector is a key industry that
contributes to a significant amount of negative
environmental effects it has become an essential
918 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
factor for the apparel sector to practice ecofriendly
activities throughout their manufacturing process
in order to increase sustainability and protect the
environment for the future generation. Minimizing
the amount of energy used, reduction of carbon
foot print, decrease of water usage and minimizing
pollution and environmental impact are main
practices to be adopted for the manufacturers in
contributing towards sustainability. This study used
a sample of thirty large scale apparel manufacturers
to obtain data required to carry out the study. When
conducting this research it became evident that
most of the companies do believe implementation
of green practices generates positive effects for the
organization as a whole as it contributes towards
the organization’s success in numerous ways. There
are many mechanisms that have been implemented in
the apparel companies related to the main key variables
energy, water, carbon footprint and the environment as
a whole to minimize the wastage and the adverse effects
generated due to the company’s manufacturing process.
As further research green practices in the small scale and
medium scale apparel manufactures and their impact on
the performance can be studied.
REFERENCES
Bank, C., 2015. Central Bank Report, Rajagiriya: Central
Bank.
Cai, W. & Li, H., 2008. Green Marketing and Sustainable
Development of Garment. Journal of Business and
Management.
Charles, J. K., 2012. Green Building Design and Delivery.
3 ed. s.l.:s.n.
Dheerasinghe, R., 2003. Garment Industry in Sri Lanka
Challenges, Prospects and Strategies, s.l.: s.n.
Dunn, S. C. & Wu, H.-J., n.d. environmentallly responsible
logistics systems. In: s.l.:s.n.
Eryuruk, S. H., n.d. Greening of the Textile and Clothing
Industry, s.l.: s.n.
Export development board, n.d. sustainability and ethical
practices. [Online].
Export Development Board, n.d. sustainability and ethical
practices. [Online].
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Hirdaramani, 2016. Hirdaramani. [Online] *Kriss, J., 2014. What is green building. LEED, 6 August.
Thilakarathna, T. & De Silva, S., 2014. Sustainability
Available at: http://www.hirdaramani.com/sustainability/ Through Building of Green Factories, s.l.: s.n.
green_zero_carbon.php
Wu, H. J. & Dunn, S. C., n.d. Environmentally responsible
Islam, M. & Rahman, M., 2014. Environmental
logistics systems, s.l.: s.n.
Sustainability Evaluation of Apparel Product: A Case
Study on Knitted T-Shirt. Hindawi journal of textiles.
Islam, S. M., 2013. Approaching Sustainability in Textile
and Garment Industries, Bangladesh , Bangladesh: Royal
Institute of Technology .

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 919
PROCEEDINGS ISBN number: 978-955-0301-41-6

LIST OF REVIEWERS

Internal Reviewers External Reviewers

Mr. WAAK. Amaratunge Prof. EA Gamini Fonseka

Ms. RMNP Rajapakse Prof. Mrs. G.J.S. Wijesekara

Dr. L Liyanage Dr. PDP Shanthi

M.T.N Wijetunge Dr. Saroja Adihetti

Ms. Krishanthi Anandawansa Mr. HAMA Hapugoda

Mrs. CJ Kothalawala Dr. Varuni Tennakoon,

Major. JPWK Abeyawickrama Dr. Asantha Aththanayaka

Ms. Nirupa Dharshani Ranasinghe M.T.M. Mahees

Ms.HMAGK Ekanayake Mrs. S Weerasinghe

Dr. Hemantha Premarathne Senior Prof Shantha K Hennayake

Mr S Satheesmohan Dr. Ramila Usuf -Thowfeek

Dr. MM Jayawardana MG Kularathne

Mr. Chandana De Silva Dr. Dhammika Herath

Ms Hasini Rathnamalala Mr. S. Sivakanthan

Prof. Amal Jayawardana Mr. M. A. N. Chandratilake

Mr. Sanath.J. Gunawardena Dr. Priyanga Dunusinghe

Ms. WMMMJT Weeraratna Rev. Ilukwela Dhammarathana thero

Ms. MKOK De Silva
Dr. M.D. Ranaweera
Prof. Nandani De Silva

Dr. KMN Kumarasinghe

920 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Management

External Reviewers Internal Reviewers

Dr C Pathirawasam, Dr. Pradeep Kalansooriya

University of Kelaniya
Dr. Chaminda Wijethilake
Dr. Jagath Munasinghe,
University of Moratuwa Dr. Hemantha Premarathna

Dr. Nalin Abeysekara, Dr. Namali Sirisoma

Open University
Dr. Upali Rajapaksha
Dr. R.A Rathanasiri
UOW Mr. Dulitha Hewadikaram

Mr. Nuresh Eranda, Mr. Wasantha premarathna

University of Peradeniya
Mr. Asela Gunasekara
Prof. Milton Rajarathna,
University of Peradeniya Ms. Chathurani Rathnayaka

Ms. Kalpana Ambepitiya

Ms. Krishanthi Anandawansa

Ms. Dushanthi Lokuge

Ms. Nalinika Rajapaksha

Ms. Romita De Silva

Ms. Imendra Wathuhewa

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 921
PROCEEDINGS ISBN number: 978-955-0301-41-6

Allied Health
Sciences
Allied Health in the Global Environment:
Challenges and Opportunities

922 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
SESSION SUMMARY
The session theme was “Allied Health in the Global En-
vironment: Challenges and Opportunities”. This session
comprised of four plenary speeches delivered by distin-
guished national and international scientists. Plenary
session of Allied Health Sciences had four emerging sci-
entific topics; “Genomics to Molecular Diagnostics to
Personalized Medicine”, “Nursing in Global Environment:
Challenges and Opportunities”, “Does dose matter?” and
“Challenges and way forward of sports and exercise med-
icine in Sri Lanka”. The plenary session of the Faculty of
Allied Health Sciences was chaired by Senior Professor
Narada Warnasuriya. Professor Narada Warnasuriya is
the Senior Professor of Paediatrics and Chairman of the
Internal Quality Assurance Unit at the Sir John Kotela-
wala Defence University (KDU). He is also the Emeritus
Professor of Paediatrics at the University of Sri Jayawar-
denepura (USJP). He was formerly the Senior Professor of
Paediatrics, Dean, Faculty of Medical Sciences and Vice
Chancellor at the USJP. The first speech was made by
Prof. Ranil Dassanayake on“Genomics to Molecular Di-
agnostics to Personalized Medicine”. He is a Professor at
the Department of chemistry, University of Colombo. He
has obtained his bachelor’s degree in Chemistry in 1993
from the Faculty of Science, University of Peradeniya and
has obtained his PhD in Biochemistry and Molecular Bi-
ology from the University of Hong-Kong in 2001. He has
served as a senior lecturer in Biochemistry and Molecu-
lar Biology at the Department of Chemistry, University of
Colombo and was a Research Associate at the Department
of Microbiology at the University of Hong Kong. Second
speech was made by Dr.Rasika Jayasekara on “Nursing in
Global Environment: Challenges and Opportunities”. He
is a senior lecturer in school of nursing and midwifery,
university of South Australia. He has obtained his bache-
lor of science in nursing degree in 1999 from the Faculty of
Natural Sciences, Open University of Sri Lanka. He com-
pleted his master of nursing science & master of clinical
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
science in evidence based healthcare from the faculty of
health sciences, university of Adelaide, South Australia
and has obtained his PhD from the University of Ade-
laide, South Australia. He has served as a nursing tutor,
school of nursing, Rathnapura, Sri Lanka & clinical asso-
ciate lecturer, The Joanna Briggs institute, Adelaide Medi-
cal School, University of Adelaide, South Australia. The
third speech was one of the most interesting studies which
took the attention from the audience and it was on “Does
dose matter?” on medical imaging done by Mark McEn-
tee. He is an Associate Professor of University of Sydney
and he is an internationally recognized researcher and a
teacher in diagnostic radiography. He was the President of
the Irish Institute of Radiography and Radiation Therapy
in 2005 & became a Senior lecturer in the University of
Sydney in 2011 and an Associate Professor in 2014.He is
an Honorary Professor of Radiography at the University
of Salford, Manchester (UK), a fellow of the Irish Insti-
tute of Radiography, and has been appointed as an expert
consultant by the governments of Singapore and Tonga.
The final speaker of the plenary session was Dr. Lakshman
Edirisinghe and the topic was on “Challenges and way
forward of sports and exercise medicine in Sri Lanka”. Dr.
Lakshman Edirisinghe is the Director General of Institute
of Sports Medicine, Ministry of Sports. He has obtained
his bachelor of surgery bachelor of medicine (MBBS)
degree from University of Colombo in 1993 & Master of
Science in community medicine from Post Graduate In-
stitute of Medicine, University of Colombo in 2002 and
Diploma in Environmental Management from Faculty of
Science, University of Kelaniya. He was the Deputy Pro-
vincial Director of Health Services – Central Province
and the Regional Director of Health Services Kurunegala
District. He is a Council Member of College of Medical
Administrators, member of National Steering Committee
of Maternal and Child Health & many more respective in-
stitutes & organizations. At the end of all speeches Prof.
10TH INTERNATIONAL RESEARCH CONFERENCE 923
PROCEEDINGS
Narada Warnasuriya opened the session for the discussion
started with the first speaker. Dr. Kottahachchi and Dr.
Chandana commented and inquired about virus activity
from Prof. Ranil Dassanayake and had a good discussion
on the topic. As the second discussion Prof. Narada start-
ed with asking about nursing professional bodies in Aus-
tralia from Dr. Rasika. Then Dr. Lakshman and Prof. Mark
commented on the same topic. Mr. Rahal inquired about
the safety and mutations happened due to the exposure
of X –ray from Prof. Mark as the third discussion. Prof.
924 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Narada Warnasuriya also commented on the safety of ra-
diography and radiotherapy profession and had a fruitful
discussion on the topic. Last discussion was on the situ-
ation of sports medicine in Sri Lanka. Prof.Ranil and Mr.
SADCS Senevirathna asked about the facilities, services
provide by the sports medicine unit of sports ministry. Dr.
Lakshman Edirisinghe commented on the issues and chal-
lenges in sports medicine in Sri Lanka. At the end of the
plenary session chairperson, Senior Prof. Narada Warna-
suriya congratulated the speakers again and stated that the
session was very interesting as well as an interactive one.
 ISBN number: 978-955-0301-41-6
Technical session I
Radiography & Radiotherapy The technical sessions on
Radiography/Radiotherapy conducted under the theme
Allied Health in the Global Environment: Challenges &
Opportunities was chaired by Dr Jayantha Balawardane
Senior Lecturer Grade I in Faculty of Medicine, Head of
the Department of Radiography & Radiotherapy, Faculty
of Allied Health Sciences and Executive Director of the
University Hospital KDU. The session commenced with
the presentation on Assessment of Frequency of Errors in
Dental Panoramic Radiographs at Dental hospital, Pera-
deniya, Sri Lanka. The collaborators of the research were
AJSP Nileema, L Thasanthan, DLBO Abayarathna. The
objective of the research was to determine the frequency
of errors that occurred during DPTs and to minimize the
further occurrence. Researches were collected 250 DPTs
and assessed for presence of errors in categories including
patient preparation, patient positioning, patient motion,
handling and processing and exposure. There were only
22 (9%) error-free DPTs. Two hundred and twenty-eight
(91%) radiographs had errors. Six percent of the DPTs
were unacceptable and had to be re-taken. Fifty-nine per-
cent images had patient positioning errors being the most
frequent error. Thirty one percent of errors were due to
processing and handling followed by 7% of errors and 3%
of errors due to patient preparation and exposure errors
respectively. There was no patient motion error during ex-
posure. The researchers concluded that the frequencies of
errors in conventional DPTs were relatively high in this
study with patient positioning being the most frequent er-
ror. Professor Mark Mcentee, University of Sydney stated
that the reviewer’s sensitivity should be checked by giving
them set of radiographs prior to the study. The study on
knowledge on breast cancers amongst females in Katu-
wawela North, Boralesgamuwa, Sri Lanka was presented
by AH Somathilake, CM Hettiarachchi, HS Niroshani, SH
Egodage. The aim of the research was to determine the
level of awareness of breast cancers among females by ex-
ploring their knowledge on risk factors, Breast Self-Exam-
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
ination (BSE) and screening modalities of breast cancers
and to find out the correlation of selected socioeconomic
characteristics with the knowledge on breast cancers. The
study concluded that an increase in the education level of
the participants was significantly associated with the in-
crease in the overall awareness level on breast cancers and
increase in the economic status of the participants also
has a significant relationship with the increase in the level
of awareness on breast cancer. Furthermore, it was found
that there is a negative relationship with the increase in age
and the level of awareness on breast cancer. This study has
depicted that majority of respondents are aware of breast
cancer as a disease entity, but their knowledge and under-
standing of the disease is very low. Therefore, a breast can-
cer awareness campaign is essential to improve the knowl-
edge on breast cancers. Professor Mark McEntee discussed
about the misunderstandings about the breast cancers
amongst the women in Katuwawela area. Dr. Kottahachchi
suggested to categorize the participants age in to groups
and bettering to find level of awareness on breast cancer in
different age groups. The study on Radiation dose received
to thyroid glands of patients undergoing mammography;
a comparison of Cranio Caudal (CC) and Medio Lateral
Oblique (MLO) views was presented by GGC Perera, WSS
Jayasinghe, WBC Nisansala, DDN Wimalarathna, GLS
Galgamuwa, G Senanayake. The main objective of this
study was to determine the association of Entrance Surface
Dose (ESD) for thyroid between the Cranio Caudal (CC)
and Medio Lateral Oblique (MLO) projections. Study has
been revealed that ESD for thyroid gland from the pro-
jections of MLO is higher than CC. In addition, the com-
pression of breast thickness is significantly associated with
ESD values of the thyroid gland. Professor Mark McEntee
stated the other methods available for dose reduction of
the thyroid gland during mammography. The study on pa-
tients’ perceptions and knowledge regarding to the nucle-
ar imaging examinations at a selected hospital in Western
Province, Sri Lanka presented by DMPR Dissanayaka, N
10TH INTERNATIONAL RESEARCH CONFERENCE 925
PROCEEDINGS
Jayabhaskar, P Sathyathas, JMC Udugama. The aim of this
study was to find out the patient’s perceptions and knowl-
edge about nuclear imaging procedures. The researchers
concluded that the median score was lower (30) for total
100 marks and majority of the patients (77%) not aware
about nuclear imaging procedures by the time of prescrip-
tion. Chi-square test showed that there were significant
differences in the perception and knowledge with patient’s
type of occupation (P=0.000), residence (P=0.000) and
educational background (P=0.000). Mann-Whitney test
926 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
showed that health related occupied patients had higher
knowledge (66.956±15.056) compared with non-health
related occupied patients (32.967±14.774). There were
differences of perception and knowledge regarding to the
nuclear imaging examinations between the urban (38.642
± 18.317) and rural (25.556 ± 15.056) patients and among
different educational categories. Majority of patients had
lack of perceptions and knowledge regarding to the nucle-
ar imaging examinations and procedures.
 ISBN number: 978-955-0301-41-6
Technical Session II
Medical Laboratory Sciences The session comprised of
research studies conducted by undergraduates of Medical
Laboratory Sciences. The session was chaired by Vidya
Jothi Senior Professor Resvi Sheriff who is currently at-
tached to the Department of Clinical Sciences, Faculty
of Medicine, KDU. Dr Chandrika Nanayakkara, a Senior
Lecturer attached to the Department of Plant Sciences,
University of Colombo, Dr. Inoka Uluwaduge, a Senior
Lecturer and the former coordinator of the Medical Lab-
oratory Sciences Degree programme University of Sri
Jayewardenepura and Dr Prasanna Galhena, Head, De-
partment of Biochemistry and Clinical Chemistry, Faculty
of Medicine, University of Colombo served as the Judges
in the session. Out of five presentations amongst the six
papers presented were delivered by five Medical Labora-
tory Sciences undergraduates of intake 30. Among them
Ms. CK Nandasena presented her research titled “Asso-
ciation between changes in liver transaminases and full
blood count among patients with RT-PCR positive Den-
gue and RT-PCR negative other febrile illnesses admitted
to Teaching Hospital-Peradeniya, Sri Lanka. The impor-
tance of usage of liver transaminases in patients suspected
for Dengue infection was highlighted in the presentation.
Comments by the audience suggested to confirm the RT-
PCR negative patients among the sample in order deliver
a sound outcome. Ms GGPN Kulasooriya presented her
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
paper on prediction of the presence of microalbuminuria
by analyzing total urine protein to creatinine ratio in di-
abetic nephropathy patients in District General Hospital,
Ampara, Sri Lanka. The research on identification and
prevalence of risk factors in bacteraemic patients at Apek-
sha Hospital-Maharagama, Sri Lanka was presented by
Ms. SLA Lenaduwe. Ms JASN Jayasinghe presented her
findings on association of hypertension and renal impair-
ment with different dietary patterns of apparently healthy
individuals in area of medical of officer of Health, Dompe,
Sri Lanka. The reasons for lack of significant elevation of
triglycerides in subjects consumed fatty foods were ques-
tioned by the Chairperson. The research titled correlation
between anemia and renal function among type 2 diabetes
mellitus patients in General Hospital Ampara was pre-
sented by Ms EMDT Edirisinghe. The lack of investigation
on other potential factors that could cause anemia in pa-
tients were highlighted by the Judges. Mr. GK Wijesinghe,
MLS graduate from the University of Sri Jayewardenepura
presented his paper on effect of laboratory culture media,
citrate encapsulated and Curcumin encapsulated layered
double hydroxides on in-vitro Pseudomonas aeruginosa
biofilm growth. It was suggested to look for any methods
to get rid of biofilm formation in future research by the
judges.
10TH INTERNATIONAL RESEARCH CONFERENCE 927
PROCEEDINGS
Technical Session III
Physiotherapy & Occupational Health The session theme
was Physiotherapy and Occupational Health. This session
was chaired by Dr. Chandana Karunathilake. He is a senior
lecturer in Pathology in Faculty of Medicine in General Sir
John Kotelawala Defence University. He obtained in
MBBS and Masters in Surgery in Sri Lanka and he is a
member of Royal College of Surgeons. While serving as a
consultant Orthopaedic surgeon over 19 years he has
engaged in Research in the areas of Orthopaedics and
Road Traffic Trauma. He has won the gold medal for
College of Surgeon oration in 2014. The judge panel of this
technical session consisted of three experts in the area of
Physiotherapy. They were Dr. Lakshman Edirisinghe-
Director, Institute of Sports Medicine- Sri Lanka, Dr.
Nishanth Kumarasinghe- Senior Lecturer in Anatomy in
Faculty of Medicine in General Sir John Kotelawala
Defence University, and Mr. TDMSB Dasanayake, Senior
Lecturer in Physiotherapy- Department of Allied Health
Sciences, Faculty of Medicine- University of Colombo. Six
speakers presented in this technical session. Firstly Mr.
MHH Sandaruwan- Department of Allied Health
Sciences, Faculty of Medicine, University of Colombo, Sri
Lanka presented under the topic of “Determinants of
Balance Confidence (BC) And Standing Balance
Performance (SBP) in Stroke survivors with Hemiplegia”.
He concluded by their research study that dominance of
affected side influences both BC and SBP. Age and gender
showed significant associations with BC and SBP
respectively. BC and SBP should be included in assessment
and rehabilitation of stroke survivors with residual
hemiplegia due to their high correlation. During the
question and answer time; judge panel questioned whether
this study participants were hemiplegic (paralyzed) or
hemiparetic (only weakness). Presenter answered that he
took only paralyzed patients at the beginning of the study.
The audience asked about the BMI of the study participants
and the speaker said BMI was not excluded and that he has
included all the people with various BMIs in the sample.
Judge panel inquired about the reason for BC had a
significant negative correlation with age that is not similar
to literature. Chairperson inquired presenter whether any
extreme figures of age were recorded. Yet presenter said he
did not find such extreme of age. Second the presentation
was on “Comparison of Balance Deficiency in Chronic
Low Back Pain Patients (CLBPPS) and Healthy Persons
928 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
(HPS)” by Miss. APGA Arunodanie from Department of
Allied Health Sciences, Faculty of Medicine, University of
Colombo, Sri Lanka. The study concluded that CLBPPs
had impairment in balance when compared with HPs.
Intensity of pain was negatively correlated with balance.
Inclusion of balance exercises to the treatment procedure
of CLBPPs is important. During the question and answer
session, the audience inquired about the association on
dominant side and the gait with LBP to which the speaker
answered that those factors were not analysed in her study.
The judge panel said the leg length discrepancy and SI
joint pathology are the main contributing factors for
balance and LBP and suggested to include such factors as
well. And one of the judges inquired about the exclusion
criteria as she excluded a lot of medical condition like
labyrinthitis by questioning by the participants. The judge
said it is not possible to diagnose those conditions only by
asking patients and it should be medically diagnosed or
should be taken from medical records. The third oral
presenter was Miss N Priyadarshani- Allied Health
Sciences Degree Programme, Faculty of Medicine,
University of Ruhuna, Sri Lanka. She presented on
“Prevalence and related risk factors of recurrent falls
among ambulatory community dwelling elderly in Galle
district, Sri Lanka”. In conclusion, she said in this study,
presence of diabetes mellitus and balance/gait problem or
foot abnormality were identified as risk factors for
recurrent falls. She said that this study emphasizes the
importance of adhering to precautionary measurements
in preventing recurrent falls among elderly to enhance the
quality of life and facilitate healthier aging with minimal
complications. During the question and answer session
the judge panel said since she had done the study in 10
divisional secretarial area only, it should not mention in
the topic as in Galle district and to change the topic as
divisional secretarial area. Further they inquired about a
result on females are more risk on falls than males. They
highlighted since she has obtained more females than
males for the study, the results figure too might favour in
females. The other question was on environmental and
behavioural factors act on risk factors for falls as results
were different with literature. They discussed on
environmental and behavioural factors as access to
bathroom and access to washing area usually act as risk
factors for falls. The panel of judges suggested it is better to
 ISBN number: 978-955-0301-41-6
study on association of falls with hyperglycemic status and
diabetes polyneuropathy rather than study on association
of falls with presence of diabetes mellitus. Audience
inquired the association of falls with poor vision yet the
study had not analysed that aspect. The chairperson
highlighted the importance of using qualitative data to
increase the validity and reliability of the study. Fourthly,
Mrs. WGRMS Nandakumara from Unit of Allied Health
Sciences, Faculty of Medicine, University of Jaffna, Sri
Lanka presented on “Factors influencing on quality of
work-life of the nurses working in teaching hospital Jaffna,
Sri Lanka”. She concluded that nearly more than half of the
population showing their dissatisfaction on their quality
of work life, even though in this study only two factors
were showing the relationship with the quality of the work
life. Thus, there is need of conducting a qualitative in-
depth interview research approach to identify the more
factors, and this may help to take necessary action to
improve the quality work of life of nurses in Teaching
Hospital, Jaffna. The panel of judges inquired about the
questionnaire since she has used an Indian one which was
not validated in Sri Lanka but she has pretested that
questionnaire with 20 participants. They discussed the
most significant work-related factors were not identified
because of using the Indian questionnaire which was not
validated. Suggestion was to include a qualitative part to
the study. Fifth Speaker was Miss. RMHM Rathnayaka
from Department of Nursing and Midwifery, Faculty of
Allied Health Sciences, General Sir John Kotelawala
Defence University, Sri Lanka. The title was on “Assessment
of awareness of three-wheeler drivers on first aid when
handling and transporting traumatic victims following
road traffic accidents in the Colombo district, Sri Lanka”.
In conclusion, this study showed they have considerable
knowledge on first aid. So, proper awareness programs,
including first aid in school curriculum and making
people aware on first aid through media will assist the
three-wheeler drivers and lay people to perform immediate
lifesaving activities thoroughly and effectively.
Chairperson commented on discussion of the study with
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
data as she used 2012 data. The panel of judges questioned
on how to quantify overall knowledge and she described
about the questionnaire. Audience asked about the
awareness on study participants on measures to prevent
bleeding and transporting the injured people. She
answered that most of the three wheel drivers knew the
importance of transporting injured people with cervical
collar and lumbar board but they are not practicing due to
practical difficulties. Then chairperson discussed on
regarding the importance of correct way of transport
traumatic victims following road traffic accidents to
prevent further injuries. He said more than 50% of Spinal
Cord injury (SCI) is secondary and also most of the
traumatic victims in Sri Lanka are transferred to the
hospital through three wheelers. He revealed that recently
a pilot study was conducted to introduce a practical
method of transport victims in three wheelers such as
keep knee hip bended and back is straight on seat. The
final speaker of this session was Miss. K.I.G. Hevawitharana
from Institute for Combinatorial Advanced Research and
Education, General Sir John Kotelawala Defence
University, Ratmalana, Sri Lanka under the topic of “Risk
factor distribution among people affected with the Chronic
Kidney Disease (CKD) In Padaviya, Sri Lanka”. She
revealed that the chronic kidney disease (CKD) remains at
epidemic proportions in north central province of Sri
Lanka where majority of the patients are reported to be of
unknown etiology (CKDu). The results suggest that 1) the
prevalence of the disease with a history of initiation risk
factor could be substantial so that the risk factor
distribution deserves attention, and 2) risk factor
management should prove useful in improving health of
the CKD/ CKDu patients. The judge panel inquired about
the literature on her result as majority is CKDu as it
contributed as a minor contribution in literature. Further
judge panel suggested reporting is not adequate in a
scientific forum and should be more scientific.
10TH INTERNATIONAL RESEARCH CONFERENCE 929
PROCEEDINGS
Technical session IV - Pharmacy
The Pharmacy session was chaired by prof. BMR Fernan-
dopulle, a professor of clinical pharmacology and thera-
peutics of KDU. She is also a member of several national
expert committees of the Ministry of Health, Sri Lanka
including the Drug Evaluation Committee of the Nation-
al Medicines Regulatory Authority, Technical Advisory
Committees of Malaria, Filariasis and TB. The chair was
introduced by Dr. Darshana Kottahachchi, Acting Dean
of FAHS, KDU. Five research papers were presented at
the session by the authors SM Abdulla, MGS Malkanthi,
R Sarveswaran, Dr WASS Weerakoon and DVGC Priya-
darshani respectively. Dr AK Chandana, Senior Lectur-
er, Department of Basic Sciences, KDU, Werahara and
Dr. Gayani Abayaweera, Senior lecturer, Department of
Chemistry, University of Colombo were the members
of the judge panel. Each presenter was given 10 minutes
to present their research findings followed by 4 minutes
question/answer session. The session commenced with
the presentation titled “Computational study of angioten-
sin converting enzyme and renin with phoenicanthusine”
presented by SM Abdulla from Industrial Technology
Institute, Colombo. She talked about in silico approach
which allows a faster and cheaper identification of promis-
ing drug candidates by the virtual screening of compound
databases. She mentioned in her study that, Sri Lankan
natural product database was virtually screened against
ACE and Renin. Potent hit has been put through refined
docking using different algorithms and an advanced scor-
ing function has been used to filter the best results. Her
study showed that, for both ACE and renin, Phoenican-
thusine exhibits favourable results. Moreover, she said
that Phoenicanthusine is an endemic natural product of
Sri Lanka, and is isolated from the stem bark of Phoen-
icanthusobliqua. Ms. MGS Malkanthi from Department
of Nursing, Faculty of Health Sciences, Open University
of Sri Lanka presented about the factors related to read-
mission of patients with schizophrenia. Objective of her
study was to examine the factors related to readmission
of patients with schizophrenia at National Institute of
Mental Health (NIMH), Angoda, Sri Lanka. She men-
tioned that Demographic, socio economic, disease related
and cultural factors related to readmission were collect-
ed from parents or guardians of the patients diagnosed as
930 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
schizophrenia. The results of the study highlighted that,
readmission of patients with schizophrenia were mainly
caused by social and economic related factors in Sri Lanka.
Therefore, health education should be carried out to im-
prove knowledge and establish positive attitudes regarding
psychiatric disorders while improving the health facilities.
The third presenter, Ms. R. Sarveswaran was from the De-
partment of Allied Health Sciences, Faculty of Medical
Sciences, University of Sri Jayewardenepura, Sri Lanka.
Her presentation topic was “Toxicity studies on aqueous
extract of flower and stalk of Aponogeton cryspus in rats”.
The aim of her study was to evaluate the short term and
sub chronic toxicity of flower and stalk of optimum ef-
fective dose for hypoglycaemic activity of A. cryspus in
Wistar rats. Rats in the test group have been orally treated
with the aqueous crude extract of combination of flowers
and stalk of A. cryspus and control group has been given
distilled water for consecutive 14 days for evaluating the
short-term toxicity and further for 28 consecutive days
for sub chronic toxicity. The study showed that the aque-
ous extract of flowers and stalks of Aponogeton cryspus
does not exert any toxic effects at a dose of 90 mg/kg. The
presentation titled “Chronic anti-inflammatory effect of
Sudarshana suspension on adjuvant-induced arthritis in
rats” was presented by Dr. WASS Weerakoon from De-
partment of Ayurveda Paediatrics, Institute of Indigenous
Medicine, University of Colombo, Sri Lanka. She talked
about Sudarshana powder (SP) which is a very effective
anti-pyretic Ayurvedic preparation, extensively used in
Sri Lanka as well as in India. The aim of her study was to
evaluate the effect of Sudarshana suspension (SS) on the
progression of adjuvant-induced arthritis in rats. Further,
she mentioned that following induction of arthritis, daily
oral treatment was started on day 14 and continued up to
day 28. Body weights (BW), hind paw ankle joint thick-
ness (AJT) and foot pad thickness (FPT) were measured
in all animals. Treatment with SS and standard drug Cele-
coxib in the arthritic animals produced significant reduc-
tions (p<0.001) in FPT, AJT, WBC count, reduction of er-
ythema and oedema in the ankle joints and foot pad of the
AIA rats. Further, the BW was normalized. She concluded
that SS possesses anti-inflammatory effects on arthritis.
The final presentation of the session was “Risk of diabetes
 ISBN number: 978-955-0301-41-6
mellitus in Ampara area using a risk prediction model: a
case control study” presented by Ms. DVGC Priyadarshani
from Department of Medical Laboratory Sciences, Gener-
al Sir John Kotelawala Defence University. Her study fo-
cused on identification of the family history and lifestyle
related risk factors with their proportionate contribution
to Diabetes Mellitus (DM) by an estimated predictive risk
value through a case control- study. She has collected socio
demographic, anthropomorphic and lifestyle related risk
factors with medical history from the DM patients and
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
controls at the Ampara General Hospital in Sri Lanka. She
concluded that the family history of DM and non -occupa-
tion could play a main role in development of DM in rela-
tion to Ampara area. At the end of the session certificates
were awarded to presenters and judges and a memento
was presented to Prof. MBR Fernandopulle.
10TH INTERNATIONAL RESEARCH CONFERENCE 931
PROCEEDINGS
Parallel Session
Nursing & Midwifery I
The first technical session of Nursing & Midwifery of
the Faculty of Allied Health Sciences was chaired by Dr
Namal Wijesinghe, Senior Lecturer in Medicine and Head
of the Department of Clinical Medicine, Faculty of Medi-
cine, General Sir John Kotelawala Defence University. The
panel of judges of the session were Dr Rasika Jayasekara,
Senior Lecturer in Nursing & Midwifery, School of Nurs-
ing & Midwifery, University of South Australia. Dr. SSP
Warnakulasuriya and Ms SMKS Senevirathna, Senior Lec-
turers in Department of Allied Health Sciences, Faculty of
Medical Sciences University of Sri Jayewardenepura. The
session consisted of six oral presentations and the first pre-
sentation was done by Mr APGC Prasad, on ‘Knowledge,
attitudes and practices towards childhood immunization
among mothers attending Child Welfare Clinics in Med-
ical Officer of Health area, Ratmalana, Sri Lanka’. Chair
congratulated the first speaker. A comment raised by the
Chair was to do a religious comparison of the above study.
The next oral presentation was done by Ms. K Abhayas-
inghe on Scientometric mapping of mental health research
publications in India and Sri Lanka, Dr Namal Wijesing-
he commented on the study and stated the importance
of including the number of papers published per million
of the population to enhance the validity of the findings.
Dr Rasika Jayasekara appreciate her study and expressed
further about the databases available for facilitate the sys-
tematic review studies. Ms KMSDK Dissanayaka, from
Department of Nursing & Midwifery, Faculty of Allied
Health Sciences did the presentation on her findings on
“Knowledge, Attitudes and Practices (KAP) about Dengue
prevention among residents in Rathmalana MOH area”.
Chair and the judge panel was stated it as a timely topic
to be addressed. Dr SSP Warnakulasuriya commented on
932 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
her sampling method. Audience commented that the same
study should be carried out among the administrative
people in public & Government places. The fourth oral
presentation was drawn the attention from the audience
since it was done by Nigerian student Ms EO Shobowale
who is attached to Department of Nursing & Midwifery,
Faculty of Allied Health Sciences. It was on “A review of
sickle cell disease in Nigeria and the way forward”. Ms
HHS Kaushalya presented on the title on “Post-Traumatic
Stress symptoms among adolescents, who were exposed
to Ammunition Depot Blast; Salawa, Sri Lanka”. Most of
the audience were interestingly gave their positive com-
ments on the study. Dr Rasika Jayasekara commented that
the importance of the continuous follow-up of the study
subjects to minimize the risk of developing PTSD among
them in future. The speaker of the final presentation of the
second technical session on Nursing was done by Ms.BS
Chathurika, title on “Knowledge and Depression literacy
among G.C.E. Advanced Level students in Kalutara Edu-
cational Zone, Sri Lanka” Dr Rasika Jayasekara stated the
importance of assessing the prevalence of the depression
among the students. At the end of the six oral presenta-
tions the chairperson, Dr. Namal Wijesinghe commented
that as necessity of including male population as he ob-
served that the majority sample was consisted with female
among many of the studies. He encouraged all the other
participants to come up with more valuable findings in
next International research conference of the KDU. Final-
ly, he congratulated the speakers again and stated that the
session was very interesting as well as an interactive one
and he expressed his sincere thanks to the panel of judges
& the all participants.
 ISBN number: 978-955-0301-41-6
Parallel Session
Nursing & Midwifery II
The second technical session of Nursing & Midwifery of
the Faculty of Allied Health Sciences was chaired by Dr
GDI Rodrigo, Senior Lecturer in Paediatrics and the head
of the department of Medical Education, Faculty of Medi-
cine, General Sir John Kotelawala Defence University. The
panel of judges of the session were Dr Rasika Jayasekara,
Senior Lecturer in Nursing & Midwifery, School of Nurs-
ing & Midwifery, University of South Australia. Dr. AATD
Amarasekara, and Dr MKDL Meegoda, Senior Lecturers
in Department of Allied Health Sciences, Faculty of Medi-
cal Sciences University of Sri Jayewardenepura. The session
consisted of five oral presentations and the first presenta-
tion was done by Ms RADS Udayangi, from Department
of Nursing & Midwifery, Faculty of Allied Health Scienc-
es, KDU, on “Knowledge and attitudes of antenatal moth-
ers regarding neonatal care in a selected Medical Officer
of Health area, Sri Lanka”. A comment was raised by the
Chair asking the reasons for thinking that the mothers did
not have knowledge regarding immunization and was an-
swered as that these mothers didn’t answer the questions
regarding the age gaps of immunization, types of vaccines
given to the babies and the importance of those vaccines.
The Chair further stated that it must be due to the frequent
changes of the EPI schedule in Sri Lanka. The next oral
presentation was done by Ms. KSS Kaushalya from the
Allied Health Sciences Degree Program, Faculty of Med-
icine, University of Ruhuna, on “Domestic Violence Ex-
posure of Married Women in Isurumuniya PHM Area in
Anuradhapura, Sri Lanka”. Chair congratulated the speak-
er on the study, as it is a difficult area to do a research and
challenging to deal with the knowledge level of those peo-
ple. Dr SSP Warnakulasuriya commented on her defini-
tions about the criteria of violence mentioned in her study.
Dr Rasika Jayasekara stated that it was a surprise to see the
collusion of the knowledge level of these women and asked
about any support system to help them. The Chair stat-
ed that ultimately a research should have some value and
asked the speaker to try and put the findings of her study
in to action to help those poor women through police and
empower them to have good knowledge regarding domes-
tic violence. The third presenter was Ms RDUP Sugatha-
pala from Department of Nursing & Midwifery, Faculty of
Allied Health Sciences, KDU. She did the presentation on
her findings on “Patterns of healthcare seeking behaviour
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
during bouts of Upper Respiratory Tract Illnesses among
preschool children in Kurunegala Municipal Council
Area, Sri Lanka”. Dr Rasika Jayasekara commented on
her data collecting method and the chair wanted to know
how she differentiate Lower respiratory tract infections
(LRTI) and Upper respiratory tract infections (URTI) and
the usage of antibiotics. The fourth oral presentation was
done by Ms N Rathnayake from Allied Health Sciences
Degree Programme, Department of Physiology, Faculty of
Medicine, University of Ruhuna. Her topic was “Impact
of socio-demographic status on quality of life of pre-and
postmenopausal women in Galle, Sri Lanka”. The chair
stated that when interpreting the findings regarding the
quality of life of postmenopausal women, we need to be
very careful since there are many confounding variables.
For an example, she raised a question as being in which
basis that we can say people with high income have more
quality of life rather than people with low income. Ms
JAD Iroshika Kumari from Teaching Hospital, Kandy was
the final presenter of the session and she presented on the
title on “Depressive symptoms among Cancer Patients
Undergoing Chemotherapy at Teaching Hospital, Kan-
dy, Sri Lanka”. At the end of the presentation, Dr Rasika
Jayasekara asked whether she could find any relationship
between duration of cancer and depressive symptoms. Ms
Iroshika stated that statistically there was none. The judge
panel commented about the importance of having coun-
selling programmes and the continuous follow-up to help
these patients in the hospital and Ms Iroshika stated that
she had already emphasized this to the ward masters there
and in some wards they have already started to use music
therapy and religious activities in order to aid the psycho-
logical and spiritual needs of those cancer patients At the
end of the five oral presentations the chairperson, Dr GDI
Rodrigo highly appreciated the talent and the enthusiasm
showing by the nursing students to conduct researches
which is even more than medical students’ capacity. She
finally congratulated the speakers once again, stating that
the session was very interesting and expressed her sincere
thanks to the panel of judges and all the participants. The
session was concluded after distributing the certificates
to the presenters and tokens of appreciation for the Chair
and the judges.
10TH INTERNATIONAL RESEARCH CONFERENCE 933
PROCEEDINGS ISBN number: 978-955-0301-41-6

Plenary
Speeches

934 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

ALLIED HEALTH IN GLOBAL ENVIRONMENT:

CHALLENGES AND OPPORTUNITIES
Dr R Jayasekara
School of Nursing and Midwifery,
University of South Australia

Healthcare is changing dramatically due to the advance-
ment of medical sciences and technology, the abundance
of clinical research and the higher demands of consum-
ers. These changes represent a significant challenge to
the healthcare professions and education in terms of
maintaining the quality of services and preparing health-
care professionals for the future. Effective systems for
regulation, education, research and management are key
to strengthening the contribution of healthcare profes-
sions in order to achieve the required improvement in
health outcomes (ICN 2017; WHO 2017; World Bank
2017). In response, healthcare professionals’ education
and services are increasingly being restructured in many
countries to prepare them for contemporary and future
demands of healthcare. It is estimated that there are 43
million global health workers in 2015, including 9.8 mil-
lion physicians and 20.7 million nurses/midwives (WHO
2017). To maintain this enormous workforce requires
approximately 3,000 educational institutions, which pro-
duce approximately 1 million new healthcare graduates
each year (WHO 2016). It is essential to ensure that new
healthcare graduates achieved the required standards and
competencies to practise in order to provide safe and ef-
fective healthcare. However, many healthcare profession-
als in particular allied healthcare professionals face poor
working environments, lower-level wages, unsupportive
management and a lack of social recognition and career
development (WHO 2016). It is therefore essential that
governments and other institutions involved in human
resources for health should establish effective strategies
designed to ensure adequate policies in order to create safe
and effective healthcare system that are able to deal with
challenges presented today and the future.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 935
PROCEEDINGS
GENOMICS TO MOLECULAR
DIAGNOSTICS TO PERSONALIZED
MEDICINE
Prof R Dassanayake
Department of Chemistry, University of Colombo,
Genomics is the study of genomes, the blue print of life,
within the field of molecular biology, and this includes
genes, regulatory sequences, and other information con-
tained within the noncoding regions of an organism’s
DNA. The genomics is one of the fastest growing areas of
Molecular Biology with constant introduction of new-ad-
vanced robust technical platforms for genome sequencing
that facilitates generation of the alphabet of life, in a rapid
and economical manner to assist investigation of the for-
mula of the life forms. Given the importance of genomics
in molecular biology and its central role in determining
the fundamental operation of cellular processes, expansion
of knowledge in this area undoubtedly facilitates medical
advances in different areas of clinical interest that may
not have been possible otherwise. Molecular diagnostics,
essentially the analysis of DNA and RNA at the molecu-
lar level, is a fast-growing business, made possible by the
growing understanding of the human genome, which has
driven growth in the diagnostics industry. Molecular bi-
ology has held out the promise of transforming medicine
from a matter of serendipity to a rational pursuit ground-
ed in a fundamental understanding of the human genome
and the mechanisms of life. Molecular biology has begun
to infiltrate the practice of medicine and genomics is has-
tening these advances. Within next few decades, compre-
hensive genomics-based health care should be the norm.
936 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Scientists will understand the molecular foundation of
diseases, be able to prevent them in many cases and design
accurate, individualized therapies for illnesses. Molecular
diagnostics involves multiple technologies to identify ge-
netic variations in individual patients. These technologies
include PCR, FISH, hybrid capture, sequencing, microar-
rays etc., and has applications in testing for infectious dis-
eases and genetic disorders, early diagnosis of leukaemia
and cancers, screening of blood and for blood disorders,
DNA fingerprinting (e.g., paternity testing, forensic test-
ing), and also in microbiology, tissue typing, and food
pathogen detection testing. Further, it has applications in
prognostic maker detections related to diseases and per-
sonalized therapy identification that leads to the selection
of the most appropriate and effective drugs according to
the individuals’ genetic makeup. Finally, Genomics, Mo-
lecular Diagnostics and Personalized Medicine can be
considered as one of the greatest intellectual enterprises
of humankind that provides the impetus to fulfil the po-
tential of understanding life processes and utilizing them
to the advantage of humanity. Although, these are the fast-
est growing knowledge-based sectors even amongst our
neighbouring countries such as China, Japan, India, Ko-
rea, Singapore, Malaysia, Taiwan, Thailand etc., Sri Lanka
has yet to harness the full potential of these areas of Mo-
lecular life science.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

DOES DOSE MATTER?

Associate Prof Mark F McEntee
Faculty of Health Sciences, The University of Sydney, Australia

On the discovery of x-rays their use in medical imaging
was immediately obvious. After the initial excitement
about the wondrous “invisible light”, came a period of
realisation where many martyrs to radiation fell. Tech-
nology for x-ray imaging developed rapidly and quickly
became safer. Then there was a new expansion into new
imaging technologies fluoroscopy, CT, and more advanced
imaging techniques like angiography, and CT fluorosco-
py. Although radiation increased with each of these ex-
aminations that was not considered to be significant as
imaging examinations were few and far between. In the
1980s-computed tomography became increasingly more
common, and patients were more likely to be exposed to
a higher dose of radiation on a more frequent basis. The
fear of radiation from x-ray examinations began to creep
into the public concern. Articles began to appear in were
leading newspapers such as the New York Times hand-
wringing about the dangers of high-dose radiation, partic-
ularly to children. Now we have the combinations of these
powerful x-ray examinations of CT with positron emis-
sion tomography (PET). The patient has an injection of a
radioisotope at the same time as being exposed to ionising
electromagnetic radiation. One theory, that of hormesis,
would indicate that the body is well-prepared for these low
doses of radiation than the normal repair pathways will
be successful in repairing radiation damage at this level.
In fact, hormesis might indicate that an initial low dose of
radiation, could signal cells to prepare for upcoming high-
er dose of radiation. Alternatively, and perhaps more wor-
ryingly, the linear no threshold model indicates that no
dose, is a safe dose. Proponents of this model indicate that
modern radiation exposure to young children is leading to
significant increases in cancer development and mortality.
As we move forward with the use of PET/ CT, how wor-
ried should we be about the use of radiation particularly
in young children?

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 937
PROCEEDINGS
CHALLENGES AND WAY FORWARD OF
SPORTS AND EXERCISE MEDICINE IN
SRI LANKA
Dr L Edirisinghe
Director General, Institute of Sports Medicine,
Ministry of Sports, Colombo 07
In the background of a dramatic demographic transition
in Sri Lanka, with the increase in life expectancy and the
steady decline in fertility (2.13 in 2014), Sri Lanka is ag-
ing rapidly. It is projected that by 2020, 20% of Sri Lan-
ka’s population will have reached age 60 or over. Mortality
rate, currently at 6.2 per 1,000 population (CDR in 2016),
has been declining since independence according to pub-
lished statistics. Sri Lanka is also in an epidemiological
transition. Malaria, tuberculosis, Japanese encephalitis,
diarrhoea, vaccine preventable diseases in childhood and
acute respiratory infections are eradicated or controlled
effectively except for dengue, but cardiovascular and cere-
brovascular diseases, diabetes, and cancer are increasing
in an alarming trend. Tobacco, substance and alcohol
abuse have also increased in magnitude over the past two
decades. Child hood obesity is becoming a huge concern
while under nutrition is still a problem in certain sectors
of the society. Under these circumstances chronic diseas-
es and diseases related to mobility leading to poor quality
of life is becoming a public health problem in Sri Lanka.
Although there is a countrywide comprehensive net-
work of health care centres, hospitals and other medical
938 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
institutions, with about 57,000 hospital beds and a large
workforce engaged in curative and public health activities,
whether our health system is oriented enough to address
the emerging epidemiological pattern should be revisited
by the all relevant stakeholders at this important juncture.
In this backdrop there is a strong argument that Sport and
Exercise Medicine will have to adapt quickly to take ad-
vantage of the broader application of Exercise Medicine
in our system without delay. We will need to demonstrate
our effectiveness in this area and usher a solution to one
of the largest problems facing our health service today:
developing sustainable prevention, treatment and rehabil-
itation models for chronic disease and conditions related
to physical inactivity. This has enormous potential for ex-
pansion of SEM services, but not without its challenges as
you will see once it is going to be implemented. We should
be ambitious in our thinking when engaging with public
health, primary and secondary care and other organisa-
tions which may offer opportunities for the specialism to
be developed in Sri Lanka. Let’s join hands to develop a
stronger and a healthier nation.
ISBN number: 978-955-0301-41-6 PROCEEDINGS

Technical
Session
(ORAL & POSTER PRESENTATIONS)

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 939
PROCEEDINGS ISBN number: 978-955-0301-41-6

KNOWLEDGE, ATTITUDES AND

PRACTICES (KAP) ABOUT DENGUE
PREVENTION AMONG RESIDENTS IN
RATMALANA MEDICAL OFFICER OF
HEALTH AREA
KMSDK Disanayaka1#, KANG Kanchana1,
EGAP Nayanajith1, KPDB Samarasinghe1,
KGG Priyangika1 and PBV Navaratne2
1
Faculty of Allied Health Sciences, General Sir John Kotelawala Defence University, Sri Lanka
2
Faculty of Medicine, General Sir John Kotelawala Defence University, Sri Lanka
# srimathidisanayaka@gmail.com

Abstract – Dengue is contributing to considerable mor-
bidity and mortality rates in Sri Lanka and that is one of
the leading causes of illness in Western Province, where
Colombo district has the highest number. This study was
planned to assess KAP about dengue prevention among
residents in Ratmalana Medical Officer of Health (MOH)
area.
A descriptive cross sectional study was carried out during
April to June 2016 among residents in Attidiya North in
Ratmalana MOH area. An adult householder from ev-
ery third house of the area was selected for the study. A
pre-tested, structured, interviewer-administered ques-
tionnaire was used as the data collection tool which in-
cluded four sections for socio demographic data, knowl-
edge, attitudes and practices regarding dengue prevention.
Data was analysed using descriptive statistics and SPSS 23
was used as the statistical software.
There were 312 participants for the study and 104 (33.3%)
were males. Almost all the participants, (306, 98.1%) iden-
tified dengue fever as a mosquito borne disease while only
86 (27.6%) participants had correctly stated the features of
the mosquito. Nearly 95% of participants stated that clear
stagnant water is the breeding place of dengue mosqui-
toes. Further, 304 (97.4%) respondents had a positive atti-
tude that dengue patients have a chance for a full recovery
by immediate treatment.
There were 188 (65%) subjects who had participated in
cleaning activities within 3 months before data collection.
Further nearly three-fourth of them had cleaned inside the
house as a mosquito control activity but only 19.1% were
interested in cleaning outside the house and road to con-
trol mosquitoes.
The study concluded that the study subjects had satisfac-
tory knowledge and favourable attitudes regarding dengue
prevention, but their practices and participation in efforts
at combating dengue outside their own premises was poor.
Keywords - KAP, dengue prevention, residents, Ratmala-
na MOH area

940 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
I. INTRODUCTION
Dengue fever (DF) is a flu-like illness, if persists can
lead to the development of potentially fatal complication
known as dengue Haemorrhagic Fever (DHF) that affects
infants, young children and adults, but occasionally causes
death (Zameer et al, 2013). There are four types of virus-
es (DENV-1, DENV-2, DENV-3, DENV-4) belonging to
the Flaviviridae family. Infected Aedes aegypti and Aedes
albopictus female mosquitoes that transmit the virus feed
both indoors and outdoors during the day time. These
mosquitoes grow well in areas with standing water, includ-
ing puddles, water tanks, containers and old tires. Lack of
reliable sanitation and regular garbage collection also con-
tribute to the spread of the mosquitoes (WHO, 2016). It
causes life threatening dengue haemorrhagic fever whose
symptoms include headache, bleeding, low levels of blood
platelets, and metallic taste in the mouth, low blood pres-
sure, muscle joint pain and rashes (Shakil et al, 2015).
Today, severe dengue has become a leading cause of hos-
pitalization and death among children in most Asian and
Latin American countries (WHO, 2012). Preventable
diseases such as dengue have the potential to cause the
greatest mortality in a developing country like Pakistan
(Itrat et al, 2008). Sri Lanka is at high risk of dengue fe-
ver. Conditions are suitable for the infection spreading
and also arrive at epidemic proportion in different parts of
the country which is associated with increased morbidity
and mortality each year. There are two peak seasons for
the spreading of dengue epidemic in Sri Lanka each year:
one is from May to July; another is from October to De-
cember. Sri Lanka has confirmed dengue cases nationally.
Most affected cities include: Colombo, Gampaha, Kandy
and Kalutara (Epidemiological News Bulletin, 2016).
During the 1st quarter 2015, 12,035 cases of DF/ DHF and
27 deaths were reported when compared to 15,140 cases
of DF/DHF and 19 deaths reported during the 4th quarter
2014. The proportion of cases notified in January, Febru-
ary and March were 52.7%, 31.0% and 16.29% respectively
(Quarterly Epidemiological Bulletin, 2015). The highest
numbers of dengue cases were reported during the last
week of January, 2015. During the last 9 months of the year
2016, 41,923 suspected dengue cases have been reported
to the Epidemiology Unit from all over the island. Approx-
imately 51.29% of dengue cases were reported from the
Western Province. The highest numbers of dengue cases
were reported during the third week of June, 2016 (Epide-
miological Bulletin, 2016).
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Dengue is contributing to considerable morbidity and
mortality in Sri Lanka and is one of the leading causes of
illness in the Western province, where Colombo district
has the highest number. The Ministry of Health Sri Lan-
ka regards vector control as a gold standard for the pre-
vention of dengue outbreaks, although this effort is often
constrained due to operational blockages with the lack of
community support and involvement in vector control
programmes. There¬fore, it is essential to enhance knowl-
edge about the community’s perception about dengue
as well as their vector control practices before and after
the initiation of any community-based vector control
programmes. This study aimed to assess knowledge, atti-
tudes, and practices regarding dengue fever among a se-
lected population in Ratmalana Medical Officer of Health
(MOH) area, Sri Lanka.
The main objective of the study was to assess knowledge,
attitudes and practices about dengue prevention among
residents in a selected community of Ratmalana MOH
area. Specific objectives were to assess the knowledge re-
garding dengue fever, vector, breeding places, clinical fea-
tures and prevention, to determine the attitudes towards
dengue prevention and to evaluate practices regarding
dengue prevention among community setting.
II. METHODOLOGY
This study was a cross sectional descriptive study carried
out among residents of Attidiya North area of Ratmala-
na MOH area. Participants were selected for the study by
using systemic sampling method. One householder was
selected from one house. List of houses was taken from
Divisional Secretariat and every 3rd house was selected
randomly out of the list. A pre-tested, structured, inter-
viewer-administered questionnaire was used as the data
collection tool which included four sections for socio
demographic data, knowledge, attitudes and practices re-
garding dengue prevention. Data was analyzed using de-
scriptive statistics and SPSS 23 was used as the statistical
software.
According to the sample size calculation the sample size
was 323. But only 312 participated for the study due to 11
householders rejecting to participate. All residents in the
selected area were included to the study and children less
than 18 years and residents who were not willing to partic-
ipate for the study were excluded from the study.
10TH INTERNATIONAL RESEARCH CONFERENCE 941
PROCEEDINGS
III. RESULTS AND DISCUSSION
There were 312 participants in this study. The male
(33.3%) to female (66.7%) ratio was 1:2 and total popu-
lation was extended over suburban areas with different
educational, economical and socio-cultural backgrounds.
When considering the education level, out of 312 respon-
dents, there were 143 (53.2%) who had education up to
G.C.E. Ordinary Level (O/L) and 98 (31.4%) had educated
up to Advanced Level and 0.6% were without any formal
education. Most of the respondents (56.7%) belonged to
middle class who earned 10,000-30,000 LKR monthly. A
lesser number of respondents (2.6%) had less than 10,000
LKR monthly income. Generally, all these respondents
had a satisfactory educational level and monthly income.
According the findings of this study, most of the respon-
dents had satisfactory knowledge about dengue fever.
Nearly 98% subjects knew the exact reason for spreading
dengue, as mosquito bite. There was no significant asso-
ciation between the level of education and knowledge
about DF transmission (p value, 0.21) as shown in Table
1. A research carried out in suburban area of Sri Lanka has
reported that 76% respondents (Gunasekara et al, 2012),
88.5% in Malaysia (Hairi et al, 2003) and 93.5% respon-
dents in Laos (Nalongsack et al, 2009) were aware that DF
is transmitted by a mosquito vector.
Table 1; Knowledge on dengue transmission
vs. level of education
Level of Edu- Knowledge on DF transmission
cation
High Average Low
Primary 15 (65.2%) 6 (26.1%) 2 (8.7%)
Up to GCE O/L 108 (75.5%) 29 (20.3%) 6 (4.2%)
Above 122 (83.6%) 20 (13.7%) 4 (2.7%)
Majority (95.2%) stated that clear stagnant water is the
breeding place of dengue mosquitoes whereas 58 (18.6%)
and 55 (17.6%) stated dirty stagnant water and both clear
and dirty stagnant water as the breeding place respectively.
The score distribution of knowledge on breeding places is
shown in Table 2. The knowledge score of breeding places
was significantly associated with the level of education (p
value, 0. 032).
942 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Table 2; Knowledge about dengue mosquito
breeding places vs. level of education
Level of education Knowledge on dengue mosquito
breeding places
High Average Low
Primary 16 (69.6%) 6 (26.1%)
1 (4.3%)
Ordinary 17 (11.5%) 101 (70.6%) 25
(17.5%)
Above 21 (14.4%) 114 (78.1%) 11 (7.5%)
Nearly a third (32.3%) gave all the correct answers on early
symptoms of dengue fever and 26 (8%) participants gave
all the correct answer on DHF. In another study carried
out in Sri Lanka, participants had answered as high fever
(94%), vomiting (69%), muscle pain (85%), headache
(72%) and 25% participant had answered as rashes as
symptoms (Gunasekara et al, 2012). Another research
conducted in Laos found that fever was the mostly given
answer as the commonest symptom (75.2%), 3% bleeding
from the nose and 18.7% skin rashes (Nalongsack et
al, 2009). One study revealed that most common signs
and symptoms were fever and skin rashes (86.0%) as
given by the participants (Hairi et al, 2003). When
considering about dengue hemorrhagic fever (DHF),
most of the respondents had heard about the DHF and
78% respondents had known that vomit with blood as a
symptom of DHF.
Majority (85.2%) of participants agreed that DF can infect
a person more than once and almost all the subjects
(305, 97.8%) had attitude that anybody can get infected
irrespective of the age. Only 49 (15.7%) participants agreed
that the rainy season is the only season for outbreak of
dengue infection and a considerable amount (255, 81.7%)
had disagreed to that. Nearly one-fourth of the sample
(73, 23.4%) had negative attitude on controlling dengue
virus infection.
According to the results, 308 (98%) had positive attitude
regarding source reduction. Most (290, 90%) of the
subjects have positive attitudes towards seeking early
medical treatments when a family member fall sick and
the importance of having adequate bed rest and liquid
intake. Majority of respondents had obtained messages
about dengue from multiple sources like television
(98.4%), family members (95%), health centres (MOH
 ISBN number: 978-955-0301-41-6
office) (93.6%) and radio (90%). Other sources like
newspapers, friends, leaflets, schools and internet also had
provided information at a satisfactory level. Television was
the most cited source of information about dengue fever in
Pakistan, Malaysia and Sri Lanka (Itrat et al, 2008, Hairi et
al, 2003, Gunasekara et al, 2012).
In the present study, results revealed that most of the
respondents use measures to prevent mosquito bites. There
were only two people who did not use any measures from
the whole sample. Mosquito nets and fans were measures
which were used by most of the respondents. Fan usage
was highly increased in suburban areas of Sri Lanka. It can
be understood by comparing with Gunasekara and others
(2012).
When considering their participation in disease prevention
activities like cleaning, within three months before data
collection, it was 60% and it should be increase. Most of
respondents (140, 74%) had enthusiasm to clean inside
their home and 168 (89%) respondents have cleaned their
own garden. But, dengue mosquito breeding places can
occur in public places like roads, draining systems, public
buildings, schools and temples. Attention to the public
places was at a very low level according to the responses
(36, 19%). According to the answers of respondents, they
were seeking government mediation (11, 6%) and some
had taken legal action to clean public drainage systems
and illegal garbage discarded places.
IV. CONCLUSION
This study was conducted at a suburban area in the
Western Province and found out the knowledge, attitudes
and practices regarding dengue prevention. According
to the study, most of the participants were educated and
had satisfactory level of knowledge regarding dengue
prevention. They had considerable level of attitudes
regarding the prevention of dengue. Based on this study, it
is recommended that health education programmes should
be continued and intensified with emphasis on improving
the practices of the urban community and educating the
community on the Aedes mosquito especially its role in
the spread of the disease as well as the biting times and
breeding habits.
A change in the approach of health education programmes
is called for based on the findings that good knowledge has
not led to a good practice of control measures. Therefore,
health personnel should be trained to give appropriate
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
counselling in an effort to change certain deeply ingrained
traditional habits like domestic water storage without
proper cover. Through the study, it has been observed
that some facts should be modified within this area such
as infrastructure facilities (proper drainage system) and
waste disposal methods. The interaction between the
community and the health sector should be increased
for dengue prevention. The coverage of house-to-house
inspections should be improved by increasing manpower
and enhancing public participation.
The study concludes that the community was familiar
with dengue prevention, but their participation in efforts
combating dengue outside their own premises was poor.
Media plays an important role in conveying increased
awareness of effective control measures among the people.
According to the results, regular visits of health personnel
to the villages should be ensured and government
mediation is a major necessity which was found through
this study to get rid of dengue.
REFERENCES
Ahmed, N and Taneepanichskul, S (2008): Knowledge,
Attitude and Practice of dengue Fever Prevention among
the People in Males’, Maldives, J Health Res 22, pp 33-37.
Anonymous (2009). Dengue and dengue Haemorrhagic
Fever, Available from: www.who.int/mediacentre/
factsheets/fs117/en/ [Accessed on 04.08.2016].
Anonymous (2012). Dengue and Severe dengue: WHO,
Available from: http://www.who.int/mediacenture/
[Accessed on 22.09.2016].
Epidemiology Unit (2016): Epidemiological News
Bulletin. Ministry of Health, available from https://www.
moh.gov.sg/publications/ [Accessed on 05.10.2016].
Gnanapragasam, SR and Cooray, TMJA (2015): Time
series models to forecast dengue fever incidences in
Western Province of Sri Lanka. Proceeding of 8th
International Research Conference, November 2015, Sri
Lanka: Kotelawala Defence University, pp 50-56
Gunasekara, TDCP, Velathanthri, VGNS, Weerasekara,
MM, Fernando, SSN, Peelawattage, M, Guruge, D and
Fernando, S (2012): Knowledge, attitudes and practices
regarding dengue fever in a suburban community in Sri
Lanka, Galle Medical Journal 17(1), pp 10-17.
10TH INTERNATIONAL RESEARCH CONFERENCE 943
PROCEEDINGS
Hairi, F, Ong, C, Suhaimi, A, Tsung, TW, Ahmad, MAA,
Sundaraj, C and Soe, MM (2003): A Knowledge, Attitude
and Practice (KAP) Study on dengue among Selected
Rural Communities in the Kuala Kangsar District, Asia-
Pacific Journal of Public Health 15(1), pp 37-43.
Itrat, A, Khan, A, Javaid, S, Kamal, M, Khan, H, Javed,
S, Kalia, S, Khan, AH, Sethi, MI and Jehan, I (2008):
Knowledge, Awareness and Practices Regarding dengue
Fever among the Adult Population of dengue Hit
Cosmopolitan, PloS one 3(7), e2620.
Kumar, VA, Rajendran, R, Manavalan, R, Tewari, SC,
Arunachalam, N, Ayanar, K, Krishnamoorthi, R and
Tyagi, BK (2010): Studies on community knowledge and
behavior following a dengue epidemic in Chennai city,
Tamil Nadu, India, Tropical Biomedicine 27(2), pp 330-
336.
Nalongsack, S, Yoshida, Y, Morita, S, Sosouphanh, K and
Sakamoto, J (2009): Knowledge, Attitude and Practice
Regarding dengue among People in Pakse, Laos, Nagoya
journal of medical science 71(1-2), pp 29-37.
News bulletin (2015): Ministry of Health, Available
from:http://www.health.gov.lk/web/ [Accessed on
11.09.2016].
Quarterly Epidemiological Bulletin Sri Lanka, First
Quarter (2015), available from http://www.epid.gov.lk/
[Accessed on 5.09.2015].
944 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Shakil, KA, Anis, S and Alam, M (2015): Dengue
disease prediction using weka data mining tool, arXiv
preprint arXiv pp 1502-5167.Available at: http://pdfs.
semanticscholar.org/ [Accessed on 08.09.2016]
Syed, M, Saleem, T, Syeda, U, Habib, M, Zahid, R, Bashir,
A, Rabbani, M, Khalid, M, Iqbal, A, Rao, E, Shujja-ur-
Rehman and Saleem, S (2010): Knowledge, attitude and
practices regarding dengue fever among adults of high
and low socioeconomic groups, Journal of the Pakistan
Medical Association 60(3), pp 243-247.
Tissera, H, Amarasinghe, A, Gunasena, S, DeSilva, AD,
Yee, LW, Sessions, O, Muthukuda, C, Palihawadana,
P, Lohr, W, Byass, P, Gubler, DJ and Smith, AW (2016):
Laboratory-Enhanced dengue Sentinel Surveillance in
Colombo District, Sri Lanka: 2012-2014, PloS Negl Trop
Dis 10(2), e0004477.
Yoba, BC and Labrague, LJ (2013): dengue Knowledge
and Preventive Practices among Rural Residents in Samar
Province, Philippines, American journal of Public Health
Research 1(2), pp 47-52.
Zameer, M, Shuja, M, Ashraf, A, Mukhtar, N and
Ahmad, BM (2013): Knowledge, attitude and practices
study of dengue viral infection and its association with
environmental factors and health issues, Lahore Pakistan,
African Journal of Environmental Science and Technology
7(7), pp 711-714.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

KNOWLEDGE AND ATTITUDES

REGARDING THE EMERGENCY
CONTRACEPTIVE PILLS AMONG
THE DEFENCE UNIVERSITY
STUDENTS IN SRI LANKA
GJI Boteju1#, MABB Samaratunge1,
BMR Fernandopulle2 and MN Priyadarshanie1
Faculty of Allied Health Sciences, General Sir John Kotelawala Defence University, Sri Lanka.
1

2
Faculty of Medicine, General Sir John Kotelawala Defence University, Sri Lanka.
s
# isurikaboteju@gmail.com

Abstract - The emergency contraceptive pill (ECP) is a
method of hormonal contraception, which is indicated
after unprotected sexual intercourse when regular
contraceptive methods are not used. There is very scanty
amount of information available in Sri Lanka regarding
knowledge on the ECP. Hence, it is an important area to
study as there are higher rates of maternal deaths due to
unsafe abortions in Sri Lanka. ECP can be a strategy to
reduce them. The objective of this study is to investigate
the knowledge and attitudes regarding Emergency
Contraceptive Pills among Defence University students.
This is a descriptive cross sectional study conducted among
395 undergraduates of Kotelawala Defence University
(KDU) using self-administered questionnaires during the
period of July to October, 2016. Study participants were
selected by stratified random sampling. Data was entered
and analyzed using Minitab version 14.
Among the total participants (n = 395), only 69.1%
had heard of ECP. The main source of information was
reported as formal education 47.1%. From the total
participants, less than half, 45.5% students knew that ECP
was important in post rape and 20.2% knew that it was
important as a preferred contraceptive method. Among
the study participants, 42.8% agreed to the idea that if ECP
was readily available it might promote promiscuity among
student.
Most of the students, 42.5% thought that it was more
effective to take ECP soon after unprotected sexual
intercourse. Only 13.7% students knew the correct time
gap between the doses which was 12 hours. Higher
number of students, 57.5% has not received information
regarding side effects or problems that might get from
ECP. Meanwhile, 13.4% stated that ECP might prevent

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 945
PROCEEDINGS
STI’s and HIV. Among the total participants, only 25
(6.5%) had used ECP. From the participants who had used
ECP, 52% of them had experienced side effects. Only 23%
knew about side effects of ECP.
The current study results on influencing factors for
knowledge indicated gender had a significant effect on
every aspect of knowledge and attitudes on ECP while,
religion did not have a significant effect on any aspect.
Age, gender and year of study have a significant effect
on knowledge on ECP while gender, year of study, and
department has a significant effect on attitudes.
Although the general awareness on ECP was reasonably
good, majority did not have a sufficient knowledge on ECP
for an effective usage. Most of the attitudes on ECP were
towards positive.
Keywords - Emergency contraceptive pills, Knowledge,
Attitude, Defence University, Sri Lanka
I. INTRODUCTION
Emergency Contraception (EC) is a method of preventing
pregnancy soon after an unprotected sexual intercourse,
after a sexual assault or rape, if a condom breaks or a
diaphragm slips out of place, a woman forgets to take
birth control pills. EC has been available for more than 30
years, (ICEC, 2012) worldwide and for about 10 years in
Sri Lanka.
ECP is not a family planning method (UMM, 2012). It
can be taken to prevent pregnancy within 72 hours of
unprotected sexual intercourse (Byamugisha et al, 2006).
It was found that the sooner the first dose was taken,
the greater the effectiveness (Puri et al, 2007). The ECP,
available in Sri Lanka is a single pill containing 1.5mg of
levonorgestrel (progestogen-only contraceptives). These
pills are also known as Postino 1 and 2.
Many pregnancies are mistimed or unwanted. These
pregnancies may carry a high risk of morbidity and
mortality, particularly in settings where safe abortion is
not accessible or where quality obstetric services are not
available for those women continuing a pregnancy to term.
According to some unrecorded data, ECP known to be a
popular and common contraceptive method in Sri Lanka.
Even though, ECP is very popular in the community
setting, malpractice and misconceptions are the main
946 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
reasons for the poor usage of ECP (Parey, 2010). If the
community is aware regarding proper usage of ECP,
then it can further reduce the maternal mortality that is
associated with unsafe abortions.
Premarital sexual relationships have increased globally.
According to a recent research the global premarital birth
percentage was 40.7% in 2012-2014. According to the
above study, Asian unmarried mother rate is less when
compared to the other ethnic groups in the world and
it is 17% (Thalagala, 2012). A previous study conducted
among advanced level students in selected districts in Sri
Lanka had pointed out that 42.9% of students had engaged
in premarital sex and among them only 12.2% of students
had used contraceptives (Perera et al, 1999). Another
study conducted in Sri Lanka, pointed out that 22% of
out of school adolescents had engaged in heterosexual
relationships (Thalagala, 2004).
The consequences of premarital sexual relationships are
as follows; increased rate of premarital unwanted and
unexpected pregnancies, increased criminal abortion rates
(Puri et al, 2007), interruption of educational process,
socio-economic burden, as well as health issues such as
increased mortality and morbidity ratios mainly due to
criminal abortions (Thalagala, 2012).
A. Abortions in Sri Lanka and Contraceptive Knowledge
In the recent decades the number of induced abortion rate
has been increased in Sri Lanka (Abeykoon, 2009). Hence,
around 1 in 10 pregnancies end up as an unsafe abortion in
Sri Lanka (Thalagala, 2012). Unsafe abortion has become
the 2nd commonest cause of maternal mortality in 2010
(Atapattu, 2012). As abortions are illegal in Sri Lanka there
are no reliable statistical data to be found. Nevertheless,
attempts have been made to estimate the incidence of
induced abortions in Sri Lanka.
In a study conducted by UNFPA and Ministry of Health in
2009, emphasized the necessity of educational programs
regarding the effective use of contraceptive methods and
side effects of abortions in order to reduce abortion related
maternal mortality in Sri Lanka (Abeykoon, 2009). The
government has recommended contraceptive methods
to lower the abortion rates. A study conducted by Dr. N.
Thalagala revealed the low levels of reproductive health
and contraceptive knowledge among adolescents even
though they are sexually active (Thalagala, 2004).
 ISBN number: 978-955-0301-41-6
The inadequate formal health education regarding
contraception, given by schools, universities and
community health institutions results in the high rate
of unintended pregnancies in Sri Lanka (Kumar, 2013).
Furthermore, overall knowledge of contraceptive among
out of school adolescents is not up to satisfactory level in
Sri Lanka (Kumar, 2013).
Even though community health services target on married
couples on family planning, very less facilities and
information are provided to the youth who are at the verge
of risky sexual behaviours (Perera et al, 1999).
The aim of this study is to determine the knowledge and
attitudes of undergraduates; cadets and day scholars in the
Defence University regarding Emergency Contraceptive
Pills.
II. METHODOLOGY
Study design is descriptive cross sectional study. It was
carried out on both male and female local undergraduate
officer cadets and day scholars of Kotelawala Defence
University.
To obtain a representative sample, stratified random
sampling was applied and students were divided into
two stratum as health science and non health science
students. Sample size was 480 undergraduates (including
25% non-response rate). Data was collected using a self-
administered Questionnaire. The pilot study was carried
out on randomly selected 48 students (10% of the sample
size) in the Faculty of Allied Health Sciences.
Ethical clearance was obtained from the Ethical Review
Committee, Faculty of Medicine, General Sir John
Kotelawala Defence University. Data was entered and
analysed using Minitab version 14.
III. RESULTS
Total of 395 students answered the questionnaire (82%
response rate). Majority (89.1%) was between the age
group 18 to 24 years. Most of the students (97.2%) were
unmarried. Majority of the respondents (66.6%) were non-
medical students while the others (33.4%) medical. Most
of the students (38.2%) were second years. Maximum
number of the respondents 345 (87.3%) were Buddhist by
religion.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
A. Distribution of Knowledge on Emergency
Contraceptive Pills
Majority of respondents (69.1%) had ever heard of ECP.
Regarding their source of information about ECP from
the total population, 186 (47.1%) reported that they got
information through formal education; 177 (44.8%) got
information through friends and neighbours; 170 (43%)
got information from printed materials, media and
internet. Only 97 (24.6%) of participants got information
through health professionals.
Higher number of students, 286 (72.4%) knew that they
can obtain ECP from pharmacy. Most of the students,
172 (43.7%) said that there was no need of a prescription
to obtain ECP while 49(12.4%) said that they need a
prescription to obtain ECP.
Out of 395 students, 95 (24.6%) agreed with the idea that
ECP was effective when taken before sexual intercourse,
while majority 200 (51.7%) did not have any idea regarding
the effectiveness of ECP before taken sexual intercourse.
Most of the students, 167 (42.5%) thought that it was
more effective to take ECP soon after unprotected sexual
intercourse. However, 159(40.5%) stated that they did not
know whether it was more effective to take ECP soon after
unprotected sexual intercourse. Meanwhile, 122 (31.2%)
students were disagreed to the idea of ECP was effective
when taken 72 hours after unprotected sex; 52 (13.4%)
students agreed and 212 (54.6%) stated that they did not
know.
Only 54 (13.7%) students knew the correct time gap
between the doses which was 12 hours. Higher number
of students, 284 (71.9%) stated that ECP prevents a
pregnancy while 80 (20.2%) thought that ECP terminates
a pregnancy. Only, 123 (31.4%) students agreed with the
idea that ECP was effective more than natural methods of
contraception.
Majority of the students, 341 (88.3%) answered that they
had never used ECP while 25 (6.5%) students stated that
they had used. Majority of the students, 156 (39.5%) stated
that ECP should be taken soon after an unprotected sex
while 127 (32.1%) students answered that it was within 24
hrs. Meanwhile, 113 (28.6%) answered that ECP should be
taken within 72 hrs. 17 (4.3%) stated that it was after 72
hrs. and 12 (3%) stated that it was within 5 days. Majority
of the participants, 268 (67.8%) did not know that how
often that they can use ECP. Only 43 (10.9%) knew that
there should be at least 3 months gap while, 21 (5.3%)
10TH INTERNATIONAL RESEARCH CONFERENCE 947
PROCEEDINGS ISBN number: 978-955-0301-41-6

answered that they can use ECP weekly or monthly and 17 C. Investigating of Influencing Factors to Knowledge
(4.3%) answered that they can use it frequently. and Attitudes on ECP
From those who had used ECP, only 13 (52%) students
had experienced side effects. Higher number of students, Table 1: Association between the knowledge
227 (57.5%) have not received information regarding side
with selected socio-demographic variables.
effects or problems that might get from ECP. Only 31
(7.9%) students received information about precautions
Socio-demographic P- Value Significant Association
regarding side effects or problems of ECP. Many students variable
188 (47.6%) had mentioned that ECP might prevent future Age with ECP aware- 0.001
pregnancy, 64 (16.2%) thought that ECP was illegal and ness √
53 (13.4%) stated that ECP might prevent STI’s and HIV.
Gender with ECP 0.006
awareness √
B. Distribution of Attitudes on Emergency
Contraceptive Pills
Department of study 0.384
with ECP awareness x
Less than half, 180 (45.5%) students mentioned that ECP
was important in a situation like post rape, 177 (44.8%) Year of study with 0.003
answered that it was important as back up when condom ECP awareness √
breaks, 121 (30.6%) stated that it was important if oral
pills were forgotten and 80 (20.2%) stated that it was Religion with ECP 0.477
awareness x
important as a preferred contraceptive method. Majority
of the students, 142 (35.9%) was uncertain that whether
availability of ECP might promote promiscuity among
students. While, 169 (42.8%) agreed to the idea that if ECP √ There is a significant association
was readily available it might promote promiscuity among x There is no significant association
student.

Most of the students, 178 (45%) agreed to the idea that

It was revealed that there was an association between age
ECP should be easily accessible. Majority of students from
and attitudes on promiscuity (P-value: 0.006). But there
those who had an opinion that ECP should be low cost,
was no association between the attitudes on promiscuity
162 (40.9%) agreed to the idea. Minority of 75 (18.9%)
and gender (P-value: 0.983), department of study (P-val-
students answered that ECP should only be available for
ue: 0.569), year of study (P-value: 0.547) and religion
victims of rape while majority of the students, 179 (44.3%)
(P-value: 0.486).
answered that it should not. Majority of the respondents,
142 (38.4%) mentioned that ECP should be available
without prescription while 89 (22.5%) mentioned that it Table 2: Association between attitudes on
should not. To the idea of that ECP should be available future usage of ECP with selected
only for women over 18 years, majority of 145 (36.7%) socio-demographic variables.
agreed while others, 107 (27.1%) disagreed.
Socio-demographic P- Value Significant Association
Majority of the respondents, 228 (57.7%) had answered variable
that they might not use ECP in the future while 120 (30.4%)
Age 0.220 x
had answered that they might. Higher number of students,
Gender 0.013
170 (43%) had mentioned that they would recommend √
ECP to a friend or a relative if needed while 114 (28.8%)
0.002
had mentioned that they would not recommend. Department of
√
Study
Year of Study 0.045 √
Religion 0.178 x

948 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Table 2: Association between attitudes on
future recommendation of ECP to a friend
or relative with selected socio-demographic
variables.
Socio-demographic P- Value Significant Association
variable
Age 0.119 x
Gender 0.007 √
Department of 0.001 √ be introduced to them. Emergency contraceptive is one
Study
Year of Study 0.006 √
Religion 0.432 x According to present study, considerable number of
IV. DISCUSSION
Young adults from various areas in Sri Lanka enter the
University for their Tertiary Education. They are from
different socio - economic backgrounds and as Sri Lanka
is a multi-cultural society, they are from different religious
backgrounds as well. After entering for tertiary education
they experience enormous freedom therefore, it instigates
them to establish new relationships. Considerable
proportion of male and female school age adolescents in
Sri Lanka has exposed to risky sexual activities within
their secondary education (Perera et al, 1999)(Thalagala,
2004). Those adolescents enters the University for their
Tertiary Education therefore, university students have a
relatively higher chance of unprotected sexual intercourse.
Usually, relationships compel the couple to engage in
various sexual activities.
In Sri Lanka abortions are not legalized. Therefore, very
less number of surveys has been conducted on unsafe
abortions. From those limited surveys, it has been proved
that the rate of unsafe abortions have been gradually
increased from 2001 to 2008 from 8% to 13%. Unsafe
abortion has been a leading cause for the maternal
mortality during recent 8 years period (Kumar, 2013).
Maternal mortality is an important health indicator which
has a huge impact on measuring a country’s development.
In order to address this health burden, the Ministry of
Health has suggested introducing contraceptive methods
to the people and those contraceptive services at the
grass root level targets married women. Even though
it has implemented likewise, previous researches has
highlighted the low levels of awareness in contraceptive
methods (Kumar, 2013).
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Furthermore, youths as a group who are at the reproductive
age, they also have a risk for unintended pregnancies. Even
though youth is sexually active, they are not targeted for
providing adequate information regarding contraceptives.
Among those who engaged in premarital sexually
activities, very few has used contraceptives (Ratnayake,
2000). Due to the fact that young women engaged in pre-
marital sexual intercourse, it is better for them to have
adequate knowledge to prevent unintended pregnancies.
Therefore, recommended contraceptive methods should
method that can be introduced to them.
defence university students had heard about ECP, 69.1%
of the respondents. Studies done in Trinidad and Iran
respectively had reported 63.1% and 95.4% awareness
(Parey et al, 2010)(Delaram & Rafie, 2012). Similar studies
had done in Uganda and Ethiopia, the awareness on EC
was respectively, 45.1% and 84.2% (Byamugisha et al,
2006)(Ahamed et al, 2012).
Main source of information was reported as formal
education and friends and neighbours. Similar studies had
shown media, (Ahamed et al, 2012) medical professionals
(Parey et al, 2010) and friends (Byamugisha et al, 2006)
as their main source of information. Less than half of the
students (43.7%) knew that they do not need a prescription
to obtain ECP. Two- third of the participants knew that
they can obtain ECP from pharmacy. In parallel studies
pointed out that 78.3% (Ahamed et al, 2012) participants
said that they can obtain ECP from health institutions
while 65.7% (Parey et al, 2010) stated that they can obtain
ECP from the pharmacy.
Less than half (42.5%) of the students knew that ECP was
more effective when taken soon after unprotected sexual
intercourse. A study conducted in Trinidad reported that
62% knew ECP were more effective when taken soon after
unprotected sexual intercourse (Parey et al, 2010). Only
13.7% students knew the correct time gap of ECP which
was 12 hours. It seems that majority of the students did not
know the correct time gap between the doses. In parallel
studies in Iran and India noted that respectively 57.7%
(Delaram & Rafie, 2012) and 38.2% (Puri et al, 2007) knew
about the correct time gap.
Most of students 71.9% knew that ECP prevents a
pregnancy. Majority of the students did not know the
effectiveness of the ECP in preventing a pregnancy. In
addition to that, majority of the students did not know
10TH INTERNATIONAL RESEARCH CONFERENCE 949
PROCEEDINGS
whether ECP is more effective than natural methods.
Majority (67.8%) of the students did not know how often
they can use ECP. Only 6.5% students had ever used ECP.
From those who had used ECP 52% had experienced side
effects .While 57.5% had not received information on
side effects and problems of ECP. Only 7.9% had received
information about precautions for side effects of ECP.
Similar studies revealed 11.1% (Puri et al, 2007) did not
know about the side effects.
Less than half had mentioned that ECP might prevent
future pregnancies while in similar studies 31% (Parey et
al, 2010) stated that ECP might prevent future pregnancies.
In current study, 16.2% had answered that ECP was illegal.
Moreover, 13.4% stated that ECP might prevent STIs and
HIV.
In current study 45.5% of participants had answered that
ECP was important in a situation like post rape. Similar
studies were reported that 82.1% (Ahamed et al, 2012)
and 32.4% (Puri et al, 2007) respectively from Ethiopia
and India ECP was important in a situation like post rape.
Less than half 44.8% of the study population had answered
that ECP was important as a backup when condom breaks.
Above mentioned Ethiopian study reported that, 49.9%
(Ahamed et al, 2012) knew the worth of ECP as a backup
when condom breaks. In our study 30.9% had answered
that ECP was important when oral pills were forgotten.
Another study revealed that only 11.1% (Ahamed et al,
2012) thought that ECP was important when oral pills
were forgotten. Considerable proportion of respondents
20.2% reported ECP is suitable as preferred contraceptive
method whereas ECP was not recommended as a preferred
contraceptive method (ICEC,2012).
If ECP should be readily available, 45.8% of students
answered that it might promote promiscuity while 13.5%
disagreed and majority of the students were unsure.
Related studies showed similar outcomes (Parey et al,
2010)(Byamugisha et al, 2006). Present study revealed that
majority of the students, 138 (34.9%) agreed to the idea
that ECP should be easily accessible. Parey et al in 2010
revealed a similar finding. As our study results indicates
majority of the students (32.1%) had an opinion that ECP
should be low cost. This is in line with another study (Parey
et al, 2010). Very less number of students thought that ECP
should only be available for victims of rape, while majority
thought it should be available for all. Maximum number
of students agreed to the idea that ECP should be available
only for women who are over 18 years. This is similar with
the study conducted in Trinidad (Parey et al, 2010). The
950 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
present study revealed higher number of students (57.7%)
responded that they might use ECP in future. In addition
to that 43% respondents intended to recommend ECP to
a friend or relative. A similar study suggested that only
33.2% (Ahamed et al, 2012) respondents had an intention
to use or suggest to a friend or relative.
The current study results on influencing factors for
knowledge indicated gender had a significant effect on
every aspect of knowledge and attitudes on ECP while,
religion did not have a significant effect on any aspect.
In addition to that, age, gender and year of study have a
significant effect on knowledge on ECP. Apart from that,
it was found that gender, year of study, and department
played a major role as influencing factors on attitudes.
V. CONCLUSION
According to the present study finding it can be concluded,
the awareness on ECP is at a considerable level. However,
there were many misconceptions on knowledge, to
illustrate, most of the students stated that ECP can be used
as a preferred contraceptive method. Most of the attitudes
on ECP were towards positive. Even though, it was found
from the present study that undergraduates had heard
about ECP through formal education, they did not have a
sufficient knowledge on ECP.
Therefore, it is recommended that health professionals
should step in and interfere with current issue and
provide sufficient knowledge to youth. In addition to
that, undergraduates had negative attitudes towards ECP.
Therefore, it is necessary to change attitudes on ECP.
REFERENCE
Abeykoon, A T P L (2009): Estimates of Abortion Rates
in Sri Lanka using Bongaarts Model of Proximate Deter-
minants of Fertility. 1st ed. Colombo, UNFPA & Ministry
of Health
Ahmed, F A, Moussa, K M, Petterson, K O and Asamoah,
B O (2012): Assessing knowledge, attitude and practice of
emergency contraception: a cross-sectional study among
Ethiopian undergraduate female students. BMC Public
Health, 12(110).
Atapattu, H (2012): Sequel of unsafe abortion. In: 45th
Annual Scientific Sessions of the Sri Lanka College
of Obstetricians and Gynaecologists. Sri Lanka, 17th-
23rd October 2012: Sri Lanka Journal of Obstetrics &
Gynaecology, 34(1), p.31.
 ISBN number: 978-955-0301-41-6
British National Formulary (2011): BMJ Group and the
Royal Pharmaceutical Society of Great Britain. London,
61.
Byamugisha, J K, Mirembe, F M, Faxelid, E and Gemzell-
Danielsson, K (2006): Emergency Contraception and
Fertility awareness among University Students in Kampala,
Uganda. African Health Sciences, 6(4).
Delaram, M and Rafie, H (2012): Knowledge and Attitudes
of Emergency Contraception among Medical Science
Students at Shahrekord university of Medical Sciences in
Iran. Future of Medical Education Journal, 2(4).
International Consortium for Emergency Contraception
(2012): Emergency Contraceptive Pills, Emergency
Contraceptive Pills; Medical and Service Delivery
Guidelines, 3rdedition, Washington DC, Available at:
http://www.cecinfo.org/about/history/ [Accessed on
13.10.2016]
Kumar, R (2013): Abortion in Sri Lanka: The Double
Standard. American Journal of Public Health. 2013 March;
103(3): pp400–404.
Parey, B, Addison, L, Mark, J K, Maurice, B, Tripathi, V,
Wahid, S, Antoine, R and Sahai A (2010): Knowledge,
attitude and practice of emergency contraceptive pills
among tertiary level students in Trinidad: a cross-sectional
survey. West Indian Medical Journal, 59(6).
Perera B, Fonseka P and Nawaratne K (1999): Knowledge
attitudes and practice of sexual health among advanced
level students in five selected districts in Sri Lanka. UNFPA
and Ministry of Health funded dissertation. Faculty of
Medicine, University of Ruhuna, Sri Lanka.
Piaggio, G, von Hertzen, H, Grimes, D A and Van Look,
P F (1999): Timing of emergency contraception with
levonorgestrel or the Yuzpe regimen. Task force on
postovulatory methods of fertility regulation. Lancet
;353:721.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Puri, S, Bhatia, V, Swami, H M, Singh, A, Sehgal, A and
Kaur, A P (2007): Awareness of emergency contraception
among female college students in Chandigarh, India,
Indian Journal of Medical Science, 61(6), pp.338-346.
Rajapakse, L C (2000): Estimates of Induced Abortion
using RRT Technique (mimio), Colombo.
Rathnayake, K (2000): An Inquiry into the Knowledge,
Attitudes and Practiceson Reproductive Health Among
Undergraduates at the University of Ruhuna, University of
Ruhuna, Sri Lanka.
Statistic Brain Research Institute (2014): Premarital
Pregnancy Statistics Available at: http://www.
statisticbrain.com/premarital-pregnancy-statistics
[Accessed on 28.08.2015].
Thalagala, N (2004): National Survey on Emerging Issues
among Adolescents in Sri Lanka. Colombo, Sri Lanka:
United Nations Children’s Fund; 2004, Available at: http://
www.unicef.org/srilanka/Full_Report.pdf [Accessed on
13.10.2016].
Thalagala, N (2012): Health burden due to unsafe abortion.
In: 45th Annual Scientific Sessions of the Sri Lanka
College of Obstetricians and Gynaecologists. Sri Lanka,
17th-23rd October 2012. Sri Lanka Journal of Obstetrics
& Gynaecology, 34(1), p.30.
University of Maryland Medical Centre (2012): Emergency
Contraception, Available at: http://umm.edu/health/
medical/ency/articles/emergencycontraception [Accessed
on 28.08.2015].
10TH INTERNATIONAL RESEARCH CONFERENCE 951
PROCEEDINGS ISBN number: 978-955-0301-41-6

LIST OF REVIEWERS

Dr MKDL Meegoda Dr Nithushi Samaranayake

Dr Neil Fernando MA Siriwardhene

Dr SSP Warnakulasuriya Dr ACM Fahim

Dr AATD Amarasekara Dr RMCRR Gamage

RAD Upul Cosman Dr Arosha Dissanayake

Dr J Jeyasugiththan Dr Sampath Gunewardana

Dr M A J C Marasinghe Dr Aindralal Balasuriya

Dr UJMAL Jayasinghe Dr JMKB Jayasekara

Senior Prof Narada Warnasooriya Mr RD Widanagamange

Professor Hemantha Peiris Dr KMN Kumarasinghe

J Kottahachchi Prof RN Pathirana

Dr PP Rasika Perera Dr D Kottachchi

Dr CM Nanayakkara Dr J Balawardane

952 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6 PROCEEDINGS

Computing
“Resolutions for Challenges & Opportunities of
Global Dynamics through Digital Innovations”

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 953
PROCEEDINGS ISBN number: 978-955-0301-41-6

Plenary
Speeches

954 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

ICT TO IOT HOW TO KEEP UP WITH

FAST MOVING INFORMATION
TECHNOLOGIES
Prof Jay Rajasekera
Vice President, International University of Japan
jrr@iuj.ac.jp

The dawn of information age has driven us to in imaginable
innovations based on technology – innovations not only
of products and services, but also strategies and policies.
When one take into account of the time that telephone was
invented by Mr. Bell, it took almost 100 years to see major
changes to the way information was generated and used.
However, the emergence of computers, particularly, the
personal computer, and the Internet dawned an era of rapid
innovations resulting in iPhones and now IoTs (Internet
of Things). Everything one can possibly think of, from
autonomous diving, artificial intelligence, social media, to
GPS in space, the information has taken over our lives. So
far the “haves” or the so-called developed countries have
been the main beneficiaries of fast moving information
age. With predictions of billions of IoTs to populate the
world, there is a great danger of “IoT Divide” taking place
between “haves” and “have-nots”. This makes it vital for
countries such as Sri Lanka to plan new ICT strategies in
order to tap into reap the benefits of new information age.
The aim of this presentation is to highlight the strategies
and options that are still out there for countries such as Sri
Lanka to explore.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 955
PROCEEDINGS ISBN number: 978-955-0301-41-6

ONLINE LEARNING:
COMMON CHALLENGES FOR
INSTRUCTORS IN LARGE
ONLINE COURSES:
STRATEGIES TO MITIGATE STUDENT
AND INSTRUCTOR
FRUSTRATION
Dr (Eng.) Lalith Liyanage
Chief Executive Officer, Lanka Logistics and Technologies Ltd,
Ministry of Defence, Sri Lanka
# lalith@lltl.lk

Teaching in the online classroom is becoming
commonplace for instructors as universities seek to grow
enrolments and tap into unexplored markets. Many
instructors, however, are often unprepared for the nuances
of distance education and apprehensive about making
the transition to online learning. This presentation aims
to discuss basic principles of online learning, advantages
and disadvantages compared to face to face learning,
common challenges for instructors of high-enrolment
online courses. Those challenges align with the areas
students commonly consider as necessary for successful
online delivery. Using examples from large online classes
and the existing research on best practices in online
education, ways to minimize those challenges will be
discussed. These suggestions include recommendations
for assignment construction, including the use of group
work, collaborative assignments, e-portfolios, as well as
for planning course design, including consistent deadlines
and course structure. These suggestions are aimed at
mitigating student and instructor frustration with high
enrolment online classes.

956 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

BITCOINS ARE HERE TO STAY:

ARE WE READY?
Dr Chamath Keppitiyagama
University of Colombo School of Computing, Sri Lanka
# chamathk@gmail.com

Bitcoin is a cryptocurrency that is not under control of
any single authority or an organization. All the bitcoin
transactions are recorded in a public, distributed ledger.
The bitcoins exist in this ledger as chains of transactions.
This ledger is represented as a chained collection of blocks-
of-transactions. A bitcoin miner adds a cryptographic
hash to a block and once this hash is accepted by a
majority of nodes in the network, it is extremely hard
(almost impossible) to change the recorded transactions.
Huge amount of computing power has to be spent by the
miner to calculate this cryptographic hash. However, once
this hash is calculated it is extremely simple to verify that
it is indeed a valid hash. This hash is a proof that the miner
has spent sufficient computing power to protect a block
and as a reward for that work the miner gets new bitcoins.
This is the only way new coins can be minted in the bitcoin
system. As time goes on mining becomes harder and
harder and the rate at which the bitcoins enters into the
system goes down. This is how the bitcoin system controls
the money supply without a central bank. The owners of
bitcoins are not really people or organizations, but public/
private key pairs. Therefore, bitcoins provide certain level
of anonymity to transactions. This anonymity can be
further enhanced by techniques such as bitcoin mixing.
There are more than 3 million unique users of bitcoins
and more than 100,000 organizations accepts bitcoins
for payments worldwide. Are Sri Lankan organizations,
law enforcement, security establishment, and regulators
ready to face the challenges and opportunities in this new
economic environment?

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 957
PROCEEDINGS ISBN number: 978-955-0301-41-6

ANALYTICS AND AI:

THE GOOD, THE BAD AND THE UGLY
Dr Srinath Perera
Vice President of Research, WSO2, Colombo, Sri Lanka
# srinath@wso2.com

Analytics let us question the data, which in effect questions lives and build biases into our systems that can marginalize
the world around us. This let us understand, monitor, and millions. In this talk, we will discuss core ideas behind
shape the world. AI let us discover connections, predict analytics and AI, their possible impact, both good and bad
the possible futures and automate tasks. These twin outcomes, and challenges.
technologies can change the world around us. On one
hand make us efficient, connected, and fulfilled. At the
same time, the change of status quo can replace jobs, affect

958 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6 PROCEEDINGS

Technical
Session
(ORAL & POSTER PRESENTATIONS)

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 959
PROCEEDINGS
SINHALA SPEECH TO TEXT LIBRARY
USING SPHINX
WTVL Gunarathne1#, TK Ramasinghe1,
DGJB Wimalarathne1, BMSH Balasuriya1 and B Hettige1
1
General Sir John Kotelawala Defence University
# vlgunarathne@gmail.com
Abstract – Speech-to-text has generated a tremendous
interest in the field of Natural Language Processing where
the ultimate goal is to build applications and systems that
has the capability to respond to the natural languages that
us humans use in a daily basis. Converting speech to text
using European languages has emerged in the world and
can be found in most modern electronic devices. But a
speech-to-text tool for the native language of Sri Lanka
which is “Sinhala” is rare to find. So, developing a STT
algorithm for Sinhala to implement an application would
help most of the local workforce for their day-today tasks
and for the development of the software industry with a
unique Sri Lankan touch. STT has become one of the best
used technologies in the modern world. It has caught the
attention of the public as it is an excellent user-friendly
feature that has been embedded to electronic devices
in the modern world. It too has caught the attention of
the organizations as they can develop much better and
attractive devices which can rise in the competing market.
Personal Computers, Hand-held smart devices and even
modern automobiles come pre-integrated with STT for the
convenient of the user and give them a better experience.
Keywords – Speech-to-Text, Sinhala, Sphinx4
I. INTRODUCTION
Speech-to-text has become an interesting phenomenon in
the modern world. Various types of end devices such as
Personal Computers, Hand held smart devices and even
modern automobiles come integrated with this feature for
the convenience of the end user. One of the basic stages of
960 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
these advance voice recognition systems is the recognition
of words spoken buy a human being or in other words,
natural language recognition. These algorithms can
identify words spoken by a user, cross reference the words
through a word dictionary for validation and generate
an output to the user. It is an alternative to typing on a
keyboard and this is something that exists today in
smartphones where one of the most known application is
“Siri”(“Apple - iOS - Siri,” 2015) for Apple products and
“Google Now”(“Google now,” 2015) for devices running
on Android OS. Speech recognition will be more and
more common in the future as the amount of data grows
and devices contain more features.
Speech-to-Text software programs work by analyzing
sounds and converting them to text. This software has
been developed in a way to provide a faster method
of writing on a computer and can help people with a
variety of disabilities. It is useful for people with physical
disabilities who often find typing difficult, painful or
impossible. Speech-to-Text software can also help those
with spelling difficulties, including users with dyslexia,
because recognized words are almost always correctly
spelled. There some major researches in the field. The first
one is a CMU SPHINIX-4 SPEECH RECOGNITION
SYSTEM done by eight personals in different four
institutions. They are Carnegie Mellon University,
USA, University of California, Santa Cruz, USA, Sun
Microsystems Laboratories, USA and Mitsubishi Electric
Research Labs, USA. They have done a marvelous research
as joint development. There are some significant features
of the Sphinix-4 decoder. It is highly modular and flexible,
 ISBN number: 978-955-0301-41-6
supporting all types of HMM-based acoustic models, all
standard types of language models, and multiple search
strategies. Algorithmic innovations in the system enable
the concurrent use of multiple information streams. The
Sphinx-4 system is an open source project. The code has
been publicly available at SourceForge™ since its inception.
The design, results, and team meeting notes are also
publicly available.
vietnam national university, the international university
school of computer science and engineering have
introduced a good method of identifying the words using
the phonetics. The words to be identified should be broken
down to its phonetics. As an example, for the following
two Sinhala words can be encoded as,
e.g.: uf.a - M AH G EH
ku - N AH M AH
So, by using this concept we have made a gram file for
Sinhala language. We have put some frequently used words
in the Sinhala language, generated their phonetic scheme
as above and created a grammar file. There is a need of
a good accurate application for Speech Recognition
regarding Sinhala language. So, we thought of developing
a library for Sinhala language that can be edited for many
purposes. We thought of making it as an open source
library so that various people can make contributions to
developing it and thereby make use of it to address Sinhala
speech-to-text requirements.
III. PROBLEM AND OBJECTIVE
The problem at hand is that there is a need for a good and
accurate Speech-to-text application for Sinhala language.
And there is a need for a text library for Sinhala language
so that it can be used inside other applications to convert
spoken Sinhala words to Sinhala text. Our main objective
in this research was to study and develop a language model
and an acoustic model for the Sinhala language which
results in a language model file and a dictionary file which
includes the phonetics of the Sinhala word pronunciation
which can be used as an externally pluggable library to any
application.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
IV. SOLUTION
The solution to the above-mentioned problem is to
develop a library package which can convert spoken
Sinhala words to Sinhala text. This can be accomplished
by making modifications to the opensource library
Sphinx4(“Sphinx-4 - A speech recognizer written entirely
in the Java(TM) programming language,” 2015) and there
by compiling a bunch of library files that can be directly
imported an application development environment.
Speech is a complex phenomenon of which people rarely
understand how it is generated and perceived. The general
conception is that speech consists of words and each word
consists of phones which happen to be the building blocks
(phonetic schemes) of the word pronunciation.
Most modern speech recognition systems are mainly
based on probability theories. That means that there are
no certain boundaries between units, or between words.
Speech to text translation and other applications of
speech are never completely correct. Speech consists of a
continuous wave form comprising stable states mixed with
dynamically changing states. Within these sequences, we
can define similar components of sounds called phones.
A single word is said to be built from phones. Context,
speaker, style of speech, dialect can be factors which would
change the acoustic properties of a waveform. Transitions
between words are more informative than stable regions,
developers often talk about diphones - parts of phones
between two consecutive phones. Sometimes sub phonetic
units (different sub-states of a phone) are also considered.
In a considered phone there can be identifiable three
states, namely preceding phone, middle part & subsequent
phone. Phones that are considered in context are called
triphones or quinphones.
The recognition process is undertaken by taking the
waveform, split it on utterances by silences then try to
recognize what’s being said in each utterance.
V. DESIGN
Speech-to-text modules and libraries have been
proliferating in the technological arena and they tend to
gain traction much more as it is a user-friendly mode of
interaction between the user and a computer. The method
that these efficient algorithms use is a statistical analysis
of the probability of a word that can appear in a recorded
voice based on the array of words spoken before that
specific word. There was no considerable success in the
10TH INTERNATIONAL RESEARCH CONFERENCE 961
PROCEEDINGS ISBN number: 978-955-0301-41-6

field of speech recognition until Lenny Baum of Princeton

University invented the Hidden Markov Model (HMM)
which provided a statistical based model to generate text
from speech. From that point, onwards many corporations
have used this model to develop their own speech
recognition systems.

Figure 2: Phoneme breakdown for Sinhala

Table 1: Breakdown of Sinhala words to

phonetic scheme
Figure 1 : Breakdown of words to phonemes
Word Phoneme
A group of scientists from the Carnegie Mellon University, M AH G EH
Sun Microsystems Laboratories, Mitsubishi Electric
uf.a
Research Labs, Hewlett Packages and contributors from ku N AH M AH
University of California and Massachusetts Institute of
Technology have built a well-known speech recognition iqks,a S UH N IH L
framework written in Java programming language called
the Sphinx4 Framework.(“CMU Sphinx,” 2015, p. 4)
The above illustration depicts how a word can be sub-
Sphinx4 is a speech recognition system based on HMM
divided into its phones. The Sinhala word “ku” is basically
which is a precise mathematical framework. The ability to
comprised of 4 phones according to which a human being
use pluggable modules in the Sphinx4 framework makes
pronounces it. The table shows the phone building blocks
it more flexible and easy to adopt. Hidden Markov Model
for some other simple Sinhala words. Notice the tags “<s>
(HMM) is a statistical model in which the system being
</s>” which are the opening and closing tags of a series if
modeled assumed to be a Markov process with unknown
Words in the dictionary (.dict) file.(“Vietnamese Language
parameters, and the challenge is to determine the hidden
Recognition with Sphinx4 | Khai Tran - Academia.edu,”
parameters from an observation parameters. In speech
2015)
recognition process, after the user’s voice is recorded, it
will be divided into many frames that we need to process
to generate the sentence in text form. Each frame is VI. PHONETIC SCHEME
represented as a state, group of some states is represented
as phoneme, and group of some phonemes is represented
ALGORITHM
as word that we need to recognize. In database known
For the Speech-to-text conversion using the Sphinx4
as linguist model, we store the reference value of state,
library, the phonetic schemes of the Sinhala words should
phoneme, and word to compare with the observed data
be recorded in a dictionary file which resides inside the
(voice).
compiled jar file of the Sphinx4 library. These phonetic
schemes are the basic auditory units that buildup a
single word. The Sphinx4 library already has a built-
in dictionary file that comprises of English words and
their corresponding phonetic schemes. But for Sinhala
words to be recognized, the Sinhala words along with the
corresponding phonetic scheme (as shown in Table 1).

962 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Through a thorough analysis and testing of the dictionary
file for the English words, we identified the phonemes in
the lower level that corresponds to various pronunciations.
There by, we could map the English phonemes to Sinhala
words depending on how the words are pronounced.
The below figure will show an example for the Sinhala let-
ter “l” and all its transformations and the relative phonet-
ic scheme for each of those combinations.

d AA
• isxy, S IH NG ld - k AA

• uf.a M AH G EH e AE le - k AE
• ku N AH M AH E AE lE - k AE
• iqks,a S UH N IH L
s IH
• rg R AH T AH ls - k IH

• Tnkak OW B AH N N AH S IY lS - k IY

• yß HH AH R IY q
UH lq - k UH
• ksjeros N IH W AH R AH D IY Q UW lQ - k UW
• hkak Y AH N N AH
• tkak EH N N AH D R lD - k R

• Tn OW B AH f EH lf - k EH
• msgqjy,a P IH T UH W AH HH AH L
l fa EY lf a - k EY

The phonetic scheme algorithm that we have developed, ff AY lff - k AY

will decompose a given Sinhala word to its phonetic
scheme as follows. fd OW lff - k OW

fda OW lff - k OW

fda AO lff - k AO

Figure 3: Phonetic scheme algorithm architecture fda AA lff - k AA

f! AW lff - k AW

! UH l! - k UH

! UW l! - k UW

x AN x

Figure 5: Formations of phonemes

Figure 4: Algorithm implementation

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 963
PROCEEDINGS
VII. IMPLEMENTATION
i. Sinhala Speech-to-Text sample application
Figure 6: Sinhala speech-to-text application
This application takes the input from the microphone and
using the phonetic scheme above, recognizes the word
and displays it to the user. This above shown is an example
of the recognized word “msgqjy,a” spoken through the
microphone.
ii. The Web Browser
Figure 7: Sinhala voice search browser
This is a browser powered by the Google Search API[10]
and completely works with Sinhala words recognized
using voice, which are passed to as a search phrase to the
google search engine which in turn, returns the search
results as a Json object. The google Gson helps to decode
the Json object to separate the titles and the URLs of the
returned results. Those results can be seen in the large
textarea in the above figure.
VIII. EVALUATION
The recognition performance evaluation of the system
must be measured on a corpus of data. A separate test
corpus, with new Sinhala words, was created from the
main corpus. The test corpus was made of 200 recorded
and labelled data. In order to test for speaker independency
of the system, some of the subjects who participated in
964 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
creation of the testing corpus had not participated in
creation of the training corpus.
We have analyzed 200+ words and the data statistics are
as follows:
The average word identification rate for the tested data
set is 65.84% with a variance of 1000.961 and a standard
deviation of 31.63797.
IX. CONCLUSION
With the proliferation of Speech-to-text technology,
it’s clear that it has become an essential technology in
almost every electronic device. So here our attempt was to
develop a system and a library to handle voice recognition
in Sinhala language. We used the CMU Sphinx4 library
to achieve this task. The Sphinx4 library consists of a
grammar file which contains English words and their
corresponding phonetic schemes. Those phonetic schemes
are mapped to various word pronunciation blocks. This
research used those blocks and mapped them to the
pronunciation patterns of the Sinhala words available in
the selected data set. There by we could develop a Sinhala
speech-to-text library which can detect a spoken Sinhala
word, cross refer it with the existing data base and if
the word is available then will give a text output of the
spoken word. The final output of the research is a java
library that can be downloaded and used to any other
application development process. The issue that we faced
was to improve the word detection percentage to a much
higher level. The reason for that would be the fact that
Sphinx4 library was originally designed to adhere to the
dialect of Europeans. Since the pronunciation of Asians
and specifically Sri Lankans is different, the decoder
finds it difficult to identify the dialect. Therefore the word
recognition rate is lower.
REFERENCES
Apple - iOS - Siri [WWW Document], 2015. URL https://
www.apple.com/ios/siri/ (accessed 7.16.15).
CMU Sphinx, 2015.
Developer’s Guide | Google Web Search API (Deprecated)
| Google Developers [WWW Document], 2015. URL
https://developers.google.com/web-search/docs/?hl=en
(accessed 11.26.15).
 ISBN number: 978-955-0301-41-6
Dragon NaturallySpeaking - world’s best-selling speech
recognition software | Nuance [WWW Document],
2015. URL http://www.nuance.com/dragon/index.htm
(accessed 7.8.15).
FreeTTS 1.2 - A speech synthesizer written entirely in the
Java(TM) programming language [WWW Document],
2015. URL http://freetts.sourceforge.net/docs/index.php
(accessed 7.8.15).
Google now [WWW Document], 2015. URL https://www.
google.com/landing/now/ (accessed 7.16.15).
google/gson · GitHub [WWW Document], 2015. URL
https://github.com/google/gson (accessed 11.26.15).
How does voice recognition software work? - Explain
that Stuff [WWW Document], 2015. URL http://www.
explainthatstuff.com/voicerecognition.html (accessed
8.29.15).
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Microsoft Speech API (SAPI) 5.4 [WWW Document],
2015. URL https://msdn.microsoft.com/en-us/library/
ee125663(v=vs.85).aspx (accessed 7.15.15).
Q6.2: How is speech recognition performed? [WWW
Document], 2015. URL http://www.speech.cs.cmu.edu/
comp.speech/Section6/Q6.2.html (accessed 8.29.15).
Sphinx-4 Application Programmer’s Guide [CMUSphinx
Wiki] [WWW Document], 2015. URL http://cmusphinx.
sourceforge.net/wiki/tutorialsphinx4 (accessed 10.21.15).
Vietnamese Language Recognition with Sphinx4 |
Khai Tran - Academia.edu [WWW Document], 2015.
URL http://www.academia.edu/1501651/Vietnamese_
Language_Recognition_with_Sphinx4 (accessed 8.29.15).
10TH INTERNATIONAL RESEARCH CONFERENCE 965
PROCEEDINGS ISBN number: 978-955-0301-41-6

APPENDIX A

Word Phoneme , L
w AH j W
wd AA I SH
we AE Y SH
wE AE i S
b IH y HH
B IY L
<
W UH F
*
W! UW
xa
t EH AA
xd
ta EY AE
xe
ft AY AE
xE
T OW
xs IH
ö    OW
xS IY
T! AW
xq UH
l K
xQ UW
. G
xD R
p CH
fx EH
P CH
fax EY
c JH
ffx AY
g T
fxd OW
v D
fxda OW
K N
fx! AW
k N
f! UH
; TH
f! UW
o DH
xx AN
m P

n B

N BH

u M

h Y

r R

966 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

KEYXTRACT TWITTER MODEL - AN

ESSENTIAL KEYWORDS EXTRACTION
MODEL FOR TWITTER DESIGNED
USING NLP TOOLS
T Weerasooriya1#, N Perera2, SR Liyanage3
1
Department of Statistics and Computer Science, University of Kelaniya, Sri Lanka
2
Department of English, University of Kelaniya, Sri Lanka
3
Department of Software Engineering, University of Kelaniya, Sri Lanka
# cyriltcw@gmail.com

Abstract - Since a tweet is limited to 140 characters, it is
ambiguous and difficult for traditional Natural Language
Processing (NLP) tools to analyse. This research presents
KeyXtract which enhances the machine learning based
Stanford CoreNLP Part-of-Speech (POS) tagger with
the Twitter model to extract essential keywords from a
tweet. The system was developed using rule-based parsers
and two corpora. The data for the research was obtained
from a Twitter profile of a telecommunication company.
The system development consisted of two stages. At the
initial stage, a domain specific corpus was compiled
after analysing the tweets. The POS tagger extracted
the Noun Phrases and Verb Phrases while the parsers
removed noise and extracted any other keywords missed
by the POS tagger. The system was evaluated using the
Turing Test. After it was tested and compared against
Stanford CoreNLP, the second stage of the system was
developed addressing the shortcomings of the first stage.
It was enhanced using Named Entity Recognition and
Lemmatization. The second stage was also tested using
the Turing test and its pass rate increased from 50.00%
to 83.33%. The performance of the final system output
was measured using the F1 score. Stanford CoreNLP
with the Twitter model had an average F1 of 0.69 while
the improved system had a F1 of 0.77. The accuracy of the
system could be improved by using a complete domain
specific corpus. Since the system used linguistic features
of a sentence, it could be applied to other NLP tools.
Keywords - Natural Language Processing, Stanford
CoreNLP, Tweet Analysis, Named Entity Recognition,
Lemmatization, Keyword Extraction, Turing Test
I. INTRODUCTION
Natural Language Processing (NLP) has seen
unprecedented development over the past two decades
(Zitouni, 2014). Keyword extraction of NLP is used during
Question and Answering (Q&A) processes.
In understanding a question, humans extract keywords
that are vital in synthesizing the answer. These specific
words can also be used to back-formulate the question. In
NLP, POS tags could be used to extract key ideas from a
sentence.
One of the most fertile grounds to put NLP to test is
Twitter. A tweet might be ambiguous and is not always
grammatically correct. Hence, conventional POS tagging
methods cannot be used to extract keywords from a tweet.
Corporate giants often answer customer support requests
through Twitter, which has 320 million active users per
month (Twitter Usage / Company Facts, 2016). In Sri
Lanka, Dialog Axiata is a prominent telecommunication
company that provides this service. Automating this
process is challenging for a machine, as interpreting a
tweet could be problematic.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 967
PROCEEDINGS
This research presents KeyXtract which is a new utilization
of the Stanford CoreNLP (Manning et al., 2014) tool,
a widely used machine learning based NLP tool. The
research was conducted in two stages. The Twitter Model
for KeyXtract presented in this paper is the extension
of Stage 1 developed at the first stage of the research. In
the first stage (Weerasooriya, Perera and S R Liyanage,
2016), Stanford CoreNLP was enhanced using parsers (to
extract essential keywords using the linguistic features
of a sentence) and a domain specific corpus (consisting
of 206 words). The second stage presented in this paper
consists of improvements made based on the evaluation
results of stage 1. The Turing test was used to evaluate the
success of this method in imitating the human logic, and
its performance was measured using the F1 score.
II. RELATED WORK
A. Extracting keywords
Mitkov and Ha, state that to extract a “ ‘keyword phrase’,
a list of semantically close terms including a noun phrase,
verb phrase, adjective phrase and adverb phrase” (Mitkov
and Ha, 1999) should be considered. In the current study,
Noun Phrase (NP) and Verb Phrase (VP) are used in
keyword extraction.
B. Current tools in NLP and POS Tagging
Currently, Stanford CoreNLP (version 3.6.0) (Manning et
al., 2014),Open NLP (version 1.6.0) (Welcome to Apache
OpenNLP, 2013) and NLP4J (version 1.1.3) (emorynlp/
nlp4j: NLP tools developed by Emory University, 2016)
are the widely used machine learning based Open Source
NLP tools for Java. These are the NLP tools with the highest
level of accuracy. The NLP tool named ANNIE POS
tagger (included with GATE, version 8.2) (Cunningham
et al., 2001) uses a rule-based approach in contrast to
machine learning methods. The present research employs
a machine learning based approach of NLP tools.
POS tagging is done using Tregex (Levy and Andrew,
2006) method and the Penn Treebank notation (Marcus
et al., 1994) is used to POS tag each word. In both cases,
the tagger uses the unidirectional model, where the tag
of the current word is predicted based on the tags of its
neighbours. A dependency network is used to perform
this task, and a word is considered as a node in the network
which is directly influenced by its neighbours (Toutanova,
Klein and Manning, 2003). POS tagging is made use of in
the current study, to identify the keywords of a tweet.
968 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
C. Lemmatization
Lemmatization uses “the vocabulary and morphological
analysis of words”, normally aiming “to remove inflectional
endings only and to return the base or dictionary form of a
word, which is known as the lemma” (Manning, Ragahvan
and Schutze, 2008, p. 32). For example, this technique is
used to obtain the common root “eat” from the following
list.
E.g.; The lemma of ‘eats’, ‘eating’ and ‘eat’ is ‘eat’
Another similar technique of obtaining the root is through
stemming. (Manning, Ragahvan and Schutze, 2008,
p. 32). In this research, the Lemma is used to expand
and isolate the subject and the verb in a subject-verb
contraction. Stanford CoreNLP Suite (Stanford Named
Entity Recognizer (NER), no date) comes with the Lemma
bundled which is used for this study.
D. Named Entity Recognition (NER) and NLP
NER is used to extract relevant information from a text
and sort them into classes. The NER (Finkel, Grenager
and Manning, 2005) included in the Stanford CoreNLP
suite has the ability to label words into a 7-class model.
The 7 classes consist of location, person, organization,
money, percent, date and time (Stanford Named Entity
Recognizer (NER), no date). The NER utilizes the POS tag
and lemma of a word to assign a class into it. The class of a
word is also used in this study to identify keywords
E. Use of NLP in Twitter
As tweets are limited to 140 characters, they tend to
“exhibit much more language variation. (Bontcheva et al.,
2013). This is one reason why previous researchers state
that tweets cannot be analysed using basic POS tagging
(Bontcheva et al., 2013). Several attempts such TwitIE
(Bontcheva et al., 2013) and TweetNLP (Owoputi et al.,
2013) have been made to develop models to analyse POS
tags of Tweets, but with limited success. In the Twitter-POS
tagger model released for Stanford CoreNLP (Derczynski,
Ritter, et al., 2013), some of the aforementioned
functionalities have been incorporated. However, there is
a need for developing accurately modelled domain specific
corpora, for the analysis of POS tagging in Tweets.
In the present study, “rule-based grammars for the
syntactic-semantic analysis of word forms and sentences”
(Hausser, 2014) is applied to extract the relevant keywords
from the tweets.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

F. Turing Test and NLP A. Data Collection and Development of the Corpora

The Dialog Axiata (Dialog Axiata (@dialoglk) | Twitter,

The Turing Test (Turing, 1950) introduced by Alan
Turing in 1950 is conducted to answer the question,
“Can Machines Think?” (Copeland, 2004, p. 479). Thus,
according to Turing, “any machine that plays the imitation
game successfully can appropriately be described as a
brain” (Copeland, 2004, p. 479). The participants of the
test are a human respondent, a human evaluator and
the machine. The test is conducted by asking a question
from the human respondent and the machine, then
the human evaluator is asked to identify the machine
generated response out of the responses from the human
and the machine (Turing, 1950; Witten, Bell and Fellows,
1998). If the human is unable to identify more than half
of the machine generated responses, the machine passes
the Turing Test (Witten, Bell and Fellows, 1998). Since
the objective of this research is to enable the machine to
imitate a human, the Turing Test was used as the method
of evaluation.
G. The Present Research
This research was developed in two stages. The research is
the extension of Stage 1 of the system.
1) Stage 1: This stage extracted the keywords by
considering the noun phrase (NP) and verb phrase
(VP). The keywords were then sent through a parser
to remove any linguistic and domain specific noise,
followed by another parser to include any domain
specific words that were not extracted from the tweet.
The method was evaluated using the Turing Test
which consisted of a sample of 6 pairs (Weerasooriya,
Perera and S.R. Liyanage, 2016).
no date) Twitter profile was used to build the system. Two
corpora were built by analysing the tweets. Terms that
refer to Dialog Axiata and its services are identified as
Domain Specific Keywords (DSK). The corpus of DSK was
manually collected by analysing tweets from the month of
February, 2016 to March, 2016. The corpus contained 206
words.
The domain specific ‘words to reject’ corpus consisted of
words that do not contribute to the meaning (E.g. hello, hi,
dear), interjections that are often wrongly tagged as verbs
(E.g. please, thanks), certain nouns, verbs, and auxiliary
verbs. This corpus consisted of 70 words.
B. NLP and POS Tagger
A machine learning-based POS tagger was selected for this
research, as it has the ability to “exploit labelled training
data to adapt to new genres or even languages, through
supervised learning” (Derczynski, Ritter, et al., 2013). The
highest token accuracy of 97.64% is recorded by NLP4J
(Nanavati and Ghodasara, 2015; POS Tagging (State of the
art), 2016). However, this accuracy is at stake in Twitter
analysis. As a result, the token accuracy of the POS tagger
declines from 97-98% to 70-75% (Derczynski, Maynard,
et al., 2013).
A POS Tagger model specifically trained for tweets
displayed a token accuracy to 90.5% (Derczynski, Ritter, et
al., 2013). Out of the above mentioned list of tools,

2) Stage 2: The second stage addressed the shortcomings

of stage one. The improvements needed were identified
by comparing the machine generated responses and
the human generated responses of stage 1.

III.METHODOLOGY
The process of developing the stages 1 and 2 is mentioned Fig. 1 Flow Chart of Stage 1
below.
this model was available only for Stanford CoreNLP
(Derczynski, Ritter, et al., 2013). Hence, Stanford CoreNLP
was used for the present research.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 969
PROCEEDINGS ISBN number: 978-955-0301-41-6

The flow chart of the methodology for Stage 1 which

was developed in 2016 (Weerasooriya, Perera and S.R.
Liyanage, 2016) is shown in Fig 1.
C. Stage 2: System Design
The result of Stage 1 for an example tweet is given below.
Tweet - @dialoglk I made my payment just after my line
got barred in themorning! And still the line hasn’t got
connected, Whats with the delay?
Keywords – made (VBD), payment(NN), line(NN),
got(VBD), barred(VBD), morning(NN), line(NN),
got(VBD), connected(VBN), delay(NN)

(Abbreviations of the Penn Treebank Notation (Marcus et

al., 1994):
CC – Coordination Conjunction, CD – Cardinal Number, Fig. 2 The Flow Chart of the Improved System
DT – Determiner, IN – Preposition or subordination, JJ –
Adjective, NN – Noun, Singular or Mass, VB – Verb, base D. Parser 4 - NER and Time Indicators
form, VBD – Verb, past tense, VBN – Verb, past participle,
VBZ – Verb, 3rd person singular present, PRP - Personal The time indicators are, in most cases, not essential for the
Pronoun, PRP$ - Possessive Pronoun, RB – Adverb, USR meaning and were not present in the human generated
– Username, WP – Wh-pronoun) keyword sets. Thus, using NER, the time indicators such
The lapses identified from Stage 1 are as follows: as “morning” were removed in the system revision. The
result of the Parser 4 is shown in Fig. 3.
i. Unnecessary time indicators included– The word
“morning”

ii. Contractions not expanded– Contractions such as “ve”

and “n’t”

iii. Negation markers absent– The word “hasn’t”

Fig. 3 The result from Parser 4
iv. Duplicate Keywords not removed– Repetition of the
word “got” E. Parser 5 – Expansion and Analysis of Contractions

The above issues were addressed using the following
methods:
i. NER - To identify and remove time indicators.
ii. Lemma – To expand and analyse the contractions.
iii. Adverbs – To include negation markers
As lemma gives the common root of a word, it was used
to expand and analyse contractions. In the example (Fig.
3), the contraction “hasn’t” is expanded to “has” (VBZ)
and “not” (RB) through lemmatization. However, “has” is
rejected as it is an auxiliary verb (see Parser 3). The result
from Parser 5 is shown in Fig. 4.

iv. LinkedHashSet – To remove duplicates

Stage 2 was also evaluated using the Turing Test and the
results were recorded. Flow chart of the improved system
is shown in Fig. 2.
Fig. 4 The result from Parser 5

970 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
F. Adjustment of Parser 1 – Negation Markers (Adverbs)
The negation markers such as ‘not’ from Parser 5 are
identified as adverbs. When the lemma of the contractions
from Parser 5 separated the ‘not’ from the rest of the verb,
the word ‘not’ is included back into the keywords list as it
is important for the meaning.
G. Removing Duplicates
he keyword list shown in Fig. 4 consists of duplicates such
as “got”. This was removed from the final keywords list by
using a LinkedHashList, which does not allow duplicates,
while retaining the sequential order in the list.
Keywords – made (VBD), payment(NN), line(NN),
got(VBD), barred(VBD), line(NN), not(RB),
connected(VBN), delay(NN)
Fig. 5 The final resulting keyword list
IV. RESULTS AND DISCUSSION
The system (also referred as the ‘machine’ in this section)
was evaluated with and without improvements. The
system without modifications is the Stanford CoreNLP
(Manning et al., 2014) with the Twitter model (Derczynski,
Ritter, et al., 2013). The system with modifications is the
system presented in this research and it is tested against
the Stanford CoreNLP with the Twitter model. The three
systems are referred to as follows in the rest of the paper.
System A : Stanford CoreNLP with the Twitter model
System B : Stage 1 of the machine
System C : Stage 2 of the machine, KeyXtract
The systems were evaluated using two methods as follows,
a) Turing Test – To evaluate if the machine could
successfully imitate human logic.
b) Performance Test – To evaluate the performance of
keyword extraction by the machine. This was measured
using the F1 Score (Derczynski, 2013).
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
The Turing Test involves asking a set of questions from a
human and the machine. Their answers are then evaluated
by a human supervisor. If the supervisor is unable to
identify the machine in at least half of the test cases, the
machine passes the Turing Test (Turing, 1950). The testing
was conducted in three phases,
a) To evaluate System A.
b) To evaluate System B.
c) To evaluate System C.
The performance was measured by comparing the
keywords sets generated by the machine with two
keywords sets produced by an English language expert and
an author of the research. The human generated keyword
sets were compared with the System A and System C using
the F1 Scores. This was used to measure the performance
of the machine.
A. Turing Test: Evaluation Methodology: Design
The evaluations based on the Turing Test were conducted
for each system as mentioned above. The System A and
System B were tested with the same set of supervisors
with a time gap of 3 months between the tests. This time
gap was to ensure the responses would not be fresh in the
minds of the human participants. Since the System C was
improved considering the previous responses, it was tested
with a new set of supervisors.
B. Turing Test: Evaluation Methodology: Participants
All three systems were tested with six test cases (each
consisting of the machine, the human keyword generator
and a human supervisor). The human participants in the
six test cases were chosen to represent six different fields.
The criteria of the test cases are given in Table I.
Keywords sets were produced by the human and the
machine, and the human supervisor tried to identify the
system generated answers.
C. Turing Test: Evaluation Methodology: Keyword
Extraction
The testing dataset consisted of 14 tweets (2 tweets per day
collected for 7 days). They were collected from the first
week of April, 2016 (3rd April to 9th April). The dataset
contained a new set of tweets. The set of tweets were given
to the human keyword generators to extract keywords and
10TH INTERNATIONAL RESEARCH CONFERENCE 971
PROCEEDINGS ISBN number: 978-955-0301-41-6

then to the machine to do the same. This was repeated for Machine Generated Keywords - buy, touch, travel, pass
Systems A, B and C. Human Generated Keywords - buy, touch, travel, pass
Among System B, and C, 5 tweets out of 14 had keyword
sets where the answers of the human and the System were
Table I - Test case criteria
identical.
Test Case Test Case Justification Minimum Require-
Number Criterion ment F. Turing Test: Evaluation Discussion: Results
1 Academics Frequent users University lecturers
of Academic who are not from the A summary of the overall evaluation results is given in
English field of English
Tables III (for System A), Table IV (for System B) and Table
2 English Lan- Competent English Language V (for System C). The total instances where the machine
guage Experts in English
language and Lecturers was successful was calculated using the given formula.
literature

3 Undergrad- Use English Individuals currently

uates for academic reading for a Bache-
purposes lor’s degree

4 Graduates Use English in Individuals who have

a professional completed a Bache-
context lor’s Degree where,
5 Computer Have an expert Computer Science T - Total instances where the system was successful
Science knowledge graduates working in
Graduates in Computer x - Instances where the Machine and Human answers are
Science the industry identical
6 Randomly se- Being Familiar Twitter Users z - Instances where the Supervisor did not detect the
lected twitter with Twitter
users
answer generated by the Machine
n - Total number of tweets
D. Turing Test: Evaluation Methodology: Evaluation of
Evaluation Results of System C is shown in Table II.
Test Cases
The responses generated at the extraction phase by the Table II - summary of turing test applied for
human keyword generators, the Systems A, B and C were system c: stage 2 of the machine
used in this stage.
Test Case Machine and Supervisor Supervisor Total in-
In the first round, the supervisor was provided with the Criterion Human an- detected could not stances
swers were the answer detect the where the
original tweet and the two sets of keywords generated by identical (x) generated answer system was
the System A and the human. The supervisor was asked to by the Ma- generated successful
chine (y) by the ma- (T)
identify the set of keywords which was generated by the chine (z)
System A (the machine). Academics 0 11 3 21.43%

English 0 4 10 71.43%
The same process was repeated with the keywords sets Language
generated by the human and System B, and the human Experts

and the System C. System C was tested last with a group Undergradu-
ates
4 8 2 85.71%

of fresh supervisors who were completely new to the Graduates 4 7 3 50.00%

research.
Computer 4 3 7 78.57%
Science
E. Turing Test: Evaluation Discussion: Identical Graduates

Keyword Extraction Randomly

selected
5 5 4 64.29%

twitter users

During the keyword extraction phase, all three systems

produced several keyword sets which were identical to the The machine was unsuccessful only with academics,
responses of the human. this could be due to their familiarity with analytical and
An example for this occurrence is given below, technical writing.
“@dialoglk Where i can buy a touch travel pass?”

972 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

An overview of the Turing test results is shown in Table The precision (P) was computed by dividing the true
III. positives (i.e. the number words which were common to
the human and the machine data set) by the false positives
Table III - comparison of the turing test results (i.e. the total number words which were extracted by the
machine). The recall (R) was computed by dividing the
System Tested Test cases that
passed
Test cases that
failed
Success rate of
the System
true positives by the total number of words which were
extracted by the human.
System A: Stan- 3 3 50.00%
ford CoreNLP
with the Twitter
model
The F1 score was computed for System A and System C.
System B: Stage 5 1 83.33%
The results for dataset by the ELE is included in Table IV.
1 of the machine

System C: Stage
2 of the machine
5 1 83.33% Table IV - F1 scores for English
language expert’s dataset
According to the Table VI, it is evident that the modified System A System C
systems have more success in imitating the human
Tweet# Word# P R F1 P R F1
in extracting keywords than the system without any
modifications. 1 9 0.40 1.00 0.57 0.50 1.00 0.67

2 11 0.43 0.75 0.55 0.60 0.75 0.67

G. Performance Test: F1 Score Evaluation Discussion 3 25 0.38 0.60 0.46 0.50 0.80 0.62

4 15 0.71 1.00 0.83 0.80 0.80 0.80

The performance of the machine was evaluated using the 5 24 0.55 0.86 0.67 0.44 0.57 0.50
F1 Score. Initially, an English Language expert (ELE) and
6 25 0.23 0.75 0.35 0.36 1.00 0.53
an author of the paper generated the controller dataset
7 25 0.25 0.40 0.31 0.45 1.00 0.63
of keywords from the 14 tweets used for the Turing Test,
from Section C. Two human generated keyword sets were 8 23 0.25 0.67 0.36 0.60 1.00 0.75

used factoring the subjectivity of the keyword extraction 9 9 1.00 1.00 1.00 1.00 1.00 1.00

process. The average of the F1 Scores from the two sets of 10 7 1.00 1.00 1.00 1.00 1.00 1.00
keywords was used for the evaluation. 11 7 1.00 1.00 1.00 1.00 1.00 1.00

12 20 0.30 1.00 0.46 0.43 1.00 0.60

The F1 Score(F) was calculated by analysing the keywords 13 13 0.83 1.00 0.91 0.83 1.00 0.91
generated for each tweet according to the formula
14 10 1.00 0.71 0.83 1.00 0.86 0.92
(Derczynski, 2013) given below,
Average 0.59 0.84 0.66 0.68 0.91 0.76

The results show that the System C has improved from a

F1 score of 0.66 to 0.76.

The F1 score computed for the dataset extracted by an

where, author of the paper is included in Table V below,
F – F1 Score
P - Precision
R – Recall

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 973
PROCEEDINGS ISBN number: 978-955-0301-41-6

Table V - F1 scores for the dataset by an author Core NLP and the Twitter-POS tagger model was used for
POS tagging to extract keywords from the tweet. The rule
System A System C based approach was used to remove unnecessary words
from the initial word group selection with the help of
Tweet# Word# P R F1 P R F1
corpora. The research was developed in two stages. Stage
1 9 0.47 1.00 0.57 0.50 1.00 0.67 2 included modifications to Stage 1 such as using NER to
2 11 0.57 1.00 0.73 0.80 1.00 0.89 remove time indicators and measures to include negation
3 25 0.38 0.60 0.46 0.50 0.80 0.62 markers.
4 15 0.71 1.00 0.83 0.80 0.80 0.80
The research was evaluated using two methods. The
5 24 0.55 0.86 0.67 0.44 0.67 0.53
ability to imitate the human logic in extracting keywords
6 25 0.23 0.75 0.35 0.36 1.00 0.53 was measured with the help of the Turing Test while the
7 25 0.50 0.57 0.53 0.45 0.71 0.56 performance was measured using the F1 Score.
8 23 0.38 0.75 0.50 0.80 1.00 0.89
The final modified system passed the Turing Test with an
9 9 1.00 0.80 0.89 1.00 0.80 0.89
overall result of 83.33%There were more instances where
10 7 1.00 1.00 1.00 1.00 1.00 1.00 the modified system produced the same set of results
11 7 1.00 1.00 1.00 1.00 1.00 1.00 as humans. Since the system from stage 3 consists of
12 20 0.30 1.00 0.46 0.43 1.00 0.60 the improvements made to the system from stage 2, the
13 13 0.83 1.00 0.91 1.00 1.00 1.00 evaluation results look quite promising. The system could
14 10 1.00 0.83 0.91 1.00 1.00 1.00
be tested with a larger population for nuance results.
Average 0.63 0.87 0.70 0.72 0.91 0.78
The performance measures of the system showed that the
F1 scores increased from 0.69 of system A (system without
The F1 score of the modified System C has increased from any modifications) to 0.77 of system C (final system with
0.70 to 0.78. The average of the two F1 Scores obtained modification). It was also evident that the system’s level of
from the two sets are summarized in the Table VI, precision was high in analysing short tweets.
Future work in the research could include the use of a
TABLE VI - SUMMARY OF F1 SCORES complete domain specific corpus and the ability to analyse
emoji, which would improve the accuracy of the keywords
System Tested ELE Dataset Author Dataset Average
extracted by keyword matching. As this approach uses
System A: 0.66 0.70 0.69
Stanford
linguistic features to extract keywords, the same approach
CoreNLP with could be applied to other NLP tools
the Twitter
model ACKNOWLEDGMENT
System C: 0.76 0.78 0.77
Stage 2 of the
machine The authors would like to thank the participants of the
Turing Test for their time and effort to evaluate KeyXtract.
The highest F1 scores are recorded from the Tweets
which of shorter length, proving that the accuracy of the REFERENCES
machine is high in short tweets.
Bontcheva, K., Derczynski, L., Funk, A., Greenwood, M.
A., Maynard, D. and Aswani, N. (2013) ‘TwitIE: An Open-
V. CONCLUSION Source Information Extraction Pipeline for Microblog
Text’.
The traditional machine learning based NLP tools have
failed to accurately classify the tokens of the tweet with Copeland, B. J. (2004) The Essential Turing: Seminal
POS tags. Writings in Computing, Logic, Philosophy, Artificial
Intelligence, and Artificial Life: Plus The Secrets of
The paper presents the Twitter Model of KeyXtract, a Enigma. Edited by B. J. Copeland. Oxford: Clarendon
system developed with a mix of machine learning and a Press.
rule-based approaches. A combination of the Stanford
974 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Cunningham, H., Maynard, D., Bontcheva, K. and Tablan,
V. (2001) ‘Gate’, Proceedings of the 40th Annual Meeting
on Association for Computational Linguistics - ACL
’02,(July), p. 168. doi: 10.3115/1073083.1073112.
Derczynski, L. (2013) ‘Complementarity, F-score, and
NLP Evaluation’.
Derczynski, L., Maynard, D., Aswani, N. and Bontcheva,
K. (2013) ‘Microblog-genre noise and impact on semantic
annotation accuracy’, Ht 2013, (May), pp. 21–30. doi:
10.1145/2481492.2481495.
Derczynski, L., Ritter, A., Clark, S. and Bontcheva, K.
(2013) ‘Twitter part-of-speech tagging for all: Overcoming
sparse and noisy data’, Proceedings of the Recent Advances
in Natural Language Processing, (September), pp. 198–
206. Available at: http://www.aclweb.org/website/old_
anthology/R/R13/R13-1026.pdf.
Dialog Axiata (@dialoglk) | Twitter (no date). Available at:
https://twitter.com/dialoglk.
emorynlp/nlp4j: NLP tools developed by Emory University
(2016). Available at: https://github.com/emorynlp/nlp4j.
Finkel, J. R., Grenager, T. and Manning, C. (2005)
‘Incorporating non-local information into information
extraction systems by gibbs sampling’, in Acl, (1995), pp.
363–370. doi: 10.3115/1219840.1219885.
Hausser, R. (2014) Foundations of Computational
Linguistics. Third Edit. London: Springer Heidelberg. doi:
10.1007/978-3-662-04337-0.
Levy, R. and Andrew, G. (2006) ‘Tregex and Tsurgeon:
tools for querying and manipulating tree data structures’,
5th International Conference on Language Resources and
Evaluation (LREC 2006), pp. 2231–2234.
Manning, C. D., Bauer, J., Finkel, J., Bethard, S. J., Surdeanu,
M. and McClosky, D. (2014) ‘The Stanford CoreNLP
Natural Language Processing Toolkit’, Proceedings of 52nd
Annual Meeting of the Association for Computational
Linguistics: System Demonstrations, pp. 55–60. Available
at: http://aclweb.org/anthology/P14-5010.
Manning, C. D., Ragahvan, P. and Schutze, H. (2008)
An Introduction to Information Retrieval. 1st edn. New
York, NY: Cambridge University Press. doi: 10.1109/
LPT.2009.2020494.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Marcus, M., Kim, G., Marcinkiewicz, M. A., MacIntyre,
R., Bies, A., Ferguson, M., Katz, K. and Schasberger,
B. (1994) ‘The Penn Treebank: Annotation Predicate
Argument Structure’, Proceedings of the workshop on
Human Language Technology - HLT ’94, pp. 114–119. doi:
10.3115/1075812.1075835.
Mitkov, R. and Ha, L. A. (1999) ‘Computer-Aided
Generation of Multiple-Choice Tests’, Proceedings of
the HLT-NAACL 03 workshop on Building educational
applications using natural language processing, pp. 17–22.
Nanavati, J. and Ghodasara, Y. (2015) ‘A Comparative
Study of Stanford NLP and Apache Open NLP in the view
of POS Tagging’, International Journal of Soft Computing
and Engineering (IJSCE), 5(5), pp. 57–60. Available at:
http://www.ijsce.org/attachments/File/v5i5/E2744115515.
pdf.
Owoputi, O., O ’connor, B., Dyer, C., Gimpel, K.,
Schneider, N. and Smith, N. A. (2013) ‘Improved Part-of-
Speech Tagging for Online Conversational Text with Word
Clusters’, Proceedings of NAACL.
POS Tagging (State of the art) (2016). Available at: http://
aclweb.org/aclwiki/index.php?title=POS_Tagging_(State_
of_the_art) (Accessed: 22 August 2016).
Stanford Named Entity Recognizer (NER) (no date).
Available at: http://nlp.stanford.edu/software/CRF-NER.
html (Accessed: 20 November 2016).
Toutanova, K., Klein, D. and Manning, C. D. (2003)
‘Feature-rich part-of-speech tagging with a cyclic
dependency network’, In Proceedings of the 2003
Conference of the North American Chapter of the
Association for Computational Linguistics on Human
Language Technology - Volume 1 (NAACL ’03), pp. 252–
259. doi: 10.3115/1073445.1073478.
Turing, A. M. (1950) ‘Computing Machinery and
Intelligence’, Mind, 49, pp. 433–460. doi: 10.1093/mind/
LIX.236.433.
Twitter Usage / Company Facts (2016). Available at:
https://about.twitter.com/company (Accessed: 30 April
2017).
10TH INTERNATIONAL RESEARCH CONFERENCE 975
PROCEEDINGS
Weerasooriya, T., Perera, N. and Liyanage, S. R. (2016) ‘A
method to extract essential keywords from a tweet using
NLP tools’, in 2016 Sixteenth International Conference on
Advances in ICT for Emerging Regions (ICTer). IEEE, pp.
29–34. doi: 10.1109/ICTER.2016.7829895.
Weerasooriya, T., Perera, N. and Liyanage, S. R. (2016) ‘A
Method to Extract Essential Keywords from a Tweet using
NLP Tools’, in 2016 International Conference on Advances
in ICT for Emerging Regions (ICTer). Colombo, pp. 29–
34.
976 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Welcome to Apache OpenNLP (2013). Available at: http://
opennlp.apache.org/.
Witten, I. H., Bell, T. and Fellows, M. (1998) Computer
Science Unplugged . . .Off-Line Activities and Games for
All Ages.
Zitouni, I. (ed.) (2014) Natural Language Processing
of Semitic Languages. New York, NY: Springer Berlin
Heidelberg. doi: 10.1007/978-3-642-45358-8.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

CHATBOTS:
THE NEXT GENERATION IN
COMPUTER INTERFACING - A REVIEW
MS Walgama1, B Hettige1#
Faculty of Computing,
1

General Sir John Kotelawala Defence University, Kandawala Road,

Rathmalana,Sri Lanka
# budditha@kdu.ac.lk

Abstract – Computer interfacing is now moving I. INTRODUCTION

towards more machine driven communications than the
Command-line or graphical user Interfacing. Human-
Computer speech through communication is now gaining
momentum as a technique of computer interaction
which paved the way for enormous developments using
natural language processing. Such systems are precisely
designed to simulate how a machine could behave as a
conversational partner. A Chatbot is a computer program
that stimulates intelligent human conversation using a
natural language. This could be done through textual or
auditory mechanisms. These Chatbots facilitates easier
learning of the domain concepts and their interrelated
relationships which make them efficient in the use of
general applications. Chatbot architecture integrates
a language model and computational algorithms to
emulate informal chat communication between a human
user and a computer using natural language. Recently,
Chatbots techniques are widely used for various practical
purposes and made available to the public. The objective
of this paper is to review significant conversational agents
that have been developed in Sri Lanka as well as in the
other parts of the world for various domains over the
past years. The development techniques, approaches and
functionalities are pointed out through this paper.
Keywords - AIML , Chatbot , Natural Language Processing
User interfaces are referred to as the graphical, textual and
auditory information that a program parses to a user. After
decades of development, at present there exist several
types of interfaces. The shift of computation intensive
design to presentation intensive design is one of the largest
steps in software development. The era of command line
interfaces which occurred in late 1960s allowed users to
respond using a visual prompt by typing in commands.
The MS-DOS prompt application of Windows is one of
the widely used. The story became interesting and wide
spread with the invention of digital computers. As a result,
Graphical User interfaces(GUI) emerged. GUI allows
users to interact with devices through graphical icons
instead of text- driven commands. Direct manipulation of
graphical elements performs actions in a GUI. The next
generation of innovation was taken to another level which
allows users to interact using voice commands. The latest
evolution of user interfaces is conversational agents. These
use natural language to communicate with the user for eg
Siri, cortana.
In 1950, Alan Turing distributed his well-known article
“Computing Machinery and Intelligence” what is
presently called the Turing test (Turing A.M, 1950)as
a foundation of intelligence. This measure relies upon
the capacity of a computer system to imitate a human

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 977
PROCEEDINGS
in a constant composed discussion with a human judge.
(French, 1990) In 1966, ELIZA which was developed by
Joesph Weizenbaum is recorded as the first ever program
to pass the Turing test. ELIZA is considered as a clinical
agent which runs on a time-sharing framework which can
automatically handle several patients in an hour.
Chatbot architecture integrates language models and
computational algorithms (Setiaji and Wibowo, 2016)to
emulate informal chat communication between a human
user and a computer using natural language. Speech based
search engines and assistants such as Cortana by windows,
Siri by Apple and Google Chrome are gaining a surge in
the commentary context.
The development of chatbots involve extensive knowledge
acquisition which is stored into the system to function
with user queries. Use of Artificial Intelligence Markup
Language (AIML) is widely seen in the development of
chatbots. AIML supports dialogs in natural language
which is then matched to an input query to given text
pattern. The use of chatbots is visible in a miscellaneous set
of applications both online and offline. Chatbots could be
used as a tool to learn, to access information in a system,
as a question and answering tool for a specific domain
and in many different fields like medicine, education,
entertainment, automobile or any other.
ALICE, Mitzuki which won the Leobner Prize in 2013,
Cleverbot are some of key chatbots while Jabberwacky,
Watson and botster some other chatbots used for different
purposes. This paper reviews many exiting systems that
use this chatbots techniques in their development. The
presented systems are from different fields proposing
to address different problem domains. The features,
functionality and methodology of these systems are also
presented in this paper.
This paper gives an overview of the existing chatbot systems
including their approaches and functions. The rest of the
paper is organized as follows. Section 2 presents about
Chatbot fundamental design techniques and approaches.
Section 3 reviews existing systems on chatbots. Finally,
conclusion is given in section 5.
II. CHATBOT FUNDAMENTAL
DESIGN TECHNIQUES AND
APPROACHES.
Building a chatbots requires many different techniques to
be implemented together. Several key techniques that are
978 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
widely used in development are also looked in to within
this content.
2.1 Artificial intelligent Markup Language (AIML)
AIML is an XML complaint language that is widely used in
designing chatbots. This is mainly based on the technology
used for ALICE. The goal of AIML language is to direct
the process of conversational modeling in to a stimulus
response process. AIML characterizes the type of data
object whose responsibility is to model conversational
patterns. These could be named as the frequent tags
and the bases used in the design of AIML chatbots who
respond intelligently. Below shows the structure of
category, pattern and template object used in AIML.
<category>
<pattern> User Input</pattern>
<template>
Corresponding Response to input
</template>
</category>
2.2 Pattern matching
Pattern matching is the techniques of checking a given
sequence of tokens for the presence of some patterns.
Chatbots use this techniques as a common practice and it
is frequently seen in question and answering systems. One
of the key ways of developing a chatbot is by analyzing the
input and finding the best match. Chatbots systems practice
these patterns matching techniques mainly because
to make the users feel that they are chatting with a real
human. ELIZA use a matching keyword and application
of transformation rules. In this process, the conditions of
all topics are compared to the goal information, the query
and the keyword list. As a result, the topics are scored and
the highest score out of all is returned as the answer.
2.3 Parsing
Parsing is the process which is used in analyzing a string
of symbols either in computer languages or natural
languages. In computational linguistic parsing is used
to analyze a sentence or another set of strings in to its
constituents which may contain semantic and other
information. This technique includes analyzing the input
text and manipulating it by using several NLP functions.
One such example is trees in Python NLTK.
 ISBN number: 978-955-0301-41-6
2.4 Chat scripts
Chat scripts are often used when the AIML does not
provide any matching in its context is a sequence of expect
strings. Chat scripts mainly focus on the best syntax out
of all the provided syntaxes given to provide a sensible
answer to system. Variable concepts, logic and facts are
some of the functionalities that chat scripts address.
2.5 ontology
Ontologies are used in chatbots to compute the concepts
of synonyms, antonyms, hyponyms and other co concepts
that prevail in any natural language
III. EXISTING SYSTEMS
Through further studies for years the researchers could
develop various efficient systems with the use of chatbots
techniques which is a catalytic section in the field of
Natural language processing. This section describes the
systems that was developed in the recent past which are
applicable for various fields such as medicine, education,
entertainment etc.
ALICE (AbuShawar and Atwell, 2015)is one of the earlier
developed chatbots. Shawar and Atwell presents an
overview of ALICE chatbot, its AIML format and their
experiments they used to generate different prototypes
of ALICE based on a corpus approach. A description of
developed software which converts readable text (corpus)
(Shawar and Atwell, 2005)into AIML format is presented
alongside with describing the different corpora that is used
in this system. A Java program that converts a readable
text to the chatbot language model format is developed.
The entire program has been divided in to four phases to
handle the linguistic annotations and filters. The approach
used is a Dialog Diversity Corpus of English along with an
after-text reprocessing and filtering using the developed
java program. A monologue bilingual corpus was also
used in this process.
Fig1 -Overview of ALICE chatbot
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Shawar, Eric Atwell (Shawar and Atwell, n.d.)present a
program developed to convert a machine-readable text
(corpus) to a specific chatbot format, which is then used
to retrain a chatbot and generate a chat which is closer
to human language. Different corpora such as dialogue
corpora, monologue corpora such as holy Qur’an (Bayan
Abu Shawar and Eric Atwell, n.d.)and FAQ s have been
used here. The main goal of this process is the ability to
generate different chatbot prototype that spoke different
languages.
A java program that converts the corpus to the chatbot
language model format has been developed which aims
to create ALICE knowledge base automatically and based
on specific corpus or domain. To handle the linguistic
annotations and fillers, the program is composed of four
phases. The system could generate AIML training data for
ALICE chatbot to serve in different domains. (Shawar and
Atwell, 2003a)
Chatbots could be used in various fields such as education,
medicine, and entertainment, business etc. Several
applications that could be used in day today life has been
developed by the researchers in modern days. The first
Sinhala chatbot (Budditha Hettige and Karunananda,
2006)developed by Hettige and Karunananda is a useful
system that has been developed to communicate in Sinhala
language. This system has been developed as an application
of a Sinhala parser that comes under a major component
of the project in English to Sinhala machine translation
system. The system has been developed using JAVA and
SWI-PROLOG supporting both Linux and Windows. As
the major components, this system comprises of a Sinhala
Morphological analyzer and a Sinhala Language parser.
(B. Hettige and Karunananda, 2006)
The Sinhala Morphological analyzer connects with three
dictionaries namely, base dictionary, rule dictionary and
concept dictionary. Knowledge identification engine
reads all the information given from Sinhala language
parsing system. It uses simple pattern matching algorithm
to identify user input and find the appropriate solutions
from knowledge base. This has been developed as an
automatically updating system where the knowledge
base is updated whenever the users use this. Further it is
stated to extend the chatbot to operate on a more specific
domain.
10TH INTERNATIONAL RESEARCH CONFERENCE 979
PROCEEDINGS
Figure 2 - Overview of the Sinhala chatbot
Abdul-Kader and Woods(Abdul-Kader and Woods, 2015)
presents a survey on the techniques used in designing
chatbots. As per survey most research work has focused
on improving recognition rates of the human voice and
the technology is now approaching viability for speech
based human computer interaction. Different techniques
that could been seen in these applications are compared.
Features and functions of the NLTK is also described here.
A responder, classifier and graph master are three
main parts in a chatbot. Parsing, pattern matching,
AIML(Richard S. W A L L A C E, 2003)chat script, SQL
and relational database, Markova chain, language and
ontologies are some of the key techniques of chatbots that
are described. Review of recent chatbots which are used
in different sectors along with their functions are also
discussed here. The ranking of several chatbots along with
their functions are presented through this.
Shawar and Atwell presents a comparison of two
chatbot systems, ALICE and Elizabeth(Shawar and
Atwell, 2002),(Shawar and Atwell, 2003)illustrating the
dialogue knowledge representation and pattern matching
techniques of each. ALICE was found to be better suited
for training using dialogue corpora because of its simple
patterns templates and simple matching technique. A
general description about ALICE chatbot, the AIML data
objects, types of categories in ALICE, pattern matching
in Alice and its algorithms are described here. ALICE is
a software robot or program that you can chat with using
natural language. Elizabeth is an adaptation of the Eliza
program, in which the various selection, substitution,
and phrase storage mechanisms have been enhanced and
generalized to increase both flexibility and adaptability.
It is concluded that they decided to train ALICE rather than
Elizabeth to learn from human dialogue corpora for two
980 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
reasons. Firstly, the AIML format is closer to the markup
format used in annotated corpora. Secondly, the simplicity
in generating patterns/templates, and applying simple
pattern matching technique. The conclusion relating to
Corpus Linguistics is that the Dialogue Diversity Corpus
(DDC) illustrates huge difference in dialogues.
ELIZA (West et al., 1985) by Joseph Weizenbaum is
a system that is implemented within the MAC time
sharing. ELIZA can converse in natural language. ELIZA
analyzes the user provided statement and generate the
corresponding response. The input is read and analyzed
with the presence of a keyword and when a specific word is
matched, the other words will be deleted and the sentence
will be transformed according to the associated rules.
ELIZA scripts exist in Welsh and German other than
English. One major problem in this system is text
manipulation. ELIZA script contains a list of structures.
The actual keyword directory is created once the script is
read. The performance of ELIZA is significant when the
first input is given by the user.
Use of technology in the field of medicine (Vales and
Sukhanya, 2009) is stepping forward rapidly. One such
initiative is “Pharmabot”(Comendador et al., 2015), which
is defined as a pediatric generic medicine consultant
chatbot. Developed by Comendador , this Pharmabot
will converse in order to prescribe, suggest and provide
information on medicine for children. A left to right
parsing algorithm is used in the development. Details about
some chatbots such as Erica which is developed for dental
practices, ELIZA which stimulates a psychotherapist and
about PARRY are described in this paper.
Visual C# has been used in the development and is
designed to run as a standalone system. The user will be
asked several questions there by the properly answered
ones will be directed to the parsing algorithm. As a
significant fact of the system this provides a dictionary
database which contains technical and medical terms for
any novice user. The calculation which has been done
through a weighted mean method depicts that this system
function and generate the same results that is expected
from a manual system.
Another application of chatbots in the medical field is
MedChatBot(Kazi et al., 2012). This is based on AIML and
Unified Medical Language System (UMLS) based. UMLS
contains around two million medical concepts including
variety of medical domains (Webber, 2005) and about 135
semantic types.
 ISBN number: 978-955-0301-41-6
Figure 3 - Relationships between Semantic Types
of Concepts in MedChatBot
This use SQL queries and a knowledge base to return the
result in natural dialog. Chatterbean, the open source java
based AIML interpreter has been used in this system for its
further development. There are three major components
in the architecture of the MedChatBot namely the front
end, the AIML parser and the database. The user inputs
the question.
One of the limitations of the system is that the at times
the system will be unable to produce accurate responses
to the queries due to lack of casual relationship between
concepts of UIML. 150 queries were gathered from five
random students accounting to 30 from each. Then these
collected queries have been tested by expert and the
system has generated results at a higher rate.
‘‘Octopus” (B Hettige and AS Karunananda, 2015)which
is a multi-agent chatbot is another application of chatbot
which use Sinhala language. This is structured with 8 sub
multi agent systems namely core system, GUI system,
Natural Language Processing system, communication
system, learning system, action system, searching system
and data access system. The development is done through
the framework MasMT which is a multi-agent framework.
Octopus functions through text input which also provides
action searching facilities such as to execute commands,
open or close applications. This has been developed
as a java application running on Windows and Linux.
The current version of Octopus is said to have limited
capability in semantic processing. Improving such aspects
and passing the turning test is stated as future work of the
system.
Bradeško and Mladenić (Bradeško and Mladenić, 2012)
compare the technologies of the chatbots that have won the
Loebner Prize. The technologies and approaches of those
chatbots are presented in this paper. The Loebner Prize
Competition is an annual competition held for chatbots
to check its capabilities, through Turing test method.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Chatbots like ALICE, CLEVERBOT, Suzette, Rosette are
some of the award winning chatbots (Konferenca Jezikovne
tehnologije et al., 2012) The technologies, language tricks,
technical approaches and their respective algorithms are
presented through this paper. Pattern matching, Markova
chain models, parsing, ontologies and chat scripts of
different chatbots are also explained through this paper.
Ontologies such as OpenCyc (Lenat, 1995)have been
widely used in chatbots. ALICE’s main technology, AIML
is most often used in many other chatbots too. Non-
sequitur, simulation of keystrokes and canned response
are some of the language tricks described here which are
used by many intelligent chatbots.
Using chatbots in the field of education can assist people
in many ways. One such application is using Bots as
Language learning tools. Wang and Petrina (Wang and
Petrina, 2013)in their paper presents how to predict and
advise the design of a language tutor called LUCY using
learning analytics. Further the paper describes student
learning methods, data trails, chat log architecture which
could be useful in designing more sophisticated language
learning bots.
Lucy comes in two forms as a commercial chatbot and
as an intelligent chatbot. Commercial chatbot Lucy is
introduced as a digital language tutor to carry extensive
conservation with learners as they speak. Lucy provides
the users with a feedback regarding their pronunciation
and provides exercises. Hosted on Pandorabots, Lucy
is an online tool which guides learning in reviewing
grammar and vocabulary (Jørgensen and Phillips, 2011).
This chatbot Lucy provides several modules such as travel
English, hotel English, restaurant English, causal talks and
helping visitors. The paper describes the importance of
applying learner analytics for understanding the design of
the intelligent chatbots.
Even though the chatbots are designed in a specific
procedure, through research it is found that the linguistic
style of chatbots is altered over time(Ali et al., 2012). The
study done by Ali shows that chatbots depicts a certain
behavioral drift in their styles. Alice, CleverBot, Hal, Jeeney,
SkyNet, TalkBot, Alan, MyBot, Jabberwock, Jabberwacky,
and Suzette have been used for the experiment. The data
for the current study has been collected over years and
contains only the chatbots that were used in their previous
study about “Evaluation of authorship attribution software
on a Chat bot corpus.”(Ali et al., 2011) The experiment
10TH INTERNATIONAL RESEARCH CONFERENCE 981
PROCEEDINGS
has been conducted through a model built for authorship
identification called RapidMiner.
The confidence value of the chatbot has been taken as the
key parameter. Jabberwacky had shown more positive
results compared to the others. These experiments had
several variations in the styles while some styles were
steady. As stated the reason could be their intelligent
algorithms. Performing additional research on these
chatbots are focused as future work in this.
Chai(Chai et al., 2001)provides an effective solution for
information access through a web based natural language
dialog system. This system allows users to find relevant
products on e-commerce sites. The main aim of this
system is to provide a better experience for the users who
are frustrated by menu driven navigation and keyword
search. The dialog system is built on a traditional rule
based technology.
The architecture is a support mixed initiative dialogue
with multiple modalities. The natural language assistant
(NLA) in this system uses a hub and spoke architecture.
The authors have chosen C for the implementation of the
statistical parser due to its efficiency. The system reports
that the average number of clicks has been reduced by
63%. Defining quantitative and objective measures of
system’s success is stated as further work by the authors.
At present chatbots are developed for various domains for
various languages. Poongkuzhaki(T.Kalaiyarasi et al ,2003)
is an intelligent chatterbot developed for Tamil language by
T.Kalaiyarasi. The main function is to produce responses
to given inputs for any existing topic. If the user is in a
pause the system will initiate the conversation. Identifying
key words in Tamil and discovering the minimal context
in the question is stated as the prominent technical
issues faced. The means to overcome such issues are also
addressed here. The time complexity of the system is of
order n where n is the input word count. Voice enabling
and advanced grammar Handling is mentioned as future
work of this project.
IV. CONCLUSION
This paper reviewed a wider range of applications where
chatbots can be used. After various advance studies
chatbots are now used in fields like education, health
982 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
care, language learning etc. Through a proper study of the
techniques and tools that prevail researches can develop
efficient and useful systems to the mankind. Rule based
approaches or pattern matching techniques could be used
to develop a successful system. There is clearly an increasing
trend matching towards to develop wide range resources
and tools to be used in this field. This paper presented
unified overview of some selected papers which involve
techniques and approaches in their development. Using
such approaches along with theories of implementing
chatbots could be used to develop numerous systems to
assist the world.
REFERENCES
Abdul-Kader, S.A., Woods, J., 2015. Survey on chatbot
design techniques in speech conversation systems. Int J
Adv Comput Sci ApplIJACSA 6.
AbuShawar, B., Atwell, E., 2015. ALICE Chatbot: Trials
and Outputs. Comput. Sist. 19. doi:10.13053/cys-19-4-
2326
ALICE(2016),<http://alice.pandorabots.com>
Ali, N., Hindi, M., Yampolskiy, R.V., 2011. Evaluation of
authorship attribution software on a Chat bot corpus,
in: Information, Communication and Automation
Technologies (ICAT), 2011 XXIII International
Symposium on. IEEE, pp. 1–6.
Ali, N., Schaeffer, D., Yampolskiy, R.V., 2012. Linguistic
Profiling and Behavioral Drift in Chat Bots., in: MAICS.
pp. 27–30.
B Hettige, AS Karunananda, 2015. Octopus: A Multi Agent
Chatbot, in: 8th International Research Conference, KDU,
Published November 2015.
Bayan Abu Shawar, Eric Atwell, n.d. An arabic chatbot
giving answers from the Quran.
Bradeško, L., Mladenić, D., 2012. A survey of chatbot
systems through a loebner prize competition, in:
Proceedings of Slovenian Language Technologies Society
Eighth Conference of Language Technologies. pp. 34–37.
 ISBN number: 978-955-0301-41-6
Chai, J.Y., Budzikowska, M., Horvath, V., Nicolov, N.,
Kambhatla, N., Zadrozny, W., 2001. Natural Language
Sales Assistant-A Web-Based Dialog System for Online
Sales., in: IAAI. pp. 19–26.
Comendador, B.E.V., Francisco, B.M.B., Medenilla,
J.S., Nacion, S.M.T., Serac, T.B.E., 2015. Pharmabot: A
Pediatric Generic Medicine Consultant Chatbot. J. Autom.
Control Eng. 3, 137–140. doi:10.12720/joace.3.2.137-140
Eliza(2016),<http://www.manifestation.com/neurotoys/
eliza.php3>
Filho, H. P. (2009). ChatterBean: Flexible Alicebot. http://
www.geocities.com/phelio/chatterbean/
French, R.M., 1990. Subcognition and the limits of the
TuringTest. Mind 99, 53–65.
Hettige, B., Karunananda, A.S., 2006. First Sinhala chatbot
in action. Proc. 3rd Annu. Sess. Sri Lanka Assoc. Artif.
Intell. SLAAI Univ. Moratuwa.
Hettige, B., Karunananda, A.S., 2006. A Parser for
Sinhala Language - First Step Towards English to Sinhala
Machine Translation. IEEE, pp. 583–587. doi:10.1109/
ICIIS.2006.365795
Hettige, B., Karunananda, A.S., 2015. Octopus:A multi
agent chatbot. Proc. 8th International research confernece
, KDU
Jabberwacky. (2016). Jabberwacky-live chat bot-AI
Artificial Intelligence chatbot.< http://www.jabberwacky.
com/>
Jørgensen, M., Phillips, L., 2011. Discourse analysis as
theory and method. Sage, Los Angeles, Calif.
Kazi, H., Chowdhry, B.S., Memon, Z., 2012. MedChatBot:
An UMLS based Chatbot for Medical Students. Int. J.
Comput. Appl. 55.
Lenat, D.B., 1995. CYC: A large-scale investment in
knowledge infrastructure. Commun. ACM 38, 33–38.
Loebner prize”. <http://www.loebner.net/Prizef/loebner-
prize.html lists>
Mitsuku(2016),http://http://mitsuku.com/
Richard S. W A L L A C E, 2003. The Elements of AIML
Style.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Setiaji, B., Wibowo, F.W., 2016. Chatbot Using a Knowledge
in Database: Human-to-Machine Conversation Modeling.
IEEE, pp. 72–77. doi:10.1109/ISMS.2016.53
Shawar, A., Atwell, E.S., 2005. A chatbot system as a tool
to animate a corpus. ICAME J. Int. Comput. Arch. Mod.
Mediev. Engl. J. 29, 5–24.
Shawar, B.A., Atwell, E., 2003a. Using the Corpus of
Spoken Afrikaans to generate an Afrikaans chatbot. South.
Afr. Linguist. Appl. Lang. Stud. 21, 283–294.
Shawar, B.A., Atwell, E., 2003. Using dialogue corpora to
train a chatbot, in: Proceedings of the Corpus Linguistics
2003 Conference. pp. 681–690.
Shawar, B.A., Atwell, E., 2002. A comparison between
ALICE and Elizabeth chatbot systems. University of Leeds,
School of Computing research report 2002.19.
Shawar, B.A., Atwell, E., n.d. CHATBOTS: CAN THEY
SERVE AS NATURAL LANGUAGE INTERFACES TO
QA CORPUS?, in: Proceedings of the Sixth IASTED
Internationa Conference. p. 183.
Turing A.M, 1950. Computing Machinery and Intelligence.
T.Kalaiyarasi, Ranjani Parthasarathi, T.V Geetha, 2003.
Poongkuzhaki- An intelligent tamil chatterbot.
Vales ,A. and Sukhanya, K “Personal healthcare assistant/
companion in virtual world, ”Department of Electrical
Engineering and Computer Science, Los Angeles,
California, 2009.
Wang, Y.F., Petrina, S., 2013. Using learning analytics to
understand the design of an intelligent language tutor–
Chatbot lucy. Editor. Pref. 4.
Webber, G. M. (2005). Data representation and algorithms
for biomedical informatics applications. PhD thesis,
Harvard University.
West, L.H., Fensham, P.J., Garrard, J.E., 1985. Describing
the cognitive structures of learners following instruction
in chemistry. Cogn. Struct. Concept. Change 29–49.
10TH INTERNATIONAL RESEARCH CONFERENCE 983
PROCEEDINGS ISBN number: 978-955-0301-41-6

INTELLIGENT NEWS READER

KLTN Perera 1#, PPNV Kumara1, and B Hettige1
Faculty of Computing
1

General Sir John Kotelawala Defence University, Ratmalana, Sri Lanka.

# starnp8@gmail.com

Abstract - News is one of the most essential requirements I. INTRODUCTION

from the beginning of the history, it is anything that is
exciting, that narrates to what’s happening in the world,
what’s happening in areas of the nation that would be of
interest to the particular audience. With the invention
of media people get to know about the current situation
of the country as well as the whole world through it.
But because of the complex routines of life they are not
able to allocate much time to listen news. Internet has
transformed the boundaries of news with fast and easy
access. But still people bothered about the validity of the
news they heard. To become a good news it should be
clear, precise and brief.
This paper reports the design and implementation of the
“Intelligent News Reader” application which can be known
as a great solution to the busy scheduled users concerning
their thirst of news. Basically this product is based on the
natural language processing, text to speech conversion,
which converts the characters in words into audio format
that sounds just as the words are spoken. News which is
needed for the software is taken from the prominent news
websites with the use of web parsers. The news taken
from the websites are finally converted into a meaningful
format, which can be easily captured by the users.
Keywords - Web Scrapping, HTML parsing, Text to
Speech Conversion, Text comparison algorithm
News has become the most important source to get a
knowledge regarding the day to day events happening
around the globe. News travels through many different
media based on printing, broadcasting, postal systems
and communication. Since majority of the society are
really busy with their routines activities, they cannot
allocate much time for listen or observe news broadcast on
television, Radio Channels or News Papers. Consequently
in the present society most of the people are not aware
about the current situation in the world. Because of that
social media transformed the way consumers receive and
participate in news. Gradually they lured people towards
social media sites, gossip sites, online reading sites which
people consider as unreliable sources. (Alejandro, 2010)
Since most of the people deal with their laptops and
handheld device all the time, and if they are able to listen
for those news while engage in their day to day works it
will be a wonderful experience for them. Intelligent News
Reader is an automated solution for the busy scheduled
users regarding the day to day events around the globe.
It affords news for its users from reliable sources (like
prominent news websites) by extracting news in an
accurate format and speak them out to its users in a clear
and precise manner. A user can separately listen to the
headlines and news in detail by using this application and
he or she can activate this program in any time period of

984 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
day and listen to the news while doing his or her routine
activities.
There are no similar applications to this solution. Currently
there are text readers built from several languages. They
are capable of reading a given text to the listener. For
that speech synthesizers are used. Several text readers
are produced from java, python and C#. Html parsers are
needed for the proposed application to grab the news from
the prominent websites. DOM-based Content Extraction
of HTML Documents is an early research done related
to the topic. Abstraction method of text summarization,
which is based on natural language generation is used
to develop this application.(Michele Banko, n.d.) The
Columbia Newsblaster1 system has been providing
summaries of topically clustered news daily and it aid
daily news browsing by providing an automatic, user-
friendly admittance to important news topics, along with
summaries and links to the original articles for further
information.
This paper reports the design and implementation of
“Intelligent News Reader which is designed to the busy
scheduled users. The design architecture and technology
methods are also given in the paper.
The rest of the paper is organized as follows. Section
2 reports brief summary of the existing frameworks
and systems. Section 3 gives on the technology behind
this application. Then section 4 gives Design and
implementation of the Intelligent News Reader with a
brief description of each module and section 5 depicts
how application works as defined. And also section 6
presents about the results of evaluation. Finally section 7
gives conclusion and further works of the project.
II. CURRENT PRACTICES
Currently there are no similar applications to this product,
but there are applications created by several researchers by
using the key technologies used to develop this application.
Here presents about the key findings which are already
available in the field of technology and technologies which
inspired to carry out this application.
A. Text to Speech Conversion
Text to speech conversion is used to create models of the
human vocal tract to generate a synthetic human voice
output conforming to input text. Text-To-Speech system
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
processes are usually different than live human speech
invention. Text-To-Speech conversion contains of two
parts; namely the Natural Language Processing (NLP)
unit and the Digital Signal Processing (DSP) unit. Natural
Language Processing unit consider phonetization and pitch
along with rhythm and it outputs a phonetic transcript of
the input text. The Digital Signal Processing unit converts
the phonetic transcript it obtains into machine speech.
(World Congress on Engineering et al., 2014)
It should contain text analyzer which contains pre-
processing block which converts numbers, abbreviations
and acronyms into full text when needed. The
morphological analysis block which classifies each term
in the sentence presence analyzed into thinkable parts of
speech, like word’s spelling. The contextual analysis module
which modernizes the list of conceivable parts of speech
of words in sentences and finally the syntactic-prosodic
parser track down the text structure. And Phonetization
model which helps to phonetic transcription of incoming
text and Prosody generation concentration on specific
parts of a sentence, such as emphasis laid on an exact
syllable, thus attributing special position or difference to
that part of the sentence.
Rhythm is a significant factor that marks the synthesized
speech of a TTS system more natural and comprehensible,
the prosodic configuration offers essential information
for the prosody generation model to generate effects
in synthesized speech. Numerous TTS systems are
established based on the principle, corpus-based speech
generation. It is very common for its high feature and
natural speech output.
When relating to text readers, there is a German text-to-
speech synthesis system Mary, which was established as a
research and development tool. It contains with modular
design and XML-based system-internal data illustration.
However, as the MARY system uses an XML based data
representation. (Schröder and Trouvain, 2003) System
receives both plain text input and input marked up for
speech synthesis. The input markup language, currently
SABLE and the W3C draft version of SSML, is interpreted
by this module into the system-internal Mary XML
format, upon which successive modules will operate. The
understanding indications uttered in the input markup are
reflected as supplements to the modules’ text-to-speech
analysis of the input. Each module enhances new or more
comprehensive information. The tokeniser cuts the text
into tokens, i.e. numbers, words, special characters and
10TH INTERNATIONAL RESEARCH CONFERENCE 985
PROCEEDINGS
punctuation marks. It uses a set of rules firm through
corpus analysis to label the meaning of dots based on the
nearby context.
B. html parsers and Web Scrapping
html parsers are needed for the proposed application to
grab the news from the prominent websites. DOM-based
Content Extraction of HTML Documents is an early
research done related to the topic. In directive to examine
a web page for content extraction, the page is first passed
through an HTML parser that modifies the HTML and
creates a Document Object Model tree illustration of the
web page.
Scrappy is intended to scrape web content from sites
that are collected of many pages of comparable semantic
arrangement.(Amir Ghazvinian1, n.d.) The system is
employed as a Firefox browser extension, and works in
three main steps to scrape web data. First, a user steers to
a page that he would like to scrape and makes a model for
the content that he would like from that page. Then, the
user chooses a set of links that point to pages matching
the content template definite by the user. First phase
in obtaining data from the web by Scrappy is to make a
template that can be functional to scrap data from multiple
pages of comparable structure. The user steers to a page
that comprises links to a set of pages corresponding the
template quantified in the previous step. Here, the user can
choice the links that he needs to scrape simply by click on
them. When the user hovers over a link in link assortment
mode, Scrappy highpoints a set of similar links. Clicking
on any of these links will choice and highlight all of them
for scraping, which saves the user time in agreeing links for
scraping. Once the user has quantified a content template
and has designated a set of links to crawl, he simply selects
an output. That is the theory behind the Scrappy which is
developed by the Stanford University.
C. Text Comparison and Processing Method
Since there are same news from different news websites,
it is essential to remove the duplicates. For that text
comparison algorithms have been used. Those algorithms
are called “String metric algorithms”. Such algorithms
are used to find the word patterns with the variances
like insertions, deletions or replacements.(Jokinen et al.,
1988)For the purpose of learning the differences between
words there are several algorithms used in the computer
science field. These algorithms are used in dynamic
programming for the purpose of having an optimized
986 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
result. (Pandiselvam.P and Marimuthu.T, n.d.)(Cohen
et al., 2003)The algorithms can be known as Hamming
distance algorithm, Levenshtein algorithm, Smith
Waterman algorithm, Boyer –moore algorithm and brute
force algorithm.
Concerning the advantages and disadvantages of several
algorithms “Levenshtien algorithm “has been used to
build this application. It calculates minimum number of
character edits required to edit two words. (Cheapest way
to transform one string to another.) This transformation
can be an insertion, deletion or replacement of a character.
This algorithm also can be known as “Edit distance
algorithm”.(Van der Loo, 2014)
When concerning about the usage of this algorithm, spell
checking applications, correction structures for optical
character recognition, and software to support natural
language translation based on translation memory, DNA
analysis, and plagiarism detection can be identified. The
Levenshtein distance can also be calculated between
two lengthier strings, but the cost to calculate it, which
is approximately proportional to the product of the two
string sizes, makes this unrealistic. Therefore, when used
to aid in fuzzy string searching in solicitations such as
record linkage, the matched strings are commonly short
to help increase speed of evaluations.
To find the similarity between the tokens “Bipartite graph”
was used. For that “Assignment problem” is used.(Kuhn,
2010) Bipartite graph is a graph whose vertices can be
separated into two disjoint sets and every edge links a
vertex in to one in and are generally called the parts of the
graph. The bipartite graph configuration can be used to
seizure a relationship between two types of objects where
the dissimilarity between the types of objects is significant.
In order to solve the assignment problem using bipartite
graph, Hungarian method is used. From Hungarian
method, a user can calculate the maximum weight of
bipartite match.
As mentioned earlier, there are several models developed
by several researchers using the technologies which used
in this application. But there is no similar application
like this “Intelligent News Reader”. By evaluating and
considering the limitations and the advantages of those
implementations “Intelligent News Reader” is developed.
Below section put forward about the technologies and
theoretical advancements which have been used to
develop this product.
 ISBN number: 978-955-0301-41-6
III. TECHNOLOGY BEHIND THE
APPLICATION
This section depicts about the theory and technologies lies
behind this application.
A. Text to Speech Conversion
Text to Speech technology is the method that used to
convert the normal text format into an audio format. Here
the sentence which needs to be read should be properly
analyzed. This analyze should cover whether there are
nouns, phrases in the sentence, tense of the sentence and
also the beginning and the end of the sentence.
When concerning about this “Speech Synthesis” process,
there are three major steps related to it. First thing is “Text
Normalization”. Here in text normalization it identifies the
phrases and the parts in the sentence. The second process
is “Linguistic Analysis.(Swetha and Anuradha, 2013) Here
it recognizes the process or the stream of the sentence.
Normally there are alterations between written and
spoken forms of a language, and these variances can lead
to indeterminacy or indistinctness in the pronunciation
of written words. Therefore the step should be properly
covered. The third step is the “Prosody Generation”.
Prosody is the set of structures of speech output that
comprises the pitch (melody or intonation), the rhythm,
the pausing, the speaking rate, the weight on words and
many other features. Producing human-like prosody is
significant for creating the sound of speech more natural
and for properly conveying the meaning of spoken
linguistic.(Dennis H. Klatt, 1987)
Text to speech applications always read sentences as single,
context-independent substances therefore it is needed
to thoroughly consider about the spellings and the text
formatting in the sentence. Commas are used to consider
the pauses in the text and it will create a significant impact
to the synthesis speech. Without commas, Text to speech
sounds too fast and abnormal. There should be intervals
between the each and every sentences to become more
natural. Therefore it is really good to consider about the
correct punctuations in the application.
B. Use of Web Scrapping and xml Document
Most of the Webpages are designed using text based mark-
up languages like HTML and XHTML. To extract the news
from the websites, reaching to its code level is essential.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
According to the requirement, grabbing the headlines and
the detailed news is the major functionality behind the
application. Therefore separate XPath queries have been
written to identify the paths to the relevant news websites.
As per the requirement the application should be able
to format the news into a meaningful manner and the
repeating groups and unnesseray points should be
removed, for this purpose the application loads the news
into a separate XML document, in order to manipulate the
proper outcome from the news. This xml document will
be helpful to analyze the news into a proper format and
gives a meaningful output to the users.
C. Text Comparison and Processing
The news required for this application are taken from
the prominent news websites. Therefore the contents
of those websites can become similar sometimes. If that
kind of situation occur in this application, it will become a
nuisance for the users and it will reduce the quality of the
application. Hence it is essential to remove the duplicative
content from the application. To remove these duplicate
content several string metric algorithms have been used.
(Gao et al., 2010)
As the first method “Levenshtien Algorithm” is used.
Here it consider about the minimum number of character
edits required to transform a one word to another.(Eric
Sven Ristad, n.d.) Here using this algorithm application
calculates this edit distance and greater the levenshtien
distance means strings are different. Since it is essential
to find the stings with smaller differences, this is the
most suitable algorithm. Since the news are created using
sentences, it is needed to divide the each sentence to peace
of information which is technically considered as “tokens”.
The characters which separates the words are recognized
as delimiters (space, commas etc.) and to identify the
delimiters “Regular Expressions are used.
Normally sentences can be considered as a list of tokens.
Therefore to identify the similarities between those
sentences “Hungarian method” was used. This method
was used as reducing the assignment problem from a
bipartite graph. The usage of Hungarian method is to
identify the total weight of this bipartite match. From this
the needed optimized output can be achieved.
Below section depicts about the design and implementation
details of the “Intelligent News Reader” application.
10TH INTERNATIONAL RESEARCH CONFERENCE 987
PROCEEDINGS
IV. DESIGN AND
IMPLEMENTATION OF
INTELLIGENT NEWS READER
This application is developed as a standalone application,
and application is developed using asp.net with the C#.
The application will be able to browse the web when
“Go” button is pressed by the user. The application will
automatically visit the prominent news websites which are
pre-defined by the user. At that time the application will
extracts news from the visited websites and categorized
them into Headlines and News-in-details into an xml
document. After that the extracted news will be processed
according to a proper format. Finally the application will
speak-out those news to the user in clear and precise
manner. The user of the application will be able to change
the volume and frequency of the application according
to the preference. The application will consist with user-
friendly interfaces which can be easily controlled by the
user.
When concerning about the process of the application, the
application can be segregated into three basic functions.
They are Select the appropriate news website, Extract the
news from the website, Order the news in an appropriate
manner, process the news and finally Read out the news
to the user.
Figure 1: Process Diagram of the Application
A. Extracting News
This is the first module of this application. Only for this
module a device with internet access is required. When user
starts the application and press “Go” button, application
will go to the news websites. Here the application will visit
to the default news websites. In addition to that user can
customize the news websites as his or her preference.
988 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
B. Text Processing and Comparison
Text analysis and summarization is the second major
module in the application. This process is happening as
an internal module, when user press “Go” button. As it is
mentioned in the previous section, when user press “Go”
button application will automatically routes to the news
websites and retrieve news from it. Those news are loaded
into an xml document. That xml document is used to do
the text analysis process.
Since application took news from several news websites,
there might be duplicative news in it, if they are available
in the application, it will reduce the productivity of the
application. Therefore it is essential to remove those
content from the application. Therefore when the match
score value is become high, it means the sentences are
similar enough, it means they are duplicative. Because of
that it is essential to remove a one news from it. Here the
application will remove a one news content when there
are same news in different news website. When the words
matching count is become 50% or more than 50% this
functionality works that is the process where it users the
text processing part.
Using the content generated in the xml document, it
analyses the similar content and remove them. After that it
keeps one news content, which is more descriptive and can
be categorized as the best one. The news will automatically
order according to the sequence of headline and news in
details.
C. Presenting News
This is the final output of the application and this is the
output which can be seen and listen by the user. When user
selects the “Play” button. Application will start speaking
V. HOW APPLICATION WORKS
This application is developed using visual C# language
as a windows form application. The only difference lies
here is, it can access internet when necessary and grab the
news. Since this is an application to use in a daily basis, it
should have attractive graphical user interfaces. Therefore
application developed with several animations and eye
catching theme. It consists with three major modules and
finally output is displayed as a single application.
The interfaces are developed to have a high user experience
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

and this is really a user friendly application. When Table 1: Results Coming from Human
user press “Go” button, application is going to specified Evaluator and from the Application
news websites and retrieve the news from those news
websites and load them into an xml document. The news Couple of sen- Match score by Match Score by application
processing part also happening in here. If user want to add tences Human

more sites to the application, user can select the settings Couple 1 71-80 69.23%
button. Using the settings button user can add or remove Couple 2 11-20 35.29%
news websites according to the preference. And also can Couple 3 71-80 71.6%
select whether the application needs to read the news as Couple 4 91-100 94.8%
headlines or detailed news or both.
Couple 5 61-70 56.09%

Figure 2: Interface of application when user

selects play button
Button integration of this application also need to
concern, all the related buttons are activated when they
are necessary to activate. Other time the buttons remains
disabled.
Using the button “Headlines” user can listen to the
headlines only. And also using the volume controller user
can adjust the volume according to the preference.
Another important thing in this application is the
presenting style. It presents news in a very good
pronunciation method and it is presenting style is very
similar to television news presenter .Therefore the
application will become a realistic experience to its user,
though this is an artificially generated application.
VI. RESULTS
The accuracy evaluation was conducted in order to find
whether the match score of “Intelligent News Reader
“ application is similar to the match score given by the
human evaluator. This criteria is really essential to remove
the duplicate news coming from several news websites.
Accuracy testing was conducted by the use of hundred
pairs of sentences. Then the each pair of sentences were
dropped to the demo application which was created for
evaluation purpose and compare the similarity by giving
a match score value. After that each pair of sentences were
given to a human being to consider the similarity of those
sentences. The results match scores of the application and
human evaluator were compared and analyzed properly.
The results of both evaluations are as follows.
Figure 3: Difference between Results Coming from
Human Evaluator and from the Application
Here the test results of human evaluator were appeared
similarly as given by the application. Therefore the selected
criteria for removing the duplicates (lesser than fifty) was
recognized as a correct value.
VII. CONCLUSION AND
FURTHER WORKS
This paper has reported the design and implementation
of the “Intelligent News Reader”. Further it has provided
the evaluation results of the application. By analyzing
the results, it is clear that Intelligent News Reader
application completely meets its expected goals. As the
future enhancement this application will be developed for
Sinhala language.
ACKNOWLEDGEMENT
I acknowledge Mr. Nandana Pathirage and Mr. Buddhitha
Hettige for encouraging me to write this paper, and also
the lecturers who inspire me to write this. The author bears
all the responsibility for any misunderstandings and/or
errors therein.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 989
PROCEEDINGS ISBN number: 978-955-0301-41-6

REFERENCES Pandiselvam.P, Marimuthu.T, n.d. A COMPARATIVE

STUDY ON STRING MATCHING ALGORITHMS OF
Alejandro, J., 2010. Journalism in the age of social media. BIOLOGICAL SEQUENCES.
Reuters Inst. Fellowsh. Pap. Univ. Oxf. 2009–2010.
Schröder, M., Trouvain, J., 2003. The German text-to-
Amir Ghazvinian1, n.d. Scrappy: Simple Web Scraping. speech synthesis system MARY: A tool for research,
Cohen, W., Ravikumar, P., Fienberg, S., 2003. A comparison development and teaching. Int. J. Speech Technol. 6, 365–
of string metrics for matching names and records, in: Kdd 377.
Workshop on Data Cleaning and Object Consolidation.
pp. 73–78. Swetha, N., Anuradha, K., 2013. Text to speech conversion.
Int J Adv Trends Comput Sci Eng 2, 269–278.
Dennis H. Klatt, 1987. Review of text-to-speech conversion
for English. Van der Loo, M.P., 2014. The stringdist package for
approximate string matching. The R.
Eric Sven Ristad, n.d. Learning String Edit Distance1.
Gao, X., Xiao, B., Tao, D., Li, X., 2010. A survey of graph World Congress on Engineering, Ao, S.I., Gelman, L.,
edit distance. Pattern Anal. Appl. 13, 113–129. doi:10.1007/ Hukins, D.W.L., Hunter, A., Korsunsky, A., International
s10044-008-0141-y Association of Engineers (Eds.), 2014. World Congress on
Engineering: WCE 2014 : 2-4 July, 2014, Imperial College
Jokinen, P., Tarhio, J., Ukkonen, E., 1988. A Comparison of London, London, U.K.
Approximate String Matching Algorithms. SOFTWARE—
PRACTICE Exp. 1, 1–4.
.
Kuhn, H.W., 2010. The Hungarian Method for the
Assignment Problem, in: Jünger, M., Liebling, T.M.,
Naddef, D., Nemhauser, G.L., Pulleyblank, W.R., Reinelt,
G., Rinaldi, G., Wolsey, L.A. (Eds.), 50 Years of Integer
Programming 1958-2008. Springer Berlin Heidelberg,
Berlin, Heidelberg, pp. 29–47.

Michele Banko, n.d. Open Information Extraction from

theWeb.

990 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

WEB-BASED EXPERT
SYSTEM FOR PERSONALIZED
PSYCHOTHERAPEUTIC COUNSELLING
S Eshwarage1# and ADAI Gunesekara1
Faculty of Computing,
1

General Sir John Kotelawala Defence University, Ratmalana, Sri Lanka

# sahanieshwarage2012@gmail.com

Abstract - Sri Lanka has one of the highest suicidal rates in
the world. It has been found that these cases were caused
mostly as a result of mental health disorders, such as
depression and anxiety. A recent study conducted at the
Rajarata University showed that approximately 5-10%
of the Sri Lankan population suffer from some form of
mental health condition that require treatment.(Jaufer,
2016) This study aims to examine the development of a
Web-Based Expert System which will provide Personalized
Psychotherapeutic Counselling for anyone in any location.
If everyone who has a suffering problem and have no one
to share their grievances could use the proposed system,
then everyone will have a hope and happiness in their
lives and the suicidal rate would decrease significantly.
The main objective of developing this system is to address
all the drawbacks of the currently available face to face
counselling system, while this system mainly focuses the
Sri Lankan community. This system will perform as a
platform to connect clients who need mental health advices,
with licensed counsellors of Sri Lanka. The significance of
the proposed web based counselling system is, that it will
present a set of standard and psychologically approved
questions to the clients before receiving the counselling
service. By using Prolog Logical Programming Language,
the answers provided by the client will be analyzed, and
then the system will predict the area of the problem of that
particular client. Then, the system will direct the client to
the relevant counsellor that has specialized on that relevant
area. Finally, the client could use a preferred method of
communication facilitated by the system, to communicate
with the counsellor. The system will also use encryption
methods to ensure confidentiality of the conversations
carried out between clients and counsellors. The proposed
system will be a fully integrated version comprised of a
Web Application, Android Mobile Application, Short
Message Sending (SMS) gateway and a Payment gateway.
Keywords - Personalized Online Counselling, Web Based
Application, Psychological Analysis, Prolog
I. INTRODUCTION
According to the statistics by the National Institute of
Mental Health (NIMH), only 20% of mental health
sufferers receive treatment, one of the main reasons
being that treatment is usually available only in cities.
In addition, the main problem is that Sri Lankan society
neglect mental health-care or counselling due to social
stigma. (Jaufer, 2016)
According to Prof. Nalaka Mendis, Professor of Psychiatry
Faculty of Medicine, Colombo and Vice President
Sahanaya, National Council of Mental Health, there has
been a definite growth in the number of those who suffer
from mental illnesses during the last ten years. Resident

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 991
PROCEEDINGS
Representative of World Health Organisation (WHO),
Dr. Kan Tun has stated that 10% of the global problem of
diseases are due to mental illness. Further he has stated that
mental illnesses are commonly investigated in developing
countries and especially in those countries where there
was a civil war. Dr. Neil Fernando, Consultant Psychiatrist
has stated that forces personnel have become mentally ill
solely due to the war. According to Prof.Mendis although
Sri Lanka has cheap and cost effective counselling services,
the country lacks experienced doctors in the field and
also the professionals in this field are mainly available in
urban areas so the clients tend to neglect their need for
counselling service and remain suffering and worrying.
Ultimately they choose suicide as a remedy to get rid-off
their untold problems. (“Mental illness on the increase
-Sahanya-Island,” 2001)
Most of the people keep away from face to face counselling
services due to many concerns. A major reason is due to
lack of knowledge. Many individuals especially in rural
areas lacking education has no any idea about psychology,
so even if they actually suffer from a mental illness they
identify it as a Mysterious force and go for mythological
remedies. Some other problems why people neglect
counselling is due to financial inabilities and lack of
transportation facilities. Some individuals with mobility
difficulties or physical inabilities such as paralysis or
problems in limbs will not be able to meet a counsellor.
So they tend to remain silent and suffer alone. Reluctance
to be exposed to the society is another main reason why
people disregard counselling. Negligence of counselling or
psychotherapy due to public humiliation is a common and
a major reason in our country. Sri Lankan society consider,
getting psychological advices or meeting a counsellor as a
shame and label such a person as a “retard”. If clients who
use to attend to continuous counselling sessions, will have
to migrate to another country, the client will no longer be
able to meet the counsellor. Boundary barriers is another
reason why clients terminate counselling.
According to my perception and study, “Online
Counselling” would be the best solution to address these
problems stated above. It is conventionally assumed
that the first professional fee-based online individual
psychotherapy started in 1995. (Metanoia,1972)
Online Counselling is a field that shows a tremendous
growth in the past ten years with the advancements in
technology and widespread use of the internet. Studies have
identified that more clients are seeking help from online
992 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
counselling services. (Murphy et al., 2009). Currently, an
Online Counselling System which focusses mainly the
Sri Lankans, is not available. The proposed web based
psychotherapeutic counselling system (YourCounsellor)
will be designed essentially focusing the Sri Lankan
community. But also the system enables any personal in
any part of the world to interact with the system and to get
counselling service. Online Counselling will never replace
the traditional method of face-to-face therapy. Researches
that have been conducted to find whether the clients
choose Online Counselling over face to face counselling
have identified that both the methods are accepted equally
by the society. E-therapy is another way of reaching people
who are unable to get any other help.
(Murphy et al., 2011), (Cook and Doyle, 2002)
II. LITERATURE REVIEW
A Google search on “Online Counseling” retrieved about
4,060,000 results, (latest result - 59,100,000 ) the top
most included famous websites which provide online
chat facility to acquire counseling services.(Dowling and
Rickwood, 2013)
Some of the most popular websites available in present day
which provide online counselling services are:
• 7 Cups of Tea
• Betterhelp
• Breakthrough System
• Butterfly
• kidsHelpline
• MindSpot
• Online Counselling directory
• PlusGuaidance
• Talkspace
• YourDOST
A.Technology
All of these web based systems uses web technology.
But some of the sites such as breakthrough, betterhelp
and talkspace provide mobile application facilities too.
The proposed system will use both web and mobile
 ISBN number: 978-955-0301-41-6
technologies to increase the efficiency and usability.
(BetterHelp, 2013), (© Breakthrough and Policy, 2009),
(Talkspace, 2014), (Baumel and Schueller, 2016).
B. Bi-Lingual Interface Navigation
The proposed system will provide Bi-Lingual Page
Content facility which means that the system enables the
clients and counsellors to interact with the system in a
preferred language, either English or Sinhala. So that even
a person who has less knowledge in English could navigate
via pages and use this system effectively to get counselling
advices. None of the above mentioned websites provide
bi lingual or multi lingual facility. So it proves that the
proposed system carries a distinct feature than the other
available Online Counselling sites.
C. Professional Counsellors
The proposed system only allows Qualified, Licensed
Professional Counsellors to be registered into the system
since the system tackles with the human brain and mind,
it is risky to allow clients to get help from therapists
who are not qualified. Among the above mentioned
systems MindSpot, betterhelp, breakthrough, talkspace,
PlusGuidance provide professional advice. Other systems
recruit any type of listener and allow them to advice the
clients. (“MindSpot Clinic | Home.”) o. The proposed
system will use both web and mobile technologies to
increase the efficiency and usability. (BetterHelp, 2013),
(© Breakthrough and Policy, 2009), (Talkspace, 2014),
(Baumel and Schueller, 2016).
D. Bi-Lingual Interface Navigation
The proposed system will provide Bi-Lingual Page
Content facility which means that the system enables the
clients and counsellors to interact with the system in a
preferred language, either English or Sinhala. So that even
a person who has less knowledge in English could navigate
via pages and use this system effectively to get counselling
advices. None of the above mentioned websites provide
bi lingual or multi lingual facility. So it proves that the
proposed system carries a distinct feature than the other
available Online Counselling sites.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
E. Personalized Counselling
The proposed system will provide therapy only by
analyzing the client’s mental level. The system will ask
a set of questions from the client so that the client’s
mental level will be measured and the basic problematic
area of the client could be identified. The client will be
provided with the most appropriate counsellor who has
specialized in that relevant area of psychological issue.
Systems such as betterhelp and PlusGuaidance also have
the feature to measure the mental level but the proposed
system is unique, that it only enables the clients to receive
counselling after the mental level analysis. This method
enables the client to get advices only from the relevant
counsellor.
F. Payment Method
Almost every other system provides therapeutic sessions
after a payment. Butterfly website will not allow to access
into any option without paying and signingup. Only a few
like PlusGuidance and 7 Cups provide a free session for
a limited time. Only the proposed system will permit a
3 Day Free Counselling Session Period which will allow
the clients to check whether the counselling session worth
enough. (“7 Cups Subcommunities.”), (“Chat Online » The
Butterfly Foundation.”)
G. Group Counselling
Free Group Therapy Platforms are only available in 7 Cups
among the mentioned websites. Some of them provide
group therapy session but for a fee. The significance of
this proposed system is that all group session will be freely
available for any user.
H.Information Security
Information Privacy or confidentiality is a major feature
that should be considered in online counselling. The
proposed system will be developed, following various
security mechanisms such as encryption, VoIP, IM
protocols and authentication mechanisms.
I. Other Features
Most of the above mentioned have complicated interfaces
and also some of the pages in websites such as kidsHelpline
has expired. Users might not be able to interact properly
and easily since the interfaces are full of bulky information
all over the page. The proposed system will have user
friendly interfaces with clear navigations and guidelines
10TH INTERNATIONAL RESEARCH CONFERENCE 993
PROCEEDINGS ISBN number: 978-955-0301-41-6

to use the system and get the ultimate services from the IV. CONCEPTUAL MODEL
system. (“Teen Online Counselling | Kids Helpline |
1800 55 1800.”)

All the users should be able to access an online counselling

system regardless the physical inabilities. In the proposed
system a feature will be included where the mouse could
be completely controlled without hand movements but
eyes or head. So that all the users will have the ability to
interact with the system. None of the above mentioned
systems have a feature as explained.
If a counsellor which the system chooses as a match to
be communicated with the client is not online, the System
will notify the counsellor via an SMS notification to that
particular counsellor. None of the above mentioned web
Figure 1. Overall Conceptual Model
based systems imposes this feature as well. Which proves
that the proposed system is unique than the currently
available Online Counselling System. V. APPROACH
The main users of this system are the clients and
III. RESEARCH METHOLOGY counsellors. In addition, a system administrator can also
access the system. According to the above illustrated figure
1, there are five main modules of this system. Namely they
Since the proposed system will be finally used by the general are; Admin Module, Client Module, Counsellor Module,
society and the counsellors, the opinion of counsellors System Module and Group Counselling Module.
and the society should be take into consideration. So as
to gather information on their views, both qualitative A. Admin Module
and quantitative methodologies will be used as data
Admin do not perform a major role in the proposed
gathering techniques to design the requirement
system. Admin will have access to view counsellor details,
specification. Questionnaires and Interviews will be the
view the number of clients registered, view feedbacks and
main data gathering methods that will be used to gather
view counsellor names who have been reported by other
the requirements and the information. All the required
counsellors and also the upgrades and maintenance will
data will be collected by distributing a well detailed
be performed.
questionnaire among the general public via both online
and offline methods, to grasp the opinion on introducing B. Client Module
the proposed system which will help to identify what
Clients will have to select the language (English or
they prefer from such a system and also to get a statistical
Sinhala) before any process. And also, clients with any
clarification to identify the percentage of people who
physical impairment can enable the accessible mode on
use mobile phones or computer along with Internet, to
the start-up page and control the mouse using eye or head
ensure whether the system could be used by common
movements throughout the system. Then, the system will
people without any difficulty and barrier. Interviews will
provide a client who is willing to obtain the counselling
be conducted with counsellors and psychiatrists to ensure
service for the first time, a set of psychologically approved
whether the proposed system will be efficient, helpful and
questions before the first session. So that the answers will
ethical enough to serve the common people.
be analysed by the system and will predict the mental level
or the problematic area of the particular client.

Then, the system will direct the client to the most

specific counsellor who has specialized on that relevant
problematic area. The clients can access the system and

994 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
obtain counselling service for free only for 3 days and then,
after 3 days the clients should get registered to the system
by creating a preferred username and a password. Clients
will be facilitated with latest methods of communication
such as instant messaging, voice messaging, video
conferencing and many other media sharing means, to
communicate with the relevant counsellors.
Clients can give feedbacks and also will have to rate the
counsellors after every session. Clients will not be able to
view all the counsellors in the system. The client can only
view all the details such as the full name, qualifications,
licence number of the counsellor who will be selected
by the system which is suitable to that relevant client.
Clients can also request another counsellor if he/she is not
satisfied with the provided counsellor. But this request can
be performed only within the free trial period. The client
could terminate at any time he/she prefers.
C. Counsellor Module
Counsellors will also be provided with Language selection
and accessible mode in the start-up page. Counsellors
will have to get registered to the system at the beginning.
The system will approve the registration by analysing the
details provided by the counsellor. Specifically, the license
number provided by a counsellor will be checked from
a dataset which contains licence numbers of legitimate
counsellors approved by the Sri Lankan government or
Internationally. The registration approval will be given
to the particular counsellor, only if the license number is
available in that dataset.
After registration the counsellor can login to the system
and carryon with the sessions. Counsellors can also give
feedbacks to the system. Counsellors can view all the
details of other counsellors registered in the system, so
that if a counsellor identifies an illegitimate counsellor by
any chance, he/she can report that particular counsellor.
D. System Module
As this is an expert system, the system itself performs huge
complex tasks. The system will analyse the answers given
by the clients to the set of standard questions and predict
the accurate problematic area or the mental level and
direct the client to the specific counsellor. And also before
allowing the counsellors to join the system, the system
will check for the validity of the provided license number.
If any counsellor is offline the system will send an SMS
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
notification to that particular counsellor’s mobile phone.
Another function is that, if a counsellor be reported by 5
other counsellors the system will remove that particular
counsellor who has been reported.
E. Group Counselling Module
Group Counselling facility will also be provided by the
proposed system. This will be an open space that allows
any person regardless whether it is a client or a counsellor,
to post their problems anonymously. Then the other
members could give their opinions.
F. Technology adopted
The most appropriate technology for developing the
system have been decided by considering the system
domain and the requirements. It is much significant
to select the technological methodologies which will
be capable to satisfy both the functional and non-
functional requirements of the system. Since the proposed
Personalized Online Counselling Expert System is a web
based system, more consideration will be given in selecting
latest technologies. The technological methodologies will
be selected in such a manner that it will help the system
to be available at anytime, anywhere and also to make it
efficient and effective.
With an in-depth analysis of the system requirements,
PHP will be used as the main programming language to
develop the backend of the system. Polog is the substantial
programming language that will be used to develop the
complex analysis and prediction functions of the system.
The knowledge base will be created by a deep medical
analysis to identify symptoms of each area of medical
disorder and the relevant counsellor type. The gathered
facts will be logically categorized and programmed for the
prediction purpose using Prolog and will be connected to
the Web Application. The database of the system will be
implemented using MySQL. The front end of the system
will be developed using the programming languages
HTML and CSS and also the bootstrap framework will
be used to make flexible and user friendly interfaces. The
validations of the forms and access controlling will be
done using JavaScript. The system will guide the users to
create a highly secured password to confirm privacy and
data security. The security of online transaction details
will be ensured by using data encryption mechanisms and
technologies such as SSL to prevent eavesdropping and
Man-In-The-Middle attacks.
10TH INTERNATIONAL RESEARCH CONFERENCE 995
PROCEEDINGS
CameraMouse Software will be embedded with the Web
Application which enables the user to move the mouse
with only eye or head movements. CameraMouse is a
simple software with a simple user interface where the
user must allow the camera to capture the real time face
image by which it allows the user to select a point on the
head, specifically eyes, and there onwards the mouse will
follow that particular point. The user could navigate the
selected point to left or right to move the cursor left or
right respectively.
VI. EVALUATION
An evaluation will be performed to verify whether the
requirements and the objectives have been fulfilled by
the system. The proposed system will be evaluated by
employing three test cases. The first test case will be done by
requesting a set of selected counsellors to interact with the
system, check the knowledge base used for measuring the
basic mental level or the problematic situation of clients,
the process of directing the clients to relevant counsellors
is medically approved and finally to get the approval to
implement the system. The second test case will be done
after receiving the approval from the professional and
reputed counsellors. Then, a selected group of common
people with various cultures, and people with different
incapability and people from different areas of Sri Lanka
will be expected to interact with the system and will be
checked whether the interfaces are user friendly and
understandable for everyone. Finally, after the approval of
the above stated test cases, the third test case will be carried
out. Here, both the clients and counsellors will be allowed
to use the system to communicate with each other “online”
and as well as “in-person” to crosscheck whether the
predictions performed by the system is accurate and also
the ethical issues will be discussed so that steps could be
taken to reduce such ethically problematic circumstances.
VII. CONCLUSION AND
FUTURE WORK
The Proposed System covers the entire online counselling
process with more enhancements than the existing online
counselling systems, such as the opportunity to receive
personalized counselling advices from the most suitable
counsellor and the possibility to use various methods of
communication by the clients and the counsellors. The
service provided by the proposed system will be beneficial
996 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
for both ordinary and handicapped individuals, since the
system has an accessible mode which allows the clients,
particularly who have less capacity to handle a mouse by
their own, to control the mouse pointer by their eye or head
movements. As well as that, any person of any part of the
world will have the opportunity to access this system. Since
this system only get the service of legitimate counsellors,
the advices provided to the clients will be consistent. This
system will also increase the quality of life of people. From
1995, Online Counselling has provided an important
service for people, to face their life dilemmas positively.
Since then, various enhancements have been done to
improve the effectivity and the interactivity of the Online
Counselling process. The proposed system will use latest
technologies to develop the system. Future enhancements
can be performed, such as, enabling Navigation in Tamil
Language and also language translation in real time
instant messaging, so that counsellors that communicate
in a particular language could easily communicate with a
client that uses any language without any barrier.
REFERENCES
© Breakthrough, © Breakthrough, Policy, I. 2009-2017 |
P., 2009. Breakthrough: Confidential video counseling and
online therapy [WWW Document]. Breakthrough. URL
https://www.breakthrough.com/ (accessed 3.25.17).
Baumel, A., Correll, C.U., Birnbaum, M., 2016. Adaptation
of a peer based online emotional support program as an
adjunct to treatment for people with schizophrenia-
spectrum disorders. Internet Interventions 4, 35–42.
doi:10.1016/j.invent.2016.03.003
Baumel, A., Schueller, S.M., 2016. Adjusting an Available
Online Peer Support Platform in a Program to Supplement
the Treatment of Perinatal Depression and Anxiety. JMIR
Mental Health 3, e11. doi:10.2196/mental.5335
BetterHelp, B., 2013. BetterHelp - Counseling [WWW
Document]. BetterHelp - Counseling. URL https://www.
betterhelp.com/ (accessed 3.25.17).
Chester, A., Glass, C.A., 2006. Online counselling: a
descriptive analysis of therapy services on the Internet.
British Journal of Guidance & Counselling 34, 145–160.
doi:10.1080/03069880600583170
Cook, J.E., Doyle, C., 2002. Working Alliance in
Online Therapy as Compared to Face-to-Face Therapy:
Preliminary Results. CyberPsychology & Behavior 5, 95–
105. doi:10.1089/109493102753770480
 ISBN number: 978-955-0301-41-6
DigitalMarketer, n.d. Internet usage statistics in Sri Lanka
[WWW Document]. URL http://www.digitalmarketer.
lk/internet-usage-statistics-in-sri-lanka.html (accessed
3.26.17).
Dowling, M., Rickwood, D., 2013. Online Counseling and
Therapy for Mental Health Problems: A Systematic Review
of Individual Synchronous Interventions Using Chat.
Journal of Technology in Human Services 31, 1–21. doi:10
.1080/15228835.2012.728508
Find Online Counselling, n.d. Find Online Counselling
Today - A list of the best in the field. [WWW Document].
Online Counselling. URL https://www.onlinecounselling.
com/ (accessed 3.25.17).
Fletcher-Tomenius, L., Vossler, A., 2009. Trust in online
therapeutic relationships: The therapist’s experience.
Counselling Psychology Review 24, 24–34.
Jaufer, A., 2016. Tackling Mental Health in Sri Lanka:
Fact vs. Stigma [WWW Document]. Roar.lk. URL http://
roar.lk/features/tackling-mental-health-sri-lanka-fact-vs-
stigma/ (accessed 4.29.17).
Jennings, R.B., Nahum, E.M., Olshefski, D.P., Saha, D.,
Shae, Z.-Y., Waters, C., 2006. A study of internet instant
messaging and chat protocols. IEEE Network 20, 16–21.
Kolog, E.A., Sutinen, E., Vanhalakka-Ruoho, M., 2014.
E-counselling implementation: Students’ life stories and
counselling technologies in perspective. International
Journal of Education and Development using Information
and Communication Technology 10, 32.
Li,Lau, P., Jaladin, R.A.M., Abdullah, H.S., 2013.
Understanding the Two Sides of Online Counseling and
their Ethical and Legal Ramifications. Procedia - Social
and Behavioral Sciences 103, 1243–1251. doi:10.1016/j.
sbspro.2013.10.453
Mental illness on the increase -Sahanya-Island, 2001. .
Island.
Metanoia, 1972. Internet Therapy Guide: History and
Survey of E-Therapy.
MindSpot Clinic, n.d. MindSpot Clinic | Home [WWW
Document]. URL https://mindspot.org.au/ (accessed
3.25.17).
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Murphy, L., Mitchell, D., Hallett, R., 2011. A comparison
of client characteristics in cyber and in-person counseling.
Studies in health technology and informatics 167, 149–
153.
Murphy, L., Parnass, P., Mitchell, D.L., Hallett, R., Cayley,
P., Seagram, S., 2009. Client Satisfaction and Outcome
Comparisons of Online and Face-to-Face Counselling
Methods. British Journal of Social Work 39, 627–640.
doi:10.1093/bjsw/bcp041
My Growth Path | 7 Cups [WWW Document], n.d. URL
https://www.7cups.com/member/ (accessed 4.29.17).
PlusGuidance, n.d. PlusGuidance -from Professional
Therapists [WWW Document]. PlusGuidance | Online
Counselling | Online Therapy. URL https://plusguidance.
com/ (accessed 3.25.17).
Priya, N.B., Juvanna, I., 2014. AN ANDROID
APPLICATION FOR UNIVERSITY ONLINE
COUNSELING.
Rhee, H.-S., Kim, C., Ryu, Y.U., 2009. Self-efficacy
in information security: Its influence on end users’
information security practice behavior. Computers &
Security 28, 816–826. doi:10.1016/j.cose.2009.05.008
Russell Elleven, Jeff M Allen, 2004. Applying_Technology_
to_Online_Counseling. Journal of Instructional
Psychology 31, 223–227.
Schrittwieser, S., Frühwirt, P., Kieseberg, P., Leithner,
M., Mulazzani, M., Huber, M., Weippl, E.R., 2012. Guess
Who’s Texting You? Evaluating the Security of Smartphone
Messaging Applications., in: NDSS.
Talkspace, 2014. Talkspace | Online Therapy | Counseling
Online | Marriage Counseling [WWW Document]. URL
https://www.talkspace.com (accessed 3.25.17).
Tanrikulu, İ., 2009. Counselors-in-training students’
attitudes towards online counseling. Procedia - Social
and Behavioral Sciences 1, 785–788. doi:10.1016/j.
sbspro.2009.01.140
Teen Online Counselling, n.d. Teen Online Counselling |
Kids Helpline | 1800 55 1800. Kids Helpline - Teens.
10TH INTERNATIONAL RESEARCH CONFERENCE 997
PROCEEDINGS ISBN number: 978-955-0301-41-6

The Butterfly Foundation, n.d. Chat Online  » The yourDOST, n.d. Online Counselling & Emotional
Butterfly Foundation [WWW Document]. URL https:// Wellness Coach | YourDOST [WWW Document]. URL
thebutterflyfoundation.org.au/our-services/talk-to- https://yourdost.com (accessed 3.25.17).
someone/chat-online/ (accessed 3.25.17).
ACKNOWLEDGMENT
Watzlaf, V.J., Moeini, S., Matusow, L., Firouzan, P., 2011.
VOIP for telerehabilitation: A risk analysis for privacy, Authors would like to express their gratitude to all the
security and HIPAA compliance: Part II. International lecturers of faculty of computing and Dr. Lakshika
Journal of Telerehabilitation 3, 3. Liyanage. Special thank goes to the supervisor Mr. ADAI
Gunasekara, for the immense support and guidance
Wells, M., Mitchell, K.J., Finkelhor, D., Becker-Blease,
K.A., 2007. Online Mental Health Treatment: Concerns
and Considerations. CyberPsychology & Behavior 10,
453–459. doi:10.1089/cpb.2006.9933

998 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
A COMPARISON OF CLASSICAL
STATISTICAL & MACHINE
LEARNING TECHNIQUES IN BINARY
CLASSIFICATION
KVU Perera1# and SD Viswakula1
1
Department of Statistics, University of Colombo, Sri Lanka
# vijiniup15@gmail.com
Abstract - Predicting a precise response for previously
unseen input variables is a vital and challenging task,
as precise predictions can minimize the risks related to
different domains by making correct decisions. The main
objective of this study was to compare the performance of
several classical statistical and machine learning techniques
by considering the prediction task as a binary classification.
The classification techniques; Logistic Regression (LR)
and Linear Discriminant Analysis (LDA) were considered
under classical statistical techniques while Random Forest
(RF), Naïve Bayes (NB), Boosting (BT) and Bagging (BA)
were considered under machine learning techniques. The
performance of those techniques were compared under
the two different aspects by using five real datasets. In
one aspect, class imbalance was artificially introduced to
the datasets by resampling. In the other aspect sampling
approaches such as undersampling, oversampling and
hybrid approach (mix of both undersampling and
oversampling) were considered, to overcome class
imbalance in the training set. Several evaluation methods
such as accuracy, precision, F-measure, G-mean and
Receiver Operating Characteristics Area Under Curve
(ROC AUC) were considered to evaluate the performance
of the classification techniques. The results indicated that
the performance of Random Forest and boosting are
better than the performance of other techniques in both
resampling and overcoming class imbalance aspects. In
many cases when the training set was balanced, not only
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
the machine learning techniques but also the statistical
techniques had better performance.
Keywords - Statistics, Machine Learning, Classification,
Resampling, Class Imbalance
I. INTRODUCTION
Predicting a response for future input variables
(prediction) is one of the intentions in analyzing data. In
order to approach this intention, the classical statistical
techniques as well as the machine learning techniques
can be used. When considering the areas such as health,
education, finance, etc., predictions play a major role
in decision making. Precise predictions can avoid
uncertainties in decision making and then the risks can be
minimized. Therefore, the most vital and challenging task
is to predict responses correctly as much as possible for
input variables which have never seen before with the aid
of existing data. Obviously, this task highly depends on the
technique (either classical statistical or machine learning)
which is used for prediction. So, this study concerns
in comparing several classical statistical and machine
learning techniques by considering binary classification as
the prediction task. Simply, binary classification is the task
in allocating the observations of a given dataset into two
categories according to a particular classification rule. The
10TH INTERNATIONAL RESEARCH CONFERENCE 999
PROCEEDINGS
classification techniques Logistic Regression and Linear
Discriminant Analysis are considered under classical
statistical techniques while Random Forest, Naïve Bayes,
Bagging and Boosting are considered under machine
learning techniques in this study (Hastie et al., 2008).
In general, both classical statistical and machine learning
techniques have the same concern, which is basically
learning from data, but in a different manner, where
classical statistical techniques check for assumptions while
machine learning techniques are used as black boxes
(Breiman, 2001). According to the literature, the amount
of data involved and the contribution from humans to
build models differ for these two techniques.
When considering a particular scenario, different
classification techniques have different performance in
classifying the response. Sometimes classical statistical
techniques might classify the response precisely by
providing a high predictive accuracy than the machine
learning techniques or in converse manner. In some other
cases, both techniques might provide the same predictive
accuracy. Therefore it is worth to identify when to use
what.
The main objective of this study is to compare the
performance of several classification methods (classical
statistical and machine learning techniques) by
considering following aspects.
• Data resampling aspect
• Overcoming class imbalance aspect
The other objective is to identify whether there is an effect
from the above mentioned two aspects for the performance
of classification methods.
II. METHODOLOGY &
EXPERIMENTAL DESIGN
A. Experimental Aspects
This study concerned on the two experimental aspects
which were depended in resampling and overcoming class
imbalance phenomena. The software R was used for this
study. The theories of used methods can be explained as
below.
1000 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
1) Resampling: The method in drawing repeated
samples from the original data called as resampling.
In resampling, the randomized cases are selected
with replacement from original data in such a way
that the each sample size is equal to the number of
observations in original dataset. Due to replacement of
observations, the drawn samples by using resampling
methods consist of repetitive cases.
2) Overcoming Class Imbalance Phenomena: The class
imbalance phenomena can be defined as, “A dataset
is imbalanced if the classification categories are not
approximately equally represented” (Chawla, 2005, p.
853). In an imbalance dataset, the class having a low
number of occurrences is named as the minority class
while the class having comparatively more number of
occurrences is named as the majority class.
According to Brownlee (2015), the problem with the
imbalance phenomena is that most of the machine learning
algorithms may favor majority class instances. That
means there is a high chance in classifying minority class
instances as majority class instances (misclassifying the
minority class instances). Hence achieve a high accuracy.
Among the approaches available to deal with class
imbalance, the sampling methods such as undersampling,
oversampling and hybrid (mix of both undersampling and
oversampling) approaches were considered in this study.
• Undersampling - This approach works with the majority
class. In order to balance the distribution of classes in a
dataset, this method removes a number of observations
randomly from the majority class. Discarding valuable
information is the major drawback of this method.
• Oversampling - This method works with the minority
class. In order to balance the distribution of classes
in a dataset, this method replicates the observations
randomly from minority class. Over fitting to data is the
major drawback of this method.
• Hybrid Approach - This method is the combination of
undersampling and oversampling approaches.
B. Cross Validation Technique
Cross validation is a model evaluation method in
determining how the outcome of the considered model
will generalize to a new dataset. The basic idea of cross
validation is to separate the dataset into two non-
overlapping sets where one is used for training the model
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

and the other set is used for testing the performance Table 1. Confusion Matrix
of learned model. The hold-out and the k-fold cross
validation methods are the two well-known cross Predicted Class
validation techniques (Schneider, 1997). According to the Positive (P) Negative
literature, many comparative studies were carried out by (N)
using the hold out method. Therefore by considering that
Positive (P) True Positive False Nega-
aspect as well as the advantages, the k-fold (k=10) cross True (TP) tive (FN)
validation technique was used in this study. Class
Negative False Posi- True Nega-
(N) tive (FP) tive (TN)
1) k-fold Cross Validation Technique: In this method, the
dataset is randomly divided into k mutually exclusive,
approximately equal size of subsets (folds). One fold
is reserved for testing while all the other (k-1) folds (1)
(all together) are used for training. Then this process
is repeated for k times, by using each fold exactly once
as the testing set. Finally, the average of the evaluation (2)
measures obtained from each time is used to assess
the performance of model. The mostly used value
for k is 10. Since there is a guarantee of using each
(3)
and every data point to train the model, the accurate
performance information can be obtained. Also the
variance of resulting measures is reduced when the
(4)
k is increased. The high computational time due to
repetition process can be considered as a disadvantage 2) Combined Evaluation Measures: The F-measure and
of this method. the G-mean are the combined evaluation measures,
considered in this study.
C. Evaluation Measures
(5)
Since this study was a comparison between classical
statistical and machine learning techniques in binary
classification task, the best technique was determined (6)
by comparing the performance between the techniques.
The performance of evolved models and methods was 3) Graphical Evaluation Measures: In this study, the
evaluated by using certain evaluation criterions such as evaluation measure, Area Under Curve (AUC) in Receiver
the accuracy and the predictability of the predicted model. Operating Characteristics (ROC) graph was considered.
Since the class imbalance phenomena was existed in some A ROC graph is a technique for visualizing, organizing
of the datasets used for this study, not only the individual and selecting classifiers based on their performance.
evaluation measures but also the combined and the Each point on the ROC plot represents a sensitivity –
graphical evaluation measures are considered. false positive rate (1 - specificity) pair corresponding to
a decision threshold. A test with perfect discrimination
1) Individual Evaluation Measures: Precision, accuracy, (no overlap in the two classes) has a ROC plot that passes
sensitivity and specificity are the individual evaluation through the upper left corner (100% sensitivity and 100%
measures which can be obtained from the confusion specificity). Therefore, the overall accuracy of a particular
matrix (Kohavi & Provost, 1998). A two-by-two test is higher when the ROC plot is closer to upper left
contingency table which is called as confusion matrix corner. In most of the cases, ROC AUC values lie between
can be shown as in Table 1. the 0.5 and 1 where 0.5 represents a worthless test and 1
represents a perfect test.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1001
PROCEEDINGS
D. Datasets
The five different datasets retrieved from well-known
database called UCI Machine Learning Repository were
used for this study. They contain real and generated data
with respect to different domains. The characteristics
of a dataset such as number of observations, number of
variables and types of variables (continuous, categorical
and mix of both) are varied for considered datasets. Since
the prediction task is binary classification, the response
variables in all five datasets contain only two categories.
The relevant links of the datasets are provided in following
Table 2.
Table 2. Downloaded links of the datasets
Dataset Downloaded Link
Gamma Telescope <https://archive.ics.uci.edu/ml/
datasets/MAGIC+Gamma+Tele-
scope>
Wilt <https://archive.ics.uci.edu/ml/
datasets/Wilt>
Bank Marketing <https://archive.ics.uci.edu/ml/
datasets/Bank+Marketing>
German Credit <https://archive.ics.uci.edu/ml/
datasets/Statlog+(German+Cred-
it+Data)>
Tic-Tac-Toe <https://archive.ics.uci.edu/ml/
datasets/Tic-Tac-Toe+Endgame>
E. Evaluation Procedures
1)Procedure in Resampling Aspect: In this aspect
classification techniques were compared while introducing
the class imbalance phenomena artificially to the dataset.
Initially, the dataset was separated into two groups by
considering the two classes in the response variable. Then
the observations were selected randomly with replacement
from the two groups to create a new dataset which the
size of it was equal to the numbers of observations in
the original dataset. When considering the response
variable in a dataset with two classes, one of them could be
considered as ‘success’(positive) class and the other one as
‘failure’ (negative) class according to the problem domain.
The number of observations that should select randomly
from each group was decided as below.
1002 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
(7)
(8)
pi : probability of success
N : total number of observation in original dataset
ns : number of observations select from ‘success’ group
nf : number of observations select from ‘failure’ group
Equation 7 was used to determine the number of
observations that should be chosen from ‘success’ group
while the equation 8 was used to decide the number of
observations that should be taken from ‘failure’ group.
Then a new dataset was formed by combining the randomly
chosen two sets of observations which contained the same
number of variables and same number of observations
as in original dataset. Since the class distribution was
not random in newly created dataset, random class
distribution was introduced by shuffling the data row wise.
When shuffling the data, the order of observations (entire
row) was changed but the values relevant to an observation
(values within a row) were not affected.
Then the 10-fold cross validation method was used.
The above mentioned classification techniques were
trained by using training set and the class outcomes
were predicted by using testing set separately for each
technique. When repetition was done for 10 times, each
time all the classification techniques were trained by using
same training set and tested on same testing set. Then
by comparing actual class outcomes and predicted class
outcomes, the evaluation measures were obtained for
each classification technique separately. Finally, the best
classification algorithm was determined by analyzing the
achieved results for evaluation measures.
This whole process was repeated for 9 times by varying the
probability of success (pi) from 0.1 to 0.9.
2) Procedure in Overcoming Class Imbalance Aspect: In
this aspect, the classification techniques were compared
under different sampling approaches to overcome the class
imbalance phenomena. The main focus was to balance the
training set. Since, the class distribution was not random
in most of the original datasets, the randomness of class
distribution was introduced by shuffling the dataset row
wise. As in previous aspect, in here also the order of
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

observations (entire row) was changed while the values Table 3. Classification techniques which
relevant to an observation (values within a row) remain have highest and lowest performance in
unchanged. Resampling Aspect
Precision
Then as same as in previous aspect, the 10-fold cross
Dataset Highest Lowest
validation method was applied. The approaches
Gamma Tel. RF LR
undersampling, oversampling and hybrid (mix of
Wilt RF & Boosting NB, LDA (0.1)
undersampling and oversampling) were applied to Bank Mark. RF NB, Bagging (0.1) LR (0.2)
balance the training set derived from shuffled dataset. The German Cr. Boosting & RF LR
same modified dataset was used under each technique (0.1,0.2,0.3)
Tic-Tac-Toe Boosting NB
separately. Finally the best classification technique was LR (0.1)
determined according to three sampling approaches. The Accuracy
only difference in here was that the use of balanced training Dataset Highest Lowest

set for the training process of classification techniques. Gamma Tel. RF LR

Wilt RF & Boosting NB
Bank Mark. RF NB
F. Special Considerations German Cr. Boosting & RF LR
(0.1, 0.2)

All the classification techniques were tested by considering Tic-Tac-Toe Boosting

LR (0.1)
NB

the five original datasets separately. LDA is one of the F-measure

statistical classification technique used in this study. The Dataset Highest Lowest
theory of LDA suggests that it is suitable when the predictor Gamma Tel. RF LR

variables are in interval scale. Therefore LDA was applied Wilt RF & Boosting NB

only for the datasets which contained continuous variables Bank Mark. RF NB
German Cr. Boosting & RF LR
except the response variable (Wilt and Gamma Telescope (0.1, 0.2,0.3)
datasets). Since bagging and boosting are ensemble based Tic-Tac-Toe Boosting, NB
LR ( 0.9)
methods (not stand alone methods), the classification tree
G-mean
was used as the single classifier. When considering on Dataset Highest Lowest
machine learning techniques, some of them use several Gamma Tel. RF LR, LDA (0.1)
parameters. The default values suggested by the functions Wilt RF & Boosting NB, LDA (0.1)

in R is used for those techniques. Also feature selection and Bank Mark. RF NB, Bagging (0.1),LR (0.2)

parameter tuning are important when training a machine German Cr. Boosting & RF (0.1, 0.2,
0,3, 0.6, 0.7, 0.8, 0.9)
LR

learning algorithm. Even in logistic regression, after fitting Tic-Tac-Toe Boosting NB

the model the best model can be selected using stepwise LR (0.1, 0.9)
ROC AUC
methods. These scenarios were ignored in order to treat
Dataset Highest Lowest
both equally for classical statistical and machine learning Gamma Tel. RF LR
techniques. That means when training the considered Wilt RF &Boosting NB, LDA (0.1)
techniques, all the predictor variables were used as well Bank Mark. RF NB, Bagging (0.1), LR (0.2)
as no parameter tuning was done to the machine learning German Cr. Boosting & RF (0.1, 0.2, LR
0,3, 0.6, 0.7, 0.8, 0.9)
techniques. Tic-Tac-Toe Boosting NB
LR (0.1, 0.9)

III. RESULTS Up arrow ( ) - represents higher values

A. Resampling Aspect Down arrow ( ) - represents lower values

The results obtained for resampling aspect are summarized
in Table 3.
In Gamma Telescope and Bank Marketing datasets,
the technique RF has the highest performance for all
probabilities of success. The techniques, RF and boosting
have almost the same highest performance for all
success probabilities in Wilt dataset. In German Credit

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1003
PROCEEDINGS ISBN number: 978-955-0301-41-6

dataset, both RF and boosting have almost the same G-mean

highest performance for all success probabilities except Dataset Highest Lowest
the probabilities mentioned in brackets and for those R U O H R U O H
probabilities RF deviates from boosting as shown by up-
Gamma Tel. RF LR
arrow and down-arrow. In Tic-Tac-Toe dataset, LR has
Wilt BT BT LR LR LDA NB NB NB
the highest performance for the probabilities mentioned
BA BA
in brackets and for all the other probabilities boosting has
the highest performance. Bank Mark. NB RF BT BT LR NB RF NB

German Cre. BT RF NB NB LR
According to the lowest measures, the techniques which
are in italic form has the lowest measures at relevant BA
probabilities mentioned in brackets and for all the other Tic-Tac-Toe BT LR
probabilities, mainly mentioned technique (which are in ROC AUC
non-italic form) has the lowest measures in each dataset. Dataset Highest Lowest

B. Overcoming Class Imbalance Aspect R U O H R U O H

Gamma Tel. RF LR
The results obtained for overcoming class imbalance
Wilt BT BT LR LR LDA NB NB NB
aspect are summarized in Table 4.
BA BA
Table 4. Classification techniques which Bank Mark. RF RF BT BT LR NB NB NB
have highest & lowest performance in German Cre. BT RF NB NB LR
Overcoming Class Imbalance Aspect NB

Precision Tic-Tac-Toe BT LR
Dataset Highest Lowest
The used symbols in these tables are as follow.
R U O H R U O H
Gamma Tel. RF RF BT RF LR
R – Before applying sampling approaches (Real case)
Wilt BT LDA LDA LDA LDA NB NB NB

Bank Mark. NB RF BT BT LR NB RF NB
U – After applying undersampling approach
German Cre. BT RF BA NB LR
O – After applying oversampling approach
Tic-Tac-Toe BT RF BT BT LR
H – After applying hybrid approach
BT
Accuracy According to the Table 4, the techniques RF or boosting
Dataset Highest Lowest have highest performances for the most of the cases in
R U O H R U O H overcoming class imbalance aspect.
Gamma Tel. RF RF
BT BT RF RF
Wilt
BT BT
NB
IV. DISCUSSION &
Bank Mark.
German Cre.
RF
RF
BT
NB
RF
BT
RF
NB
NB BA
LR
BA BA
CONCLUSIONS
Tic-Tac-Toe BT LR
F-measure
When considering the resampling aspect, resampling was
Dataset Highest Lowest
done to the whole dataset by changing the probability of
R U O H R U O H success from 0.1 to 0.9. Though the dataset was balanced
when success probability was equal to 0.5, there was no
Gamma Tel. RF LR
guarantee about the balance of training set. Here, for
BA BT RF RF NB
Wilt
BT BT
Gamma Telescope and Bank Marketing datasets the
RF BT RF RF NB BA BA BA
technique RF had the highest performance for all success
Bank Mark.
BT
probabilities. When considering all success probabilities,
German Cre. RF NB RF BT LR
two techniques (RF and boosting) had the same highest
BT NB
performance for the Wilt dataset. In some datasets
Tic-Tac-Toe BT LR (German Credit and Tic-Tac-Toe), there was no one such
method which could be considered as the best method for

1004 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
all success probabilities. The obtained results suggested
that either RF and boosting separately or both of them as
the best methods for all success probabilities.
Also it was noticed that in resampling aspect, the G-mean
and ROC AUC values, were approximately similar to
each other in all five scenarios respectively. According to
those two evaluation measures, almost all classification
techniques had higher performance at middle probabilities
(eg: 0.5) than the extreme probabilities (eg: 0.1 and 0.9).
In overcoming class imbalance aspect, the main focus
was balancing only the training set not the whole dataset.
Not like in previous aspect, there was an assurance that
the classification techniques were trained by using a
balanced training set. The sampling approaches such as
undersampling, oversampling and hybrid approach (mix
of both undersampling and oversampling) were used to
balance the training set. The performance of classification
techniques were analyzed individually before and after
applying the sampling approaches to make the training
set balanced under each dataset separately. The techniques
RF and boosting had the highest performance for many
cases in this aspect. Since the sampling approach which
maximize the performance of classification techniques
was depended on domain, the best sampling approach
couldn’t specifically mentioned.
Also majority of evaluation measures (precision, G-mean,
ROC AUC) suggested that the performance of most
classification techniques were higher after applying the
sampling approaches. The evaluation measures, accuracy
and F-measure had lower values after applying sampling
approaches for most of the classification techniques. The
performance of positive class may low due to issues such
as loss of valuable information and over fitting in sampling
approaches. Since the accuracy and F-measure are relied
only on positive class, they might provide lower results
after applying the sampling approaches.
Though LDA and LR are classical statistical techniques,
they were sensible to sampling approaches. However
for most of the cases, the performance of LR also had
been improved after applying sampling approaches but
sometimes it was considerably very low when compared
to the performance of other techniques. That was probably
due to the characteristic of domains. Not only the machine
learning techniques, but also statistical techniques had
higher performance by using the balanced training set.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Not only after training classification algorithms by using
the balanced training set but also in resampling aspect, RF
and boosting performed well for most of the cases. It might
happened probably due to the sampling structure in their
classification algorithm. When considering the obtained
results, it is difficult to make a general conclusion about
the best classification algorithm due to the dependency
in performance of classification technique on problem
domain.
When the classification classes are not equally represented
(imbalanced) in dataset approximately, the accuracy
of machine learning techniques are biased towards the
majority class. That means if higher proportion (about
90% or above) of data belongs to one particular class
(called majority class), then classification algorithm tends
to predict the majority class often. Hence achieves highest
accuracy while the decisions can be misled. Since the class
imbalance was considered in both aspects, relying only on
accuracy was not appropriate. That was the reason for using
several evaluation measures such as precision, F-measure,
G-mean and ROC AUC. When considering the precision,
it is based on high performance of only one class. According
to equation 5 and 6, F-measure is combination of precision
and sensitivity while G-mean is combination of sensitivity
and specificity. ROC AUC gives the area under the curve
which represents sensitivity – (1-specificity) pairs. Though
the F-measure was suggested as an alternative for the class
imbalance, it was relying only on success (positive) class.
The evaluation measures, G-mean and ROC AUC can
handle the class imbalance scenario well as they consider
the performances of both classes at once. Therefore, use
of several evaluation measures was important as some of
them can handle the issues like class imbalance and can
provide much confident conclusions.
REFERENCES
Aha D (1991). UCI Machine Learning Repository: Tic-
Tac-Toe Endgame Data Set. [online] Archive.ics.uci.edu.
Available at: https://archive.ics.uci.edu/ml/datasets/Tic-
Tac-Toe+Endgame.
Bock R and Savicky P (2007). UCI Machine Learning
Repository: MAGIC Gamma Telescope Data Set. [online]
Archive.ics.uci.edu. Available at:https://archive.ics.uci.
edu/ml/datasets/MAGIC+Ga
mma+Telescope.
10TH INTERNATIONAL RESEARCH CONFERENCE 1005
PROCEEDINGS ISBN number: 978-955-0301-41-6

Breiman L (2001). Statistical Modeling: The Two Cultures.
Statistical Science, 16(3), pp.199-231.
Brownlee J (2015). 8 Tactics to Combat Imbalanced
Classes in Your Machine Learning Dataset. [online]
Machine Learning Mastery. Available at: http://
machinelearningmastery.com/tacti cs-to-combat-
imbalanced-classes-in-your-machine-learningdata set/.
Chawla NV (2005). Data Mining for Imbalanced Datasets:
An Overview. In: O. Maimon and L. Rokach, ed., Data
Mining and Knowledge Discovery Handbook, 1st ed.
Springer US, pp.853-867.
Hastie T, Tibshirani R and Friedman J (2008). The
Elements of Statistical Learning. 2nd ed. California:
Springer Series in Statistics.
Hofmann H (1994). UCI Machine Learning Repository:
Statlog (German Credit Data) Data Set. [online] Archive.
ics.uci.edu. Available at: https://archive.ics.uci.edu/ml/
datasets/Statlog+(G erman+Credit+Data).
Johnson B, Tateishi R and Hoan N (2013). A hybrid
pansharpening approach and multiscale object-based
image analysis for mapping diseased pine and oak trees.
International Journal of Remote Sensing, [online] 34(20),
pp.6969-6982. Available at: https://archive.ics.uci.edu/ml/
datasets/Wilt.
Kohavi R and Provost F (1998). Glossary of Terms.
Machine Learning, 30, pp.271-274.
Moro S, Cortez P and Rita P (2014). A data-driven approach
to predict the success of bank telemarketing. Decision
Support Systems, [online] 62, pp.22-31. Availableat:
https://archive.ics. uci.edu/ml/datasets/Bank+Marketing.
Schneider J (1997). A Locally Weighted Learning Tutorial
using Vizier 1.0. [online] Available at: https://www.ri.cmu.
edu/pub_ files/pub2/schneider_jeff_2000_1/schneider_
jeff_2000_1.pdf.
ACKNOWLEDGMENT
We wish to express our sincere gratitude to the colleagues
at Department of Statistics, University of Colombo and
University of Colombo School of Computing who helped
us in numerous ways to make this research a success.

KVU Perera is currently working as a Temporary Instructor at Department of Statistics, Faculty of

Science, University of Colombo. (BSc. in Statistics with Computer Science special degree with 2nd
class – Upper division)

DrSD Viswakula is a Senior Lecturer (Grade II) at Department of Statistics, Faculty of Science,
University of Colombo. His research interests are Biostatistics, Computational Statistics and
Probabilistic Modeling.

1006 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

PREDICTIVE CUM ADAPTIVE SYSTEMS

DEVELOPMENT METHODOLOGY
FOR HYDRO GIS TOOL DEVELOPMENT
RMM Pradeep1#, NTS Wijesekara 2
1
Faculty of Computing General Sir John Kothelawala Defence University, Sri Lanka
2
University of Moratuwa ,Sri Lanka
# pradeep@army.lk

Abstract - In system development methodologies, main
consideration has being changed from processes to users
since waterfall development. As the business logic became
less complex due to familiarity; Extreme Programming
methodologies like Agile and Scrum become popular
among the programmers. Nevertheless, the engineering
processes like hydrology modelling, which are still
evolving, remain in the same complex. As well historically
established hydrology calculations, which are base to
evolving models, are remaining complex. When model
development, hydrologists have to identify and sequencing
those established calculations that best suited to the model
scenarios. Then as the final step, they have to calibrate
and validate the model, which takes considerable time
and effort, before apply to decision making. Hence, once
the models are crated, those cannot be changed very
easily whilst the automation. As well as the programmers
has to pay a considerable attention to the get the 100%
accurate result. However, the most users of engineering
applications are novice and required more user centric
tools same as other users. Hence, there is a difficulty to
develop such systems following either predictive (focus
on process) or adaptive (focus on users) methodologies
which presently available.
The present work’s main objective is to identify and
calibrate the most suited combination of methodologies
to development of a HydroGIS (Hydrological Geographic
Information System) tool, which should accurately
automate the complex hydrology process in GIS
environment whilst satisfying the user requirements.
Whilst developing the tool, it carries out two parallel
developments (1) Automation of engineering process and
(2) Achieving the user-friendliness. Whilst automating
the processes, a comprehensive devotion was paid to
calibrate the accuracy of the calculations. When achieving
the maximum user-friendliness, a repetitive developing
prototype was used. Once the both parallels come to
the accepted level, it amalgamates the prototype with
engineering processes. Then integration and system
testing were carried out before releasing the final product.
The developed tool named GIS2MUSCLE.
The HydroGIS tool which developed, demonstrates 100%
accuracy in hydrological and GIS calculations whilst
92% user-friendliness in tool operation. Presently this
calibrated methodology, which follows a process centric
development to gain user centric tool (PcD.UcT), is being
verified with six software development projects.
Keywords - Engineering System Development
Methodology, HydroGIS tool, Predictive cum Adaptive
I. INTRODUCTION
A. System Development Methodologies
System Development Methodology refers to the approach
in implementing the system development life cycle (SDLC)
phases in the practical software development. Planning,
analysis, design and implementation are the four basic
steps of SDLC. In the planning phase, it identifies and

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1007
PROCEEDINGS
reasoning why the system should build and determining
how it will go about building it. Through studying
the present systems and identifying the problems &
opportunities, team conceptualize the new system whilst
the analysis phase. In the design phase, team finalise how
the system operate by the means of functional and non-
functional requirements. At the implementation stage,
team builds the system. (Dennis et al., 2009)
These methodologies have being classified according to
different perspectives. Following are few examples (1) Plan-
driven/traditional or heavyweight and Agile/ lightweight
(2) Predictive and Adaptive (3) Process centred and Data
centred.(Dennis et al., 2009; Fowler, 2001; Picek, 2009).
The present study considers the predictive (scope of the
project can be expressed accurately) and adaptive (scope
and requirements are difficult to clearly expressed early in
the SDLC) perspectives.
B. Engineering Applications
Engineering is a matured profession with experiences.
Comprehensive planning, designing and drawing are
inherent characteristics of the engineering process. Due to
the failure-costs are immensely unbearable, the engineers
invest considerable time and resources in these initial
stages.
Once the software engineering profession emerge in
1960’s, it was a subset of engineering profession which
women’s job (Meyer, 2013). In that time, planning, design,
implementing and maintaining of the hardware was the
main computing work whilst software development
was a painting work. Since then, the initial software
development methodologies such as waterfall and
parallel development models emerge with the increasing
of the software utilization in the computing. The phases
in these development methodologies are based on same
characteristics of general engineering process such as
comprehensive study, design and then development.
Nevertheless, with the vast distribution of the users
among the different knowledge level of computing,
the user interface of the software required to be more
user-friendly. As well due to the uncertainty of the
requirements, the development of the software using
predictive methodologies becomes hectic work. Then as
a solution, adaptive development methodologies such as
extreme programming, agile development versions got the
popularity.
1008 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
The automation of the engineering applications also
subjected to same user requirement variations. The
situations become worst, as the non-technical decision
makers are prone to use the engineering applications when
decision-making. As the engineering calculations become
more complex with the development of profession, it
needs more time and resource in the analysing, design and
development phases. Therefore, engineering applications
required predictive as well as adaptive development
methods simultaneous which is not applicable.
C. Hydro GIS Tool Development Methodology
The present work considers a HydroGIS tool, hydrological
engineering application for urban decision-making.
The decision-making users in the local authorities
need to suggest and stress the public when urban land
modifications for the purpose of urban flood prevention.
Then a series of hydrological calculations need to be
performed to evaluate the affect of modifications on the
flood generation and select a suitable preventive option. As
it is impractical to employ a hydrologist in the process, the
requirement is to automate the hydrological calculations,
which can perform by local decision makers. Then the
automation process should allow users to incorporate
land modifications in spatial format, perform hydrological
calculation, display output in spatial format and opt a
preventive option (Pradeep and Wijesekara, 2012).
Considering hydrologists are the most naturalistic
environmental modellers. The hydrology models
developed for natural phenomena are widely accepted and
more established than other environmental models (Sui
and Maggio, 1999). For the Sri Lankan context, most of the
hydrology calculations based on the Ponrajah’s guidelines
published in 1980s, which have being successfully utilised
since then. Such basic calculations, which verified, not
only through data but also through time, have become
the base for most of todays developing hydrology models.
(Chemjong and Wijesekara, 2017; Dahanayake and
Wijesekera, 2017; Keerthirathne and Wijesekara, 2017;
Sakthivadivel et al., 1997; Thakuri and Wijesekara, 2017).
Whilst hydrology model development, the most suitable
set of established hydrological calculations is selected.
Then the sequence of the calculations is being arraigned
to get the required results. Once complete the calculation
process sequencing, start the model calibration with
available data. Then calibrated model subjected to validate
with the real time data. Automation of the process is
allowable only after this process, which urged considerable
 ISBN number: 978-955-0301-41-6
time and resource. Whilst the automation process,
the results of each, intermediate to final hydrological
calculations steps, has to be verified and confirmed to
the standards and norms. Hence, automation becomes a
process centric development, which need to carryout with
close relation with hydrologists.
As well, hydrological models are based on the geographical
distribution of features such as slope, soil, landcover,
ground water levels, soil moisture etc. Therefore when
perform the hydrological models it required to manipulate
the geographic information too. For this purpose, GIS,
a specific tool developed to manipulate the geographic
information, become a supporting tool in hydrological
model calculations (Maidment, 1992). Therefore,
to receive the accurate result in automation of the
hydrological models it need to develop coding not only for
hydrological calculations but also for geographical layer
manipulations. As the hydrological systems are based on
time and GIS based on the space the automation process
get more complex. Then developers need to develop codes
with the consultation of the GIS professionals too.
Apart from the complex process, the potential users of the
HydroGIS tool are varying from very few highly technical
hydrologists to large number of non-technical decision
makers like government officials in local government
authorities. Importantly the decision makers required
the accurate result with minimum interaction with the
tools. Specially, decision makers need to make attribute
modifications and viewing the intermediate results, calls
trial-and-error process, to arrive the concluding
D. The Problem
Then the final problem is, how a HydroGIS tool that process
and user-friendliness both are having same importance,
can be developed accurately and satisfactorily.
E. Objective
Therefore, the main objective of the present work is to
identify and calibrate the most suited combination of
methodologies to development of a HydroGIS tool, which
accurately automate the complex hydrology process in
GIS environment whilst satisfying the non-technical user
requirements.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
II. LITERATURE SURVEY
A. System Development Methodologies
A large number of system development methodologies are
available as shown in the table 1.0. All these methodologies
based on the phases of system development cycle;
planning, analysis, Design and implementation. As
well, the waterfall method can be considered as the first
most development methodology that extensively used in
1970s. As an alternative in 1980s, to satisfy the unsatisfied
requirements in development the prototype evolved.
Finally, when the “Requirements of users” evolved in
2000s the Agile methodologies were introduced (Avison
and Fitzgerald, 2006). Therefore the methodologies
shown in the Table 1.0 can be describe as the versions of
waterfall and prototype developments.
When study the development methodologies, it
can identify the attention has being changed from
process automation to user engineering with the time.
Nevertheless, it can be considered, waterfall methodology
provides the foundation to all the methodologies. Even
today, the popular methods like scrum show the features
of waterfall and prototype methodologies. However,
attention to the user requirement should have a limit.
It has identified the user engineering may have a risk of
excessive software development with gold-plating or
bells-and-whistles or mission/feature/scope/requirement
creeping. Then the repercussions may be negatively effect
on system development project schedule, quality and cost
(Shmueli and Ronen, 2017). Hence, a balance between
process automation and user requirement satisfaction
should to be maintained.
Therefore, the developers need to a make decisions
to select the best methodology to their development
process. When deciding the satisfaction of promises
(expected positive impacts) and practises (essential steps
of methodology) of the selected methodology to the
developers’ requirement is important. There are number
or evaluation mechanism such as Cost-benefit analysis,
Scoring evaluation, Feasibility study, Value Analysis
and Multi- Objective Multi-Criteria methods and so
on are available (Mohagheghi, 2008). But the present
work evaluate the methodologies based on much simple
approach, analysis the ease of automating the process as
well as user requirements. Then the suitability of the each
approach for the HydroGIS tool development is shown in
the Table 1.0.
10TH INTERNATIONAL RESEARCH CONFERENCE 1009
PROCEEDINGS ISBN number: 978-955-0301-41-6

Table 1.0 System Development Methodologies

15 Fea- P The larger project develop-
Srl Meth1 On2 Suitability for Hydro- ture-driven ment process which is hav-
GIS tool develop- ing less complex processes.
1 Waterfall P Best suited for process ment Need a complete staff of
automation, difficulty in- developers
corporate user require- 16 Model-driv- P Model oriented automa-
ments later en engi- tions, user requirements are
2 Prototyping U Best for user friendly neering difficult to handle
development, effects on
workflows 17 Crystal U Develop based on the devel-
3 P Shorter waterfall steps, Methods opers capability over user
Iterative and
but difficult in incorpo- Methodol- requirement
incremental
ration of user require- ogy
ments 18 Joint Appli- D Developer centric, not much
4 Spiral P The methodology is cation De- describe as a development
velopment methodology, but as a tool
based on risk reduction
19 Adaptive U Based on User requirement
5 Rapid appli- U Time boxing approach software satisfaction
cation devel- based on fulfilling the develop-
opment requirements ment
6 Extreme pro- U Directly automate the 20 Open D Developer centric, not much
gramming user requirements source soft- describe as a development
7 V-Model P Based on the testing, ware devel- methodology, but as a tool
difficult in user require- opment
ment facilitation 21 Microsoft D Developer centric, not much
8 Scrum U Rather small projects Solutions describe as a development
and scrum based Framework methodology, but as a tool
9 Cleanroom U Iterations with box struc- 22 Agile Devel- U A adaptive approach based
ture opment on the user requirement
10 Dynamic U Time boxing approach satisfaction
systems de- based on fulfilling the 23 Scrum U A adaptive approach based
velopment requirements on the user requirement
11 Rational Uni- U A complicated system satisfaction, development of
fied Process with iterative develop- agile methodology
ment 24 Kanban P Provide the most required
12 Lean soft- U Automate the minimum (Just-in- part of the software to the
ware devel- requirements with users time) correct time with quality.
opment Based on pre defined user
requirement.
13 Test-driven U Development based on
development the testing Meth1: Methodology On2: Focus on
U: User P: Process D: Developer
14 Be- U User behaviour centric
Source: Author after Despa, 2014; Jamsheer, 2016;Leiet
haviour-driv-
al., 2015
en develop-
ment

1010 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

B. HydroGIS Tool Requirements cognitive walkthrough, Formal usability inspections,

Pluralistic walkthroughs and etc; are available. The main
aim of the usability evaluation is find the problems in user-
Pradeep and Wijesekara (2012) study has found that
friendliness and fix those before release the final product.
a requirement to develop a HydroGIS tool to manage
(“Introduction to User ­Centered Design,” 2017; Nielsen
the urban flash flood. Hence urban flash flood is a
and Molich, 1990; Nielsen, 2012, 1994)
repercussion of urban land allotment modifications, the
tool need to identify the affect of the modification on the
stormwater generation. Further tool should allow the users
to arrive dynamic engineering solution to manage excess
stormwater. They have identified three process modules to
carryout the entire task, (1) Incorporation of land parcel
modifications (2) Calculated storm generations pre and
post scenarios (3) incorporation of detention pit. This tool
urged to have a user centric map based interface with on-
screen capability in data input and dynamically modify
the attributes/land parcel modifications. As well due to
the manipulations are done in rather smaller urban land
extents, the accuracy of the results becomes sensitive. The
Important factor is, the potential users of the tool are non-
hydrological land managers.

C. User Centric Design

As the users feel User Interface is the System, development

of user interface for tools are very important. User-Centred
Design (UCD) is the process of designing a tool, from the
perspective of how it will be understood and used by a
human user which formulated around 1980s. To learn the
software users have to adapt their attitudes and behaviours.
But when UCD, the software designed to assist potential
users’ existing attitudes, and behaviours. To achieve this,
it places users at the centre of the design process from the
stages of planning and designing the system requirements
to implementing and testing the product. This result an
efficient, satisfying, and user-friendly tool. (Abras et al.,
2004; Baek et al., 2008)
The user experience design (UXD) and usability are the
other two terms with the UCD. UXD needs to understand
the users through a research. The research includes user
observations, interviews, and different techniques to
capture the users’ emotions, motivations, and underlying
concepts and beliefs. Then this knowledge will be used
to develop user interfaces which align and support user
behaviour. Usability is a measure of the interactive user
experience associated with a user interface. It is a evaluation
of user-friendliness which is easy-to-learn, and easy-to-
use capability. The usability measures evaluating the users
capability of use of the tool without any assistance. For the
evaluation there are number of methods such as heuristic,
Figure 1.0 Overall Development Methodology
III. METHODOLOGY
A. Overall Methodology
The present work methodology flowchart is shown in
the Figure 1.0. The work starts with the comprehensive
literature survey to identify existing system development
methodologies. Then, it identifies the requirements of the
tool such as user needs, data needs, process / calculation
needs and technology needs. After that, it categorises those
requirements in to different features. In the next step, it
analyses the system development methodology over the
tools requirements to identify which methodology to be
used. If it could be able to identify the methodology, then
follow the same. Else, identify the suitable methodologies
to automate the categorised feature by feature. Once all of
the different features automated, integrate them together
and carryout integration, system and acceptance testing
till reach the satisfactory level, before release the tool.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1011
PROCEEDINGS
Once the tool release, progressively scan through the
methodologies practises to develop the tool and identify
successive path to development of an accurate and user
friendly HydroGIS tool.
B. User Requirement Automation with Prototype
When user requirement automation, the tool underwent
three kinds of user evaluations; software adequacy,
formative and summative evaluations as shown in the
Figure 2.0. Through the requirement analysis, it identified
the basic functionalities of the tool then a prototype was
developed. The developing prototypes were evaluated
with the users through questionnaire. The objectives of
the evaluations are shown in the Table 2.0. This software
adequacy questionnaire evaluates the useability of the
initial functions of the tool. Then with the user commented
modifications second prototype was developed and again
it was tested two times, which until satisfy the users. The
acceptable prototype then integrated with the calculations
modules. The final system was subjected to perform a
summative evaluation. In the each evaluations, the tool
was modified based on the view of usability. These process
and the evaluations were align with the National Research
Council (2007) guidelines of human-system integration in
system development process.
Figure 2.0 Prototype Development Process
1012 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Table 2: Basic Requirements gathered
from the questionnaires
Basic Question target to acquire the
Requirement Users’ satisfaction on
Assess the Installation of the tool, Start the
achievement tool , Layer selection, Modify the
of Objectives selected layers and attributes, Up-
date modifications , Modification
of onscreen map and attributes,
Do the modification in all four
layers, Printed outputs generation,
Secure the operation
Assess the Ease of navigation, Zooming and
usability of the panning, Scale facility, Permit the
developed GUI user to navigate while keeping
with GIS con- track of current reference frame,
cepts / usage Provide tools for capturing, editing,
and printing maps, Map in a larger
percentage of the screen area
Assess the User centred design , Visual
General Prin- clarity, Consistency, Explicitness,
ciples of GUI Appropriate functionality, Flexibility
development and control, Error prevention and
correction, Compatibility / Porta-
bility, User guidance and support,
Informative feedback
Source : (Pradeep and Wijesekara, 2015)
C. Process development with waterfall development
In the calculation process automation, it identified the
required hydrology models and calculation sequences.
As well, a specific attention paid to automate the dynamic
calculation modules. For that, it incorporates Rational
Method to calculate composite runoff coefficient. As well,
the concept of inflow hydrograph attenuation is being
used to determine the detention storage size (Pradeep and
Wijesekara, 2012). These two complex processes run based
on the inputs from not only users or results of intermediate
calculations but also GIS manipulation outcomes. Then the
entire process run through manually and all the final and
intermediate results were recorded in excel sheets. This
process can be considered as the study the manual system
and identify the bottle-necks in exiting system which is a
step of waterfall development. Then automation started
and each step evaluated against the results in the Excel
sheets for accuracy. This become more difficult work, as
once it found the errors in the manual system, then the
 ISBN number: 978-955-0301-41-6
manual system has to perform from the very beginning
and as well, the automation had to reengineered. Then a
considerable time spent to evaluate the manual process
then start the automation. A module level evaluation of
the software output against the manual results carried out
throughout the automation. The development of foresaid
excel sheet was a different research of University of
Moratuwa.(Wijesinghe and Wijesekera, 2010)
D. Integrating the User Centric Tool with Process
Centric Algorithm
Once the user friendliness reach to a acceptable level, 70%,
and need to integrate the process for further evaluations,
it decided to stop prototype modification. But the
process automation carryout until reach 100% against
the recoded set of results. Once it satisfied the process
accuracy condition, then integrated the user interfaces
with the process codes which have developed using same
languages on same platforms. Whilst the integration, the
user modification requests received at the 70% satisfaction
level, were incorporated and went to the final summative
evaluation.
IV. RESULTS & DISCUSSION
A. The development model - Process Centric
Development to User Centric Tool (PCD.UCT)
Figure 3.0 PcD.UcT Development Model
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Once the satisfied tool resulted, it evaluates the software
development methodology, which it has actually practised.
Then it could be able to identify a hybrid development of
waterfall and prototype development. The user centric
tool development shows the characteristics of prototype
development whilst the process centric algorithm
development process shows the characteristics of
waterfall development. The practised and proved software
development methodology was named Process Centric
Development to User Centric Tool (PcD.UcT) model and
shown in the Figure 3.0.
B. User Centric Tool
The HydroGIS tool which developed through the described
process shows a 92% user acceptance as shown in the Table
3.0. User evaluation 1,2 and 3 were performed during the
user centric tool development and final evaluation was
done at the optimization evaluation after the integration
of prototype to process algorithms.
C. Process Centric Development
The result has shown the followed methodology could
able to provide 100% of the process. It has tested 80 test
cases whilst the algorithm development and optimization
evaluation. Result of five samples out of 80 carried out at
the optimization evaluation are shown in the Table 4.0.
10TH INTERNATIONAL RESEARCH CONFERENCE 1013
PROCEEDINGS ISBN number: 978-955-0301-41-6

Table 3: Evaluations Result

7 On-Screen Operational Capability
Srl Main Consid- User Friendliness Eval- Ease of navi- 70% 75% 78% 86%
erations and uations gation
attributes 1st 2nd 3rd 4nd Scale facility 69% 75% 78% 90%
1 Continuity in Operation Zooming and 68% 75% 78% 89%
Process Liberty 70% 70% 78% 90% panning
of other GIS Modification 67% 72% 75% 93%
functions of onscreen
2 Error handling & Accuracy Confirmation map and attri-
Error prevention 70% 72% 78% 98% butes
and correction Visual clarity 65% 68% 73% 98%
3 GIS software version Compatibility Map display 5% 65% 67% 89%
Compatibility / 70% 78% 78% 99% size
Portability 8 Tested and Verified Results
4 Information Security Modification 65% 70% 75% 85%
Spatial Data Se- 0% 0% 35% 89% in all required
curity layers
5 Non-GIS User Operation Capability Average 50% 63% 70% 92%
Update modifica- 80% 80% 80% 97% Source : (Pradeep and Wijesekara, 2015)
tions
Flexibility and 70% 70% 79% 88%
control Table 4: Evaluation accuracy result
Appropriate 68% 70% 77% 90% Test Runoff coefficient Runoff coeffi- Accuracy
functionality case (manual method) cient (Tool)
No Before After Before After Before After
Modify the se- 68% 77% 77% 97%
Mod: mod: mod: mod: mod: mod:
lected layers and
1 0.548 0.695 0.548 0.695 100% 100%
attributes
2 0.984 0.816 0.984 0.816 100% 100%
Explicitness 67% 75% 76% 97% 3 0.327 0.364 0.327 0.364 100% 100%
4 0.228 0.389 0.228 0.389 100% 100%
6 Easy operation Capability
5 0.200 0.376 0.200 0.376 100% 100%
User centered 65% 65% 73% 88%
design D. Effect on Software Project Management
Consistency 55% 69% 71% 89%
According to the practised development model (Figure
Informative feed- 5% 68% 70% 99% 3.0) it has identified three different phases (1) Requirement
back clarification (2) Development and (3) Evaluation and
User guidance 3% 40% 49% 90% modification. The recoded average weeks (1 week is equal
and support to 14 hours effort of a programmer) taken for each phases
are showing in the Table 5.0.
Printed outputs 3% 30% 43% 95%
Provide tools for 3% 30% 43% 91%
capturing, edit-
ing, and printing
maps

1014 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
The Requirement Clarification of Prototype was taken
much shorter time than the Process Development, as the
requirement clarification is in the Process development
was a hydrological model development activity.
In the development, it took average equal time for both
development processes. However due to the requirement of
following standards such as user-friendliness development
and usability guidelines, prototype taken much longer
time.
Whilst the evaluation, Prototype was taken longer time
not only as it need to meet the users and get the feedback,
but also need to satisfy users requirements which arises at
each evaluation.
Table 5: Average Weeks Taken to Complete
Phases
Phase Prototype Development Process Development
Requirement 2 5
Development 3 2
Evaluation 4 1
Considering the all three phases are having equal weights
towards the completion of the development described
in this sub section, Figure 4.0 shows the comparison of
phases of two different developments over the time.
Figure 4.0 Development process
completion comparison
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
E. Discussion
The present work attempted to develop the codes for
process whilst developing the model. Then the work
had trouble in time-taken reengineering with re-coding
when a model has to correct. Hence, it has realised that
requirement of tough patient, until the model development
finish to start the coding of the processes / calculations.
Then without wasting time for model calibration, the work
started the user interface development and evaluations for
usability. Nevertheless, before finalise the prototype, sub
process module out of 3 process modules were available
to automate.
Then the present work started automates the process
modules parallel to the prototype development that
realised the methodology is an innovative use of available
methodologies.
Even the model developers confirm the completion of
the modules’ accuracy, the work did not develop the user
interfaces with the process automation, but carried out the
automation of processes in software modules which can
fix to prototype at anytime.
The work realised that, in each user evaluation a new user
requirements are creating which are out of the scope of
development. Due to this scope creeping requirements,
more than 70% user friendliness was considered as an
acceptable level to conclude the prototype development.
Once integrating the user interface to the developed
process the final summative evaluation or the optimization
evaluation was carried out. Then, as the tool is providing
the actual results and satisfied the observations, which
users made at the 2nd formative evaluation, user
acceptance raised to 92%.
The processes and prototype were integrated once the
acceptable levels of each reached. Then the expected errors
were with only coding problems in integration. As it uses
standards, publicly expressed variable descriptions and
module level algorithm development, it was became easy
task.
The time taken information in the Figure 4 shows late
completion of the prototype development after the
process development. Due to the two different natures of
developments, it is not reasonable to match each other.
However, Figure 4 shows the development phases of two
developments start one after other. Hence this parallel
10TH INTERNATIONAL RESEARCH CONFERENCE 1015
PROCEEDINGS
development methodology utilise the developer for coding
continuously which results early Software Development
Project completion.
V. CONCLUSION
The present work practised a hybrid development
methodology, UcT.PcD, a combination of waterfall,
prototype - repetitive development methodologies. This
was a two parallel phased development methodology,
(1) automate the engineering process through waterfall
development and (2) develop the user interfaces using
prototype - repetitive development methodologies. Once
the accuracy reach to maximum and user-friendliness
reach to a accepted levels, the two set of codes were
integrated. The final evaluation carried out to confirm the
success.
Then the present work demonstrate the capability
of combine the approaches in system development
methodologies that are following entirely different
sequence to achieve the required outcome if the developers
understating the desired outcomes.
REFERENCES
Abras, C., Maloney-Krichmar, D., Preece, J., 2004. User-
centered design and evaluation of a next generation
fixed-split ergonomic keyboard. Bainbridge, W. Encycl.
Human-Computer Interact. Thousand Oaks Sage Publ.
37, 445–456. doi:10.3233/WOR-2010-1109
Avison, D., Fitzgerald, G., 2006. Methodologies for
Developing Information Systems  : A Historical Perspective.
Past Futur. Inf. Syst. 1976–2006 Beyond,Springer US 27–
38. doi:10.1007/978-0-387-34732-5_3
Baek, E.-O., Cagiltay, K., Boling, E., Frick, T.,
2008. User-centered design and development.
Handb. Res. Educ. Commun. Technol. 1, 660–668.
doi:10.1145/1273961.1273973
Chemjong, H., Wijesekara, N.T.S., 2017. Evaluation
of Irrigation Water Issue Practice for Better Water
Management at Rajangana Reservoir, Sri Lanka, in:
UMCSAWM Water Conference. UNESCO Madanjeet
Singh Centre for South Asia Water Management,
Moratuwa, pp. 41–51.
1016 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Dahanayake, A.C., Wijesekera, N.T.S., 2017. Watersheds
with an Emphasis on Multi User Concepts – Demonstrating
an Application at a Watershed in Dampe, Sri Lanka, in:
UMCSAWM Water Conference. UNESCO Madanjeet
Singh Centre for South Asia Water Management
(UMCSAWM), Moratuwa, pp. 21–28.
Dennis, A., Wixom, B.H., Tegarden, D., Wiley, J., Tavares,
K., Wolfman-robichaud, S., Dodds, K., 2009. Systems
Analysis and Design with UML Version 2.0, 3rd Editio.
ed. John Wiley & Sons, Inc.
Despa, M.L., 2014. Comparative study on software
development methodologies. Database Syst. J. V, 37–56.
Fowler, M., 2001. The New Methodology. Wuhan Univ. J.
Nat. Sci. 6, 12–24.
Introduction to User ­Centered Design [WWW
Document], 2017. . Usability First. URL http://www.
usabilityfirst.com/about-usability/introduction-to-user-
centered-design/ (accessed 3.3.17).
Jamsheer, K., 2016. 12 Best Software Development
Methodologies with Pros &amp; Cons. Softw. Dev. India.
Keerthirathne, W.H., Wijesekara, N.T.S., 2017.
Determination of a Design Rainfall Pattern by Comparing
with its Effect on Streamflow on Greater Colombo
Watershed in Sri Lanka, in: UMCSAWM Water
Conference. UNESCO Madanjeet Singh Centre for South
Asia Water Management, Moratuwa, pp. 35–40.
Lei, H., Ganjeizadeh, F., Jayachandran, P.K., Ozcan, P., 2015.
A statistical analysis of the effects of Scrum and Kanban
on software development projects. Robot. Comput. Integr.
Manuf. 43, 59–67. doi:10.1016/j.rcim.2015.12.001
Maidment, D.R., 1992. Grid-based Computation
of Runoff  : A Preliminary Assessment. Hydrologic
Engieering Center, Davis, California.
Meyer, B., 2013. The origin of “Software engineering”
[WWW Document]. URL https://bertrandmeyer.
com/2013/04/04/the-origin-of-software-engineering/
(accessed 3.1.17).
Mohagheghi, P., 2008. Evaluating Software Development
Methodologies Based on their Practices and Promises,
in: New Trends in Software Methodologies, Tools and
Techniques. pp. 14–35. doi:10.3233/978-1-58603-916-5-
14
 ISBN number: 978-955-0301-41-6
National Research Council, 2007. Human-System
Integration in the System Development Process: A New
Look. The National Acedemies Press, Washington D.C.
Nielsen, J., 2012. Usability 101 : Introduction to Usability
[WWW Document]. Nielsen Norman Gr. URL https://
www.nngroup.com/articles/usability-101-introduction-
to-usability/ (accessed 4.17.17).
Nielsen, J., 1994. Usability inspection methods, in:
Conference Companion on Human Factors in Computing
Systems - CHI ’94. pp. 413–414. doi:10.1145/259963.260531
Nielsen, J., Molich, R., 1990. Heuristic Evaluation of
user interfaces, in: CHI ’90 Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems. pp.
249–256. doi:10.1145/97243.97281
Picek, R., 2009. Suitability of Modern Software
Development Methodologies for Model Driven
Development. J. Inf. Organ. Sci. 33, 285–295.
Pradeep, R.M.M., Wijesekara, N.T.S., 2015. Modification
of User Friendliness in to a HydroGIS Tool, in: 8th
Intrernational Research Conference 2015. Sir John
Kothalawala Defence University, Kandawala, Ratmalana.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Pradeep, R.M.M., Wijesekara, N.T.S., 2012. GIS Tool to
Mitigate Urban Flash Floods Through Capturing Land
Allotment Modifications and Incorporating Conceptual
Detention Pit, in: ITRU 2012. Universiry of Moratuwa,
Moratuwa.
Sakthivadivel, R., Fernando, N., Brewer, J.D., 1997.
Rehabilitation Planning for Small Tanks in Cascades: A
Methodology Based on Rapid Assessment. Colombo.
Shmueli, O., Ronen, B., 2017. Excessive software
development: Practices and penalties. Int. J. Proj. Manag.
35, 13–27. doi:10.1016/j.ijproman.2016.10.002
Sui, D.Z., Maggio, R.C., 1999. Integrating GIS with
hydrological modeling: Practices, problems, and prospects.
Comput. Environ. Urban Syst. 23, 33–51. doi:10.1016/
S0198-9715(98)00052-0
Thakuri, P.S., Wijesekara, N.T.S., 2017. Climate Change
Impacts and Adaptation Measures for Pahala Divul
Wewa, Anuradhapura, Sri Lanka, in: UMCSAWM Water
Conference. UNESCO Madanjeet Singh Centre for South
Asia Water Management, Moratuwa, pp. 59–62.
Wijesinghe, W.M.D., Wijesekera, N.T.S., 2010. Reserch
Monograph - Conceptual Detention Storage for
Stormwater Management. Moratuwa.
10TH INTERNATIONAL RESEARCH CONFERENCE 1017
PROCEEDINGS
MACHINE LEARNING
OPTIMIZATION FOR COLOUR IMAGE
RECONSTRUCTION FROM
THERMAL/INFRARED IMAGES
WTVL Gunarathne1 and DMR Kulasekara1
1
General Sir John Kotelawala Defence University, Sri Lanka
# vlgunarathne@gmail.com
Abstract - Machine Learning has generated a tremendous
interest in research and development under the umbrella
of Artificial Intelligence. It was a field that evolved from
pattern recognition and computational learning in
Artificial Intelligence. Machine Learning algorithms are
capable of identifying how to accomplish certain tasks
by generalizing from real world examples. In comparison
to manual programming, this is often feasible and cost-
effective. In this context, this paper focuses on the use of
these optimized machine learning algorithms in order to
reconstruct colour images from thermal/infrared images.
With the capability of these machine learning algorithms
to identify patterns in existing data sets, such an algorithm
can be used to reconstruct a colour images based on the
features that are given to the algorithm. Most night time
photography uses thermal imagery because of its capability
of capturing thermal radiation from the human body.
So these data can be used to reconstruct a colour image
based in feature recognition form the thermal image. In
conclusively it is our intention to use the power of machine
learning techniques to build a system that can generate
colour images by analysing the small amount of details that
are present in thermal images.
Keywords - Thermal Imagery, Facial Image reconstruction,
Machine Learning
1018 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
I. INTRODUCTION
Machine learning field and machine learning algorithms
have gained immense popularity due to its capability of
improving itself through experience[1]. It has become
one of the most growing technical fields being nurtured
by a mix of computer science, data science, statistics and
artificial intelligence. Machine learning algorithms are
capable of identifying patterns and how to accomplish
certain tasks by generalizing from the real world examples.
It all depends on the amount of data that is being fed to
the algorithms. More problems can be tackled as more
and more data is available. This is one of the interesting
adoptive properties of machine learning algorithms. This
is often feasible and cost effective. Machine learning is used
in Web search, spam filters, stock trading, recommender
systems, fraud detection, ad placement, credit scoring,
drug design, and many other applications.[2] Under the
umbrella of Artificial Intelligence, machine learning has
flourished as the method of developing systems such as
speech recognition, natural language processing, robot
control, computer vision and many other things. Training
a system with examples of desired input-output behaviour
has become more feasible than to program and algorithm
to anticipate a desired output based on all possible inputs.
Machine learning algorithms can be used to tackle
learning problems. Learning problems can be defined as
an improvement of the threshold of performance when
executing a task over and over again and there by learning
from it. For example, a machine learning algorithm
 ISBN number: 978-955-0301-41-6
trained to differentiate between male and female faces
which will flag a given image, whether the person is a
male or a female, will improve its performance over time
when encountering more and more data sets and through
processes such as active learning.[3]
Figure 1 Machine Learning Mechanism
The focus of this paper is the use of these machine learning
techniques to understand the minute details of the face
image data from thermal images of human faces and using
the machine learning optimization to build a colour image
model of the face. When the thermal image is submitted to
the algorithm, it will analyse for the features of the image
and based on the existing knowledge base that it has, it
will reconstruct a face based on the data. Thermograms
are the images that are generated by detecting radiation in
the long infrared range of the electromagnetic spectrum.
A thermal infrared camera is being used for this purpose.
Simply said, the camera can identify the heat signatures
emitted by any object that is inside its view finder. After
pre-processing the thermogram using image processing
techniques, certain physiological features can be extracted
based on the blood perfusion data. Blood perfusion data
are based on the way that blood vessels are distributed
under the skin. Distribution of blood vessels is unique
for each individual and a set of extracted minutiae points
from a blood perfusion data of a human face would be
unique for that face. So these unique features will assist
in reconstructing the facial data and finally output a facial
image of a person.
II. EXISTING SYSTEMS
Biometrics is a Greek origin word meaning measure of life.
It means that the measures or the metrics that we take of
the human beings. Facial data analysis is a part of this field.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Biometrics use various physical characteristics to match
with the data in the database.
Figure 2 Biometric Uses
Abundantly used physical biometrics are fingerprints,
facial features, hand geometry and eye features etc. Among
these many biometric measures, facial detections are a
significant trait that is being used in the modern world.
One significant importance is that no physical interaction
is needed for facial images to be used. The system can
capture the facial image of a person while a person is
walking through a door or while he is standing, there by
revoking the requirement for the individual to touch a
panel, keep a fingerprint of any other physical interaction.
In comparison of visual images with thermal images, the
visual images have a higher preference in many aspects.
Following can be considered as a few set of advantages of
visual images.
• Location and extraction of features can be done easily.
• Optical cameras are less expensive.
But factors like illuminations and viewing directions
have a noteworthy impact on these visual images[4].
One solution to overcome these limitations is using
3D data obtained from 3D vision devices. They are less
independent on illuminations but the downside to it is the
massive cost. So a simple, feasible solution to this would
be infrared or thermal images which do not depend on
illuminations and other factors. Colour which comprises
the longest wavelength of the visual spectrum is red.
“Infra” which means below in Latin suggests that Infrared
is below Red.
The wavelengths of different IR spectrums are shown in
the following table[5].
10TH INTERNATIONAL RESEARCH CONFERENCE 1019
PROCEEDINGS ISBN number: 978-955-0301-41-6

Spectrum Wavelength Range III. TECHNOLOGIES USED

Near-Infrared (NIR) 0.7-1.0 µm There can be various scenarios where visual images
Short-wave Infrared (SWIR) 1-3 µm become un-resourceful and are incapable of meeting out
Mid-wave Infrared (MWIR) 3-5 µm certain requirements. Specially for night time surveillance
imagery, acquisition of visible light images is a bit difficult
Thermal Infrared (TIR) 8-14 µm due to the lack of visible light. This is the place where
Figure 3 Wavelength changes for different thermal imagery plays a major role. Thermal imaging
IR Spectrums which uses mid-wave or long-wave infrared radiations
Following mentioned advantages can be obtained from which are naturally emitted by the human body can be
using thermal imaging. utilized to overcome the shortcomings of visual images.
• The cost of IR cameras has been reduced recently due The drawback here is that thermal images lack the wealth
to the implementation of CCD (Charged Coupled of information that is available in visual images and hence
Device) technology[6]. it is difficult to identify features like facial cues in a thermal
• Different lighting conditions doesn’t affect thermal image to recognize a person. So this section focusses on
imaging, even complete darkness. the technologies that are used to develop a system that
uses machine learning algorithms to study features that
• IR face images contains details about the basic are common to visual and thermal images from a set of
anatomical information about the face. data and try to reconstruct a visual image of a face from
• Facial detections, localizations and segmentations are
a given thermal image. Matching thermal face imagery to
much easier. the existing databases of facial images therefore requires
the development of across modality face recognition
• Has much better accuracy because it uses facial algorithms and methods.
temperature variations.
In relevance to current thermal imaging systems, they Due to the gap caused by the wavelength difference
can be classified into three main parts namely, image between visible radiation and thermal radiation, thermal
acquisition, image processing and classification. to visual face reconstruction can be challenging. The
existing systems[4], [7], [8] sectioned detailed about
the usage of NIR images vs visual images for facial
recognition. Details about the NIR-to-visible[9] face
recognition and the SWIR-to-visible[10] face recognition
was also briefly described. Since both NIR and SWIR
require active illumination it is not very practical to night
time surveillances.

The key point in solving the thermal-to-visible imaging

drawback is the development of an algorithm or transform
space that well-correlates the thermal and visible face
signatures.

Figure 4 Block diagram of a present system Figure 5 Illustration of visible and thermal
resresentationsof the physical face

1020 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Here our main focus is the matching of thermal images
to visual images so that the machine learning algorithm
can be trained to analyse and detect patterns between the
two image domains. We convert this face identification
problem of matching thermal images to visible images as
a multi-modal face recognition problem. Before tackling
this, various pre-processing techniques such as self-
quotient[11] images, difference-of-Gaussian[12] filtering
and various feature transforms are being used.
A.PRE-PROCESSING
Due to the different signatures that thermal and visual
images contain, pre-processing is essential in finding a
solution to the problem. Pre-processing consists of two
main stages. Namely,
• Thermal image normalization, and
• Local variation reduction for thermal and
visible imagery
Simple median filtering prior to image normalization is
used to remove the dead pixels within the thermal images.
Step 01: Normalizing thermal signatures
Normalize the thermal signatures by its mean and
standard deviation to reduce the temperature offset and
statistical variation across thermal images.
Figure 6 Original Figure 7 Normalized
Image Image
Step 02: Adjusts the thermal and visible imagery for local
variations.
For visible imagery, illumination primarily induces the
local variations, whereas for the thermal imagery, the
varying heat distribution within the face produces the
local variations. Self-quotient image and difference of
Gaussian
Filtering is commonly applied to reduce illumination
variations in visible face imagery.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
B.FEATURE TRANSFORMS
Computer vision applications is crucially dependant on
the selection of good features. There are many feature
descriptors available to facilitate face recognition. Local
Binary Patterns (LBP) is a popular texture and a successful
feature descriptor under local illumination variances.
LBP are compact and can be easily compared by various
histogram metrics. The most popular extension is multi-
scale LBP (MSLBP) descriptor.
The above figure shows two pre-processed images along
with the original images in the visible and thermal
domains. As can be observed, SQI emphasizes the edge
information in the thermal imagery while DOG filtering
blurs the visible imagery. The Gabor wavelets are also
effective face descriptors which capture global shape
information centred on a pixel. The convolution of multiple
Gaussian-like kernels at different scales and orientations
captures information insensitive to expression variation
and blur at a pixel’s location. We consider all these features
for thermal-to-visible face recognition. We also compare
the results of using raw intensity values as a feature as in
some previous works.
IV. DISCUSSION
Performing basic pre-processing consisting of dead pixel
removal, affinely warping the face by four fiducial points
(two eyes, nose tip, mouth), cropping to face regions,
10TH INTERNATIONAL RESEARCH CONFERENCE 1021
PROCEEDINGS ISBN number: 978-955-0301-41-6

and resizing to 80x100 pixels gives a much better way to Gang Hua, Face Recognition by Discriminative
generate visual images from thermal images. All these Orthogonal Rank-one Tensor Decomposition. INTECH
techniques gave a proper insight as to how the task of Open Access Publisher, 2008.
thermal image reconstruction should be accomplished.
X. Chen, P. J. Flynn, and K. W. Bowyer, “IR and visible
light face recognition,” Comput. Vis. Image Underst., vol.
V. CONCLUSION 99, no. 3, pp. 332–358, Sep. 2005.
In this study, I have investigated the thermal-to-visual
X. Zou, J. Kittler, and K. Messer, “Face Recognition Using
image reconstruction. The novel combination of pre-
Active Near-IR Illumination.,” in BMVC, 2005.
processing, feature transforming and PLS-DA recognition
framework gives promising results with a high accuracy.
Z. Lei and S. Z. Li, “Coupled spectral regression for
A data set of these visual images and their corresponding
matching heterogeneous faces,” in Computer Vision and
thermal images can be fed into a machine learning
Pattern Recognition, 2009. CVPR 2009. IEEE Conference
algorithm which can be trained to analyse the patterns
on, 2009, pp. 1123–1128.
between those image pairs. Thus, that algorithm will be
able to generate a new visual image for a thermal image
B. F. Klare and A. K. Jain, “Heterogeneous Face Recognition
that is provided. In conclusion these techniques can be
Using Kernel Prototype Similarities,” IEEE Trans. Pattern
used to implement a machine learning algorithm that can
Anal. Mach. Intell., vol. 35, no. 6, pp. 1410–1422, Jun. 2013.
reconstruct a visual image based on the thermal image
that is being provided to the algorithm.
H. Wang, S. Z. Li, and Y. Wang, “Face recognition under
varying lighting conditions using self quotient image,”
REFERENCES
in Automatic Face and Gesture Recognition, 2004.
Proceedings. Sixth IEEE International Conference on,
E. Horvitz and D. Mulligan, “Data, privacy, and the greater
2004, pp. 819–824.
good,” Science, vol. 349, no. 6245, pp. 253–255, 2015.
X. Tan and B. Triggs, “Enhanced local texture feature sets
P. Domingos, “A few useful things to know about machine
for face recognition under difficult lighting conditions,”
learning,” Commun. ACM, vol. 55, no. 10, pp. 78–87, 2012.
in International Workshop on Analysis and Modeling of
Faces and Gestures, 2007, pp. 168–182.
N. Lavesson, Blekinge tekniska högskola, and Department
of Systems and Software Engineering, “Evaluation and
analysis of supervised learning algorithms and classifiers,”
ACKNOWLEDGEMENT
Blekinge Institute of Technology, Karlskrona, 2006.
I would like to express my great appreciation to Prof.
S. G. Kong, J. Heo, B. R. Abidi, J. Paik, and M. A.
AS Karunananda for his valuable and constructive
Abidi, “Recent advances in visual and infrared face
suggestions during the planning and development of
recognition—a review,” Comput. Vis. Image Underst., vol.
this research work. His willingness to give his time so
97, no. 1, pp. 103–135, Jan. 2005.
generously is very much appreciated. I would also like to
thank Mr. DMR Kulasekara for his advice and guidance
Anne Cleary, Face Recognition without Identification.
that helped me to make this possible. I would also like to
INTECH Open Access Publisher, 2011.
thank Mr. A Gunasekera and staff of the Department of
Computer Science at KDU, for their valuable and precious
time, which is generously and highly admired.

1022 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
A MACHINE LEARNING APPROACH TO
CLASSIFY SINHALA SONGS BASED ON
USER RATINGS
HMT Paranagama1#, MKA Ariyaratne1 and
SCMDS Sirisuriya1
Faculty of Computing,
1
General Sir John Kotelawala Defence University, Ratmalana.
# tharindrapara@gmail.com
Abstract - In the music industry there is a need to analyse
the significant features that distinguish highly rated songs
from lower rated ones. Then an artist can test their music
tracks to check whether it will gain potential popularity,
before mass production and if the rating is lower they
can focus on the significant features present in popular
tracks. Our study address this by developing a machine
learning approach to classify music tracks based on user
ratings. There were many research performed in the area
of music genre classification, music recommendation
using vanilla neural networks, recurrent neural networks
and convolutional neural networks. The research mainly
focuses on the classification of Sinhala songs. Our
dataset is consisting of 11,000 Sinhala music tracks each
having several attributes. From each track we extract 3
meaningful features. For the feature extraction process we
used a python library. The output has three distinct classes
that specify the user rating. A Multi-layer neural network
was implemented. 500 training epochs with 60 neurons in
each hidden layer were used. Initially, with 3031 training
tracks and 1299 testing tracks we achieved an accuracy
of 86%. With this, we conclude that the development of
a multilayer neural network to automate the process of
determining the rating for a song is in a successful stage
compared with the existing approaches.
Keywords - Artificial Neural Networks, Classification,
Clustering, Feature Extraction
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
I. INTRODUCTION
Music has a great influence on humankind from reducing
stress to enhancing cognitive power of human beings,
While conveying strong messages as well as feelings to
the audience in a unique manner. According to forecasts
done by PriceWaterhouse Coopers, the global music
industry is supposed to generate 47.7 billion U.S. dollars
in revenue by 2020.(Statista,1 May 2017) In this type of
promising market musicians can easily benefit given
that they can satisfy audience expectations. In the context
of Sinhala music it has the influence of many cultures
such as the British, Indian and Buddhist.(wikipedia,1 May
2017) With these influences the music developed in Sri
Lanka has greatly evolved and as a result of this evolution
today we experience a more unique production of music.
In the past the music produced was more of classical type
while modern music incorporates a highly sophisticated
level of beat patterns as well as a variety of instruments
which generate powerful and complex tones. With the
improvement in technology the music industry has
taken a huge step forward with more convenient ways to
produce, manipulate as well as generate music for the ease
of musicians. With these developments in this industry
it has made it extremely difficult to analyze and interpret
user preferences thus leading to many musicians to fail
in the industry while a few highly dominate the industry
through their gifted talent. Through research done by a
collaboration of the following universities and institutions,
10TH INTERNATIONAL RESEARCH CONFERENCE 1023
PROCEEDINGS
McGill University, the University of Cambridge, Rutgers
University, City University of New York, and the Stanford
Graduate School of Business, They have found out that any
music produced can be effectively categorized with the use
of three significant features which are Arousal (intensity
and energy in music), Valence(Spectrum of emotions in
music) and Depth(Intellect and sophistication in music).
(Digital Music News,1 May 2017) There are many
research which mention different types of characteristics
of music that determine user preferences. But there are no
proper tools that have been implemented to successfully
provide a solution to this problem of determining the
potential user rating for a given song based on the
underline characteristics. As a rating determines the
success of the music track inability to effectively determine
the rating will be a barrier to determine the success of a
music track. Without any awareness of the success of the
product a musician( especially for amateurs) may not be
willing to enter the industry or may leave the industry,
because of this the objective of our study was to develop
a multilayer artificial neural network to successfully
determine the rating of a given Sinhala song. In this
process of determining the rating we also optimize the
algorithm(s) and source code to provide a more efficient
and accurate result.
II. METHODOLOGY
There are extensive research done in the area of music genre
classification and sound classification. In a research done
by Aaqib Saeed, used machine learning to classify urban
sounds to its relevant classes. He used a dataset which
consist of various 8732 urban sounds each consisting
of duration of 4s separated in to 10 folders from www.
freesound.org from which first three folders were used for
training and testing purposes. The dataset has 10 classes
which are air conditioner, car horn, children playing, dog
bark, drilling, engine idling, gun shots, jack hammer, siren
and street music The features that were extracted with the
use of a python library known as Librosa and these features
were melspectrogram, mel-frequency cepstral coefficients,
chroma-stft, spectral contrast and tonnetz .
The extracted features and labels were fed to a multi-layer
neural network on the training data. And the testing data
were later fed after completion of the training. And here
only the features of the testing set were provided to the
neural network and the network was supposed to predict/
determine the relevant label pertaining to the given
features based on prior learning experience. The accuracy
1024 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
is calculated based on the number of correct predictions
over the total testing set.
They have used Gradient Descent Optimizer which is
an optimisation algorithm used to reduce the cost by a
fraction of the learning rate during the training process.
Thus optimizing for higher accuracy. Tensorflow has been
used in the development of the neural network and the
cost function they use is the cross entropy function.
This system does not provide an acceptable level of
accuracy. It only has an accuracy of approximately 12%
and further the computational cost of this implementation
is significantly high(Saeed,1 May 2017).
In a research done by Carlos N. Silla Jr., Alessandro
L. Koerich, and Celso A. A. Kaestner which was to
automatically classify music into genres.In this system
they use multiple feature vectors and pattern recognition
techniques. A set of binary classifiers which produces the
final result/genre based on a merge of multiple attributes
were used to obtain result which were afterward merged
to determine the genre. Most popular machine learning
algorithms such as k-nearest neighbors, support vector
machines, Naive-Bayes are some of them which they used.
An important feature identified through their research
was the most significant features used for the classification
task depended on the origin of the music signal(Carlos N.
,Koerich and Silla Jr,2008).
In a music genre classification research done by Michael
Haggblade, Yang Hong and Kenny Kao. In their research
they have discussed about three classification algorithms
which are k-means,multi-class SVM, and K-Nearest
Neighbor(KNN).further they have chosen four
classes(jazz, classical, metal, pop) of which one will be
assigned with the output. They have used Mel Frequency
Cepstral Coefficients(MFCCs) as their feature. They
have also extended their system to cluster music into
genres based on image features. They have performed
the extension using k-means algorithm for clustering and
Fourier-Mellin 2D transform to extract features. They
have noted an important fact which is that the accuracy
decreases if the number of classes exceeds 4. They
stated that K-means algorithm faced a huge difficulty in
distinguishing between classical and jazz. thus leading
to 36% misclassification. They have stated that when
considering overall accuracies the KNN and K-means
displayed similar accuracies of 80% while the SVM gave
them a 87% accuracy and finally the neural network giving
them a 96% accuracy(Kenny Kao , Michael Haggblade and
Yang Hong ).
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

In a research done by Sam Clark, Danny Park and Adrien

Guerard, they have used eight summary features, growing
neural gas (an algorithm for reducing the computational
costs when using a complex neural network) and a
neural network for music genre classification. They have
considered five classification classes as rap, reggae, classical,
and country. Their hypothesis was that the growing neural
gas would lead to an improved classification accuracy. This
hypothesis was proven to be true with the receipt of nearly
14% increase in training accuracy and a 3% increase in test
Figure 2: Architecture of the multi-layer neural
accuracy (Adrien Guerard, Danny Park and Sam Clark,
(2012)) . tempo, MFCC and harmonic element. The hidden
layers do the processing and outputs the relevant
In a research done by Cory McKay using neural networks label(excellent,moderate,poor).
for music genre classification, he has developed a
system which can automatically classify MIDI files into C. Approach
hierarchically organized parent genres and sub genres.
For his classification he has used primarily twenty original 1. Users: Any artist or music listener who would like to
features and achieved an accuracy of 85% for parent genres
determine the rating of a given song.
and 65% for sub-genres(Cory McKay).
There also exist research done by Bo Shao, Dingding 2. Inputs: The 3 extracted features from the audio file.
Wang, Tao Li, Mitsunori Ogihara in the area of music
recommendation. They state that collaborative filtering 3. Process:There will be a web page which will provide
and content based recommendations are the most popular the user with the facility to upload their audio file and
and recommended methods for music recommendation. then the feature extraction method will run in the
But the aforementioned researches see some significant back end and display the extracted feature vector on
disadvantages of these methods which leads to an ineffective the web page’s output area.
recommendations been made. These disadvantages are,
the collaborative method will be effective only if we use a Then the user can input this feature values to the neural
large dataset that consist of user history of user activity on network available on another page to determine the
the web. And the content based method lacks the ability to predicted rating for the input vector.
identify user preferences and interests. To overcome these
limitation they propose a novel approach which takes into D. Implemantation
account content features and user access patterns(Bo Shao,
Dingding Wang and Tao Li, (2009)). 1. Creation of the datasets and preprocessing

B. Conceptual Design The dataset contains 11,000 mp3 music files from a variety
The following diagram describes the entire process of of musicians, genres and centuries. This diversified dataset
determining the rating for a given song. is a good representation of the entire population of Sinhala
songs.

First we would extract the features from each of the

audio file and save it to a CSV file. So that it would be
computationally less expensive compared to processing
the audio files for feature extraction during program
execution.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1025
PROCEEDINGS
Figure 5: Clustering the features using K-Means
After the CSV file is created we would use a K -Means
clustering to determine the labels of the dataset. For this
we use the scikit-learn library which provides an easy
implementation of this algorithm.
Then this dataset would be split into 3 portions as follows
70% as training ,15% as testing and 15% as validation.
Finally the training data and testing feature vectors should
be reshaped to be separately fed to the neural network
during training and testing.
And also the labels should be converted in to one hot
vectors because we cannot do comparisons with nominal
values and also using one hot encoding facilitates ease of
processing. Here we use an encoding which is an array of
the length equal to the number of classes in our problem
and we give an activation to the relevant position in the
array depending on the value of the label. for example if the
rating poor has a label of 0 this can be represented as a one
hot vector as follows,
1026 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
[1,0,0] where the position of zero is active to denote the
label. When training and testing the neural network we
provide it with the features and one-hot encodes instead
of labels.
Now since we have constructed our inputs in a suitable
manner for feeding we have to now move on to
development of the neural network.
2.Neural network design
The neural network has an input layer, 2 hidden layers and
an output layer.The hyper parameters of the network are as
follows,10 neurons in each hidden layer a learning rate of
0.003 and 600 training epochs.
As we are using tensorflow for the construction of the
neural network we use 2 placeholders which is a data type
that allows us to feed data to the neural network during
run time. These are very useful as we can sequentially feed
each feature vector along with its corresponding one hot
encode to these two separate placeholders.
when the input(feature vector) is fed to the first layer
of the neural network its multiplied (using matrix
multiplication )with the weight matrix which is of size
[#features,#neuronsInHiddenLayer1] and provides the
ability for each feature to be multiplied by each neuron
.The values of the weight matrix are initialised with
random values at the start of the training process. After
we multiply the weight matrix with the feature vector we
add biases to each output from the matrix multiplication.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

The biases matrix is of size [#neuronsInHiddenLayer1].
now this net input is sent through an activation function
which will out put a vector of size [#features] which will
be the input to the next layer ,the most popular activation
function is Sigmoid and we have also used it because it
gives an output within the range of 0&1.the Next layer also
operates in the same way and progresses to the output layer
in which it goes though a Softmax activation function to
produce a probability as an output.
determine the correct predictions by comparing the actual
label with the predicted label on the testing features by
the network. Then we cast this boolean value to a float to
determine the number of correct predictions as a value.
then we display the accuracy as the number of correct
predictions over the total number of test samples.

E. Evaluation

After developing the model we can use the model to make

new predictions for that we need to save the desired model
and restore it when we need to do prediction through it
and the prediction can be made on a new record as follows

Figure 10: Construction of the multi-layer neural

Now we define the cost function as cross entropy which is

also a popular choice among cost functions.

III. RESULTS
Figure 12: Mathematical formula for cross A.Performance Indices

where q(x) is the predicted output and p(x) is the expected the main performance indices we consider is the accuracy
output. To convert this function into a minimisation and the cost. We managed to achieve an accuracy of 50%
function we put a minus in front. and a smoothly decreasing curve. The following diagrams
illustrate the above explanation. It should be also noted
And then we should optimise our cost function during that the following results were obtained after a number of
the training process for this we use the Gradient Descent training runs by changing hyper parameters to find this
Optimization algorithm. It should be noted that during temporary satisfiable solution.
the training of the network over the 600 epochs we update
the weight an biases after each input is fed to the network
via a dictionary of feature values and labels. This updating
is done by a fraction of the learning rate for this we use the
optimization algorithm and we specify the cross entropy
to be minimized. Thus allowing the network to converge
to a particular function during the learning process.

Then we also need calculate the accuracy so for that we

need to know the number of correct predictions. So we

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1027
PROCEEDINGS
Fig. 18: Changes in performance with respective
IV. SUMMARY & CONCLUSION
According to the extensive research I performed I
observed that the currently there is no solution for the
aforementioned problem in the context of Sinhala music.
So we have not only built a solution for the system but we
have optimized the solution and the code in various ways
such as using clustering in determining the labels, and
using prestored data for feeding input.
ACKNOWLEDGEMENT
its with great honor that I would like to thank
MR.Chandrasekara who was grateful enough to offer me
a large repository of diverse Sinhala mp3 song collection.
Further I would also take this opportunity to thank my
supervisor Ms.M.K.A.Ariyaratne for her valuable insights
and guiding on this project and also Dr.T.G.I.Fernando for
showing me this problem for which I was able to come up
with a decent solution.
REFERENCES
Statista. 2017. • Global music industry revenue 2016 |
Statistic. [ONLINE] Available at: https://www.statista.
com/statistics/259979/global-music-industry-revenue/.
[Accessed 01 July 2017].
Music of Sri Lanka - Wikipedia. 2017. Music of Sri Lanka
- Wikipedia. [ONLINE] Available at: https://en.wikipedia.
org/wiki/Music_of_Sri_Lanka. [Accessed 01 July 2017].
1028 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Digital Music News. 2017. All Music Can Be Categorized
by Just Three Attributes. [ONLINE] Available at: https://
www.digitalmusicnews.com/2016/05/17/music-genres-
three-attributes/. [Accessed 01 July 2017].
Aaqib Saeed. 2017. Urban Sound Classification, Part 1.
[ONLINE] Available at: https://aqibsaeed.github.io/2016-
09-03-urban-sound-classification-part-1/. [Accessed 01
July 2017].
A Machine Learning Approach to Automatic Music
Genre Classification - Kent Academic Repository. 2017. A
Machine Learning Approach to Automatic Music Genre
Classification - Kent Academic Repository. [ONLINE]
Available at: https://kar.kent.ac.uk/24022/. [Accessed 01
July 2017].
Haggblade, M., Hong, Y. and Kao, K. (2017). Music
Genre Classification. [ebook] p.http://cs229.stanford.
edu. Available at: http://cs229.stanford.edu/proj2011/
HaggbladeHongKao-MusicGenreClassification.pdf
[Accessed 1 Jul. 2017].
Clark, S., Park, D. and Guerard, A. (2012). Music Genre
Classification Using Machine Learning Techniques.
[ebook] Available at: https://www.cs.swarthmore.
edu/~meeden/cs81/s12/papers/AdrienDannySamPaper.
pdf [Accessed 1 Jul. 2017].
Bo Shao, Ogihara, M., Dingding Wang and Tao Li (2009).
Music Recommendation Based on Acoustic Features
and User Access Patterns. IEEE Transactions on Audio,
Speech, and Language Processing, 17(8), pp.1602-1611.
McKay, C. (2017). Using Neural Networks For Musical
Genre Classification. [ebook] Montreal: McGill University.
Available at: http://www.music.mcgill.ca/~cmckay/
papers/musictech/GenreClassification1.pdf [Accessed 1
Jul. 2017].
 ISBN number: 978-955-0301-41-6
RESEARCH DIRECTIONS FOR
NETWORK BASED VIDEO
STREAMING WITH EMPHASIS FOR
LIVE SCREEN MIRRORING
SC Gajanayaka1, DMR Kulasekera1 and ADAI Gunasekera1
1
General Sir John Kotelawala Defence University, Sri Lanka
# sinhalokaya@gmail.com
Abstract - Recent advancements in the field of embedded
systems and network speeds in general – particularly
the realization of concepts of image processing, live
video streaming and Internet of Things have redefined
the technology world, as we knew years ago. In terms
of video streaming, the rapid increase of speeds of
networks around the world have enabled fast and efficient
streaming of live content. Embedded systems have made
it simpler to build portable devices which can stream
live video via a network. This review analyses these
developments in detail, outlining the potential left for
development. In doing so, the paper looks at techniques
and technologies that have been used for streaming videos
over networks. Special emphasis is placed on using the
available technologies for screen mirroring and live video
broadcasting. Development in these concerns will assist in
developing lightweight wireless systems for TV stations to
personal video uploading in social media.
Keywords - wireless video transmission, embedded
systems, screen mirroring, live collaboration tools
I.INTRODUCTION
Video transmission over networks has been a contended
topic of research ever since the birth of the Internet.
Though many technologies have developed over time, one
significant shortcoming which marred the propagation
and the popularisation of network video streaming is the
practical issue of low bandwidth in networks. In the current
sphere, however, most countries have achieved significant
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
network bandwidths with the development of ADSL,
3G and 4G networks around their countries. In such a
context, live video streaming has become a crucial area of
research as it now involves the added concern of managing
millions of connected devices at once. It is predicted that
there will be over 25 billion connected devices by the year
2020 (“The Evolution of Mobile Technologies,” 2014).
As of 2017, major corporations are focussing strongly
on live video streaming. As such, “streaming” is moving
into the mainstream. This focus has resulted in apps
like Periscope (“Periscope - Live Video - Android Apps
on Google Play,” 2017), YouTube and Facebook Live
which have made it easy for anyone with a smartphone
to start streaming. They have normalized the idea of live
streaming video and as a result millions of people are
now used to this type of content. That growing ubiquity
represents an opportunity for other businesses. Live video
can be a powerful tool for marketing, sales, training,
entertainment, or media, as it is a compelling technology
for everyone, in effectively transmitting a message of
pages in a few seconds (“Youtube Live vs Facebook Live
compared to Online Video Platforms,” 2017).
Though this is the status quo, the birth of all these
technologies run years behind. This paper thus discusses
the development of wireless (online) video streaming
technologies and engages in a comparative analysis of the
same.
Another important aspect discussed in this paper is the
parallel development of live collaboration software tools.
10TH INTERNATIONAL RESEARCH CONFERENCE 1029
PROCEEDINGS
These tools have been the artefacts to show how much
technology has triumphed in its application. Online
collaboration tools have become instrumental in many
areas including business and education.
Though technology has developed at an alarming rate, one
underlying factor of limitation when it comes to streaming
live video is the bandwidth problem. To overcome this,
either smaller packets of data containing the video data
should be transmitted or more effective algorithms must
be used. These however should be done in such a way to
ensure that the quality of the video is not diminished.
This paper therefore looks at the above-mentioned aspects
and how these technologies can be integrated to deliver
higher quality streaming experiences to the end user. The
essential hypothesis is to see if such integration is possible
within technologies available as of today and to assess
potential issues associated with the same. The objective
of this research is therefore, to identify the applicability of
network based video streaming for live screen mirroring.
In general, the paper analyses related aspects, as mentioned
above, in detail.
The paper is divided into 3 main sections – literature
review – which highlights the research and the journey
thus far in above technologies, discussion – analysis of the
contents of the literature review, and finally the conclusion
– concluding remarks as to what developments may be
undertaken by future researchers.
II. LITERATURE REVIEW
A. Video Transmission Techniques
Firstly, literature clearly shows that there has been a
constant interest to research about the various video
transmission techniques which can be deployed for
network based video transmission. Networks function
in layers in which a data packet goes through a stack of
different layers which perform distinct functions to the
data packet. These layers include both hardware and
software. Therefore, it could be seen that optimisation for
video transmission could be achieved at both a hardware
level and at a software level. One of the first approaches is
seen in (Cherkassky et al., 2002), where the project focuses
on the issues in wireless transmission of image and video
data – video compression and network prioritization
which are the key factors to consider when making efficient
use of the limited bandwidth. In (Shan, 2005), a network
layer-based video transmission technique is discussed
1030 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
which uses equal-sized radio link protocol (RLP) packets.
Quite an extensive analysis and a mechanism into
“optimizing” video streaming for wireless transmission
is discussed in (Lu, 2009), where a method described as
“PROTAR” has been used. This again is a layer-based
approach. A packet scheduling algorithm for wireless
video streaming has been discussed in (Kang and Zakhor,
2002), which is based on unequal deadline thresholding.
First, the proposed video packet scheduling algorithm
is efficient in achieving unequal loss rate between video
packets with different importance. Second, the motion-
texture discrimination is more efficient for large motion
clips and small quantization step, i.e., large size of
texture fields. Though a key assumption in developing
this algorithm has been fixed round trip time, it is a safe
assumption to make in today’s context of high speed
networks. Wireless network modes are analysed in depth
in (Ketkar et al., n.d.), where they claim that WiMAX
could be more efficient to transmit videos over networks
as opposed to Wi-Fi and ADSL. A voice and video over
Wireless LAN technique is discussed in (Iyer et al., 2013)
and the results show how a WLAN is in fact more efficient
than using a LAN for intra-organization communication
of video and voice. In (Ramya et al., 2015), a system has
been devised to act as a framework to stream cloud based
video streams to a “CCMN” – Cloud-Centric Media
Network. This is capable of a multi-screen application
but what it does it telecasting “one screen” to many other
screens and not vice versa.
Another popular protocol proposed for video transmission
is Transmission Control Protocol. TCP is widely used by
commercial video streaming systems. When a packet has
not arrived by its playback time, a typical practice in these
commercial systems is that the client simply “stops and
waits” for this packet, and then resumes playback. This stop-
and-wait playout strategy is easy to implement. However,
stopping playout due to late packet arrivals renders the
viewing experience unsatisfactory. A continuous playout
strategy, i.e., continuing playout regardless of late packet
arrivals, also leads to unsatisfactory viewing experience,
since late packet arrivals cause glitches in the playback.
The performance of both the stop-and-wait and the
continuous playout strategies therefore depends on the
frequency of late packet arrivals during the playback of the
video. In (Wang et al., 2003), where a model is proposed
overcome these issues, discrete-time Markov models have
been developed to evaluate the performance of live and
stored video streaming using TCP. Based on these models,
guidelines have been provided as to “when” using TCP
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

for streaming leads to satisfactory performance. A crucial video source rates can adapt to the time-varying wireless
finding in this research has thus been that the fraction link capacities. They note that many problems remain,
of late packets is similar for long videos in constrained particularly in the context of wireless mesh networks
streaming while it decreases with the video length (after a – for instance, it is still unclear whether the stringent
short duration of increasing trend at the beginning of the latency constraint (usually less than a second) for video
playback) in unconstrained streaming. streaming can be met when packets need to be delivered
over multiple hops of time-varying wireless links in a
A more recent development has been the RTS (Real Time
mesh network. Typically, the wireless network is shared
Streaming) Protocol. The RTSP protocol is based on the
by both video streaming and other applications such as
HTTP protocol, and is commonly used to manage media
file downloading. The problem remains to be addressed
content streaming. This protocol doesn’t directly deal
as how to optimally allocate network resources among
with the streaming content, but uses the RTP protocol
heterogeneous traffic types, each bearing a different
to handle content transmissions. (Peltotalo et al., 2010)
performance metric (e.g., completion time for file
discusses a peer-to-peer streaming application using this
downloading versus video quality for streaming). For
protocol. The effective real-time P2P streaming system
example, in a case of where a wireless projecting device
for the mobile environment presented in this paper is
is connected to a machine, there is a possibility that the
an alternative solution to traditional client-server-based
network being used to be used for various other purposes
streaming applications. However, it also highlights that
too (Zhang et al., 2001), while broadcasting the screen
more advanced laboratory tests with different latencies and
to the projector. Such a situation can surely limit the
throughputs between peers are still needed to highlight
bandwidth required to broadcast the screen at a higher
system bottlenecks and usability issues.
frame rate of more than 25-30 fps. Even more problematic
is the situation where “many” clients would be trying to
RTMP (Real-Time Messaging Protocol) is also a widely-
connect to the projector for multi-screen sharing. This
used protocol nowadays for live video transmission.
problem presents a totally new research problem of not
RTMP is the Adobe’s network protocol used to transmit
only having to stream the videos fast, but also having to
audio, video and data between its Flash platforms (Adobe
combine them in an effective manner. Back in 2001, as
Incorporated, 2011). RTMP consists of two important
per (McCrohon et al., 2001), it was predicted that it would
structures, namely Message and Chunk. RTMP belongs
be possible to stream live lectures, “with expected future
to the application-level protocol, and usually TCP is
developments in networking technology, the quality of
accompanied with it as the transport-level protocol. The
streamed video will soon be of even higher quality
basic unit of the RTMP transmitting information is the
ensuring video streaming a promising role in the delivery
“Message”. During transmission, for consideration of
of online education.”. However, even with technologies
multiplexing and packetizing multimedia streams, each
available 15 odd years later, it is still doubtful if a “real-
Message will be split into some “Chunks” (Lei et al., 2012).
time” output can be achieved – to transmit the video feeds
of many device and combine them in one device.
Another example of P2P (Peer to Peer) streaming is seen
in (Tran et al., 2003), where first, an end-to-end transport
Dynamic Adaptive Streaming over HTTP (DASH) (also
architecture for multimedia streaming over the Internet
known as MPEG-DASH) is quite a modern technology
is presented. Second, a new multimedia streaming TCP-
which is now used for high quality streaming of media
friendly protocol (MSTFP), which combines forward
content over the Internet delivered from conventional
estimation of network conditions with information
HTTP web servers. Like Apple’s HTTP Live Streaming
feedback control to optimally track the network conditions,
(HLS) solution, MPEG-DASH works by breaking
is discussed. As the said protocol is TCP based, it would
the content into a sequence of small HTTP-based file
still have the inherent issues which TCP communication
segments, each segment containing a short interval of
has always had.
playback time of content that is potentially many hours in
duration, such as a movie or the live broadcast of a sports
In (Seeling et al., 2004), a discussion is available on how
event. This is probably the best solution that is available
to evaluate the network performance for single-layer
for live streaming of content at the moment. The MPEG
and two-layer encoded video using “traces”. As (Zhu
DASH standard was published by the ISO in April 2012
and Girod, 2007) notes, in network video transmission,
(Andy Salo, 2012).
cross-layer information exchange is required, so that

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1031
PROCEEDINGS
A comparative evaluation of most of these techniques
is seen in (Aloman et al., 2015). They have conducted a
comparative performance evaluation of MPEG DASH,
RTSP, and RTMP streaming protocols over 4G and Wi-Fi
(IEEE 802.11g/n) real networks in terms of QoE (Quality
of Experience), tested both video on demand and live
video streaming. Results in this analysis has suggested that
RTSP is more efficient than MPEG DASH for starting the
video playback, but at the expense of decreasing QoE due
to packet losses. In addition, the long pre-loading time
interval needed by MPEG DASH or RTMP permits to
alleviate the impact of the packet losses which take place
during the transmission, as revealed by a lower number of
re-buffering events for these two protocols. MPEG-DASH
surely the best available protocol currently, for streaming
content such as TV series and / or live events. This is for
requirements that demand both high quality and speed at
the same time. However, this obviously requires a network
with high bandwidth. Therefore, for other lesser needs,
specifically for purposes of screen mirroring and such,
RTMP is a lucrative option.
B. IoT Based Embedded Systems for Video
Transmission
Secondly, we should look at the developments of the IoT
sphere as well. Transmitting video over the web using
embedded devices has surely been an area of crucial
significance for the development of the IoT ecosystem.
Most of the integrations among various technologies have
been possible due to the rapid accession and research on
development boards such as Raspberry Pi and Arduino.
A classic integration of embedded systems with IoT and
image processing is seen in (Kulkarni et al., 2014). Here
a surveillance robot has been developed which is capable
to transmit a video stream over the web. To be able to do
this on a simple chipset such as Arduino, clearly shows
that complex systems could now be implemented using
available technologies. Image processing has also become
more viable on a Pi as it contains both a Graphics Processing
Unit (GPU) along with an internal memory. This aspect
has been explored in (Ujjainiya and Chakravarthi,
2006),where the system is used to detect the objects in
front of a vehicle by using a camera module. The camera
detects the picture and operation is performed in OpenCV
to detect the edges of the detected picture. Another image
processing example is seen in (Manasa et al., 2015), where
the paper proposes the design and implementation of
object counting algorithm based on image processing by
1032 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
using the RPi on real time basis. RPis are also ideal for
the construction of surveillance systems both in-house
and remote. This is explored in (Chuimurkar and Bagdi,
2016), where a video stream is transmitted via the web to
a mobile device.
A recent device which professes IoT capabilities is the
Intel® Compute Stick shown in Figure 1 (Pete Carey, 2015),
which came to the market as late as 2016. This is a complete
computer in a “stick” with a HDMI port. This too, would
be ideal for streaming video via a wireless method, but
the only drawback of this device is that it is quite costly
compared to the RPi. Research done using this device is
yet to be seen and as thus, the reliability and performance
is yet to be assessed.
Figure 1: Intel Compute Stick features
C. Screen Mirroring / Projection
Nowadays, screen mirroring devices (McGill et al.,
2014) have been built by integrating the wireless video
transmission techniques discussed above and using
various chipsets. There is also a possibility where projectors
will soon become completely wireless, minimizing the
many practical restrictions (not being able to project
more than one screen, issues setting up wire cords etc.)
that traditional projectors undergo. Therefore, a review of
these available devices is apt.
We now look at existing projectors out in the market. The
existing projectors provide the basic functionality of where
one wired device can be projected. In the last 3-4 years,
many projectors, however, have been developed with the
capabilities of WLAN (BenQ, n.d.) (Epson, n.d.), meaning
the projector can connect to a client using a wireless mode.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

A few of these are in fact capable of screen sharing like However, as it may be, there has not yet been a model
most of the Epson® PowerLite versions using the Epson capable of connecting to the internet to pull a feed from
iProjection™ App (Epson, 2015). These algorithms and a device not located in the current location. Furthermore,
implementations are proprietary and allow for a limited almost all these models use proprietary software meaning
scope of application. they are only compatible with the set list of devices
produced by the manufacturer. Also, there is no way that
Apart from these commercial products, there is little a customer with a projector with no wireless capabilities
literature available on projects conducted at an academic can enjoy these features unless s/he buys a new projector.
level relating to projectors itself. This is probably due
to the nature of the projectors being predominantly a D. Online Collaboration Tools
commercial device. However, a couple of research projects
is to be noted. In (Chheda et al., 2013), a Raspberry Pi has Next is to consider commercial software solutions which
been used to transmit the video feed from a desktop to the address the need of online collaboration. Many tools are
Pi. The mode of transmission is Wi-Fi and this could be available such as GoToMeeting (“GoToMeeting,” n.d.),
identified as a basic implementation of media streaming Skype for Business (Skype, n.d.), Join.me (Join.me, n.d.),
over a network. No “enhancing” procedure is available screenleap.com (screenleap, n.d.) and so forth. A few of
in transmitting the frames and this project depicts and these provide multi-user interaction while some provide
predicts (in further research), how the same concept interfaces for customized viewing such as manipulating
could be used to create an adapter-like device for already PDF files, presentation slides etc. All these solutions,
available projectors. however, limit their usability and reliability on the
availability of their proprietary systems. A summary of
When considering the technical aspect, it could be seen these systems is seen in Figure 3:
that there are many projectors available in the market
deploying one or more of the above technologies intended Software Multi user Depen- Image
to achieve real-time wireless video transmission. However, Platform dant on a optimisa-
most of the companies who have involved themselves web plat- tion tech-
in developing projectors of this sort have provided a form? niques
profound description of the technicalities of the same. Skype for Yes Yes No
Leading projector brands such as BenQ (BenQ, n.d.), Business
Optoma (Optoma, n.d.), Epson (Epson, n.d.), Panasonic
Join.me No Yes No
(Panasonic, n.d.) and Acer (Katie Scott, 2008) seem to
have built wireless projectors as of late. A few of the Screenleap. No Yes No
models offered by these brands have the “multi-screen” com
split feature (Shown in Figure 2). Many include an app for TeamViewer Yes Yes No
mobile devices (Android and iOS), and come equipped
with proprietary algorithms. The NEC MultiPresenter Figure 3: Comparison of software solutions for
Stick (NEC, 2016) is a 2016 device with quite sophisticated screen sharing
features but it still requires a NEC compatible projector to
work with. E. Pre-processing Before Streaming

Is it also important to look at research that has gone

into pre-processing image frames before being streamed
via a network. This means that the streams received by
whatever device (from clients) will undergo a stage of
pre-processing to “enhance” the frames received. This
may vary from changing basic properties such as contrast,
brightness, mid-tones and so forth. This could be achieved
via existing graphic processing libraries such as OpenCV
(open source - (“OpenCV library,” n.d.)) and CUDA
Figure 2: Application of screen-split option in a
(proprietary - (“CUDA Zone,” 2015)). Both libraries are
wireless projector
quite powerful in their own ways.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1033
PROCEEDINGS
(Di Salvo and Pino, 2011) has highlighted how applications
have been parallelized in various areas using CUDA, to
achieve very high-performance in time processing keeping
the same performance in terms of accuracy. They further
note that the main portion of the available CUDA based
approaches deal with the parallelization of generic image
processing operations, whereas much more work should
be done in the biomedical and video-surveillance fields,
where this is mostly used. OpenCV on the other hand
is more prominently used for “detection purposes” in
videos. This is shown in many research literature including
(Pulli et al., 2012), (Shah, 2014) and (Farhadi-Niaki and
Mehrvar, n.d.).
III. DISCUSSION
The methodology of this research centres around the
performance measures of different video transmission
techniques and how each performs for the purpose of live
screen mirroring. Such factors include transmission speed,
reliability, quality of decoding, platform dependability,
scalability and usage of image optimisation techniques.
Firstly, when referring to the status quo of video
transmission protocols MPEG-DASH and RTMP seem to
be two protocols which are in the forefront when it comes
to real-time video transmission.
Another concern is “how” the video frames should be
transmitted. Should they be transmitted as “frame by
frame” or should it be after a pre-compression technique
has been applied such as after converting the stream to a
format like “mp4”? What are the already available packages
which can do this? “FFmpeg” (“FFmpeg,” 2017) is one
such strong open source library available for streaming
audio / video which supports both above techniques (Wan
and Dai, 2016). This easily supports Linux based systems
and thus runs fine on embedded chipsets discussed here.
Furthermore, this library also supports mobile platforms
such as Android. In (Fu et al., 2010) an Android-based
codec application is designed and implemented using Java
with FFmpeg. Therefore, this would be an ideal solution
for anyone planning to implement a video transmission
algorithm.
Secondly, what is a possible embedded system which can
be optimized for video transmission? In comparison,
Arduino has limited support for video streaming.
ComputeStick is surely a high-end option which supports
video streaming but when looking at other concerns such
1034 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
as the cost and cross platform capabilities, the ideal chipset
for mini scale projects, specifically for screen mirroring
concerns would be the Raspberry Pi. Another similar
concern is what methodology is most appropriate for
network connectivity. For this, the most intuitive solution
would be to use the technology which offers the “fastest”
bandwidth. However, it should be noted that video
transmission does not solely depend on the speed of the
network, but rather on concerns of priority accorded to
video packets. WiMAX + LTE technologies warrant the
most optimized connectivity (Iyer et al., 2013) for video
streaming.
Thirdly, when looking at the aspect of screen mirroring,
as discussed, most projectors are now becoming wireless.
It is a matter of time for all our devices to come void of
HDMI / VGA ports. More academic research should be
undertaken in this area, though, to ensure that the most
optimized methods of screen mirroring maybe identified
and studies done, as opposed to comparing commercial
solutions, due to the restrictions posed via proprietary
algorithms. Further research should look at extending
its functionality to intelligently analyse the video streams
received and provide some basic image corrections
(Sukthankar and Mullin, 2000) before it is broadcast.
IV. CONCLUSION
This paper discussed, at length, the various advancements
in the fields of wireless video transmission techniques,
IoT and embedded systems which can facilitate such
techniques, screen mirroring techniques and various
pre-processing that can be done to livestreams using
image processing algorithms. Though one may claim
these are distinct research areas, that is not the case in
reality. Many real-world applications are being modelled
using these aspects. The development of chipsets has now
allowed a massive level of data manipulation, graphics
data processing and storage abilities. It is important that
these chipsets are used to achieve their true potential
by researchers and it could be said that much has been
already done. However as mentioned, there is room for
improvement in integrating all these technologies to build
useful embedded systems for day-today use. Such research
need to be undertaken with the constant thought in mind
that the limitation of network bandwidth should not
hinder the performance and speed of the streams being
broadcast over a network.
 ISBN number: 978-955-0301-41-6
REFERENCES
Adobe Incorporated, 2011. Adobe® Flash® Media Server
4.5.
Aloman, A., Ispas, A.I., Ciotirnae, P., Sanchez-Iborra,
R., Cano, M.D., 2015. Performance Evaluation of Video
Streaming Using MPEG DASH, RTSP, and RTMP in
Mobile Networks. IEEE, pp. 144–151. doi:10.1109/
WMNC.2015.12
Andy Salo, 2012. MPEG DASH: A Technical Deep Dive
and Look at What’s Next.
BenQ, n.d. BenQ W1500 Home Entertainment Wireless
Projector [WWW Document]. Cable Free Easy Set-Up.
URL http://www.benq.com/product/projector/W1500/
features/
Cherkassky, V., He, X., Balasubramanian, P., 2002. Wireless
Transmission of Image and Video Data.
Chheda, D., Darde, D., Chitalia, S., 2013. Smart Projectors
using Remote Controlled Raspberry Pi. Int. J. Comput.
Appl. 82.
Chuimurkar, M.R., Bagdi, V., 2016. Smart Surveillance
Monitoring And Automation System Using Raspberry PI
and PIR Sensor.
CUDA Zone [WWW Document], 2015. . NVIDIA Dev.
URL https://developer.nvidia.com/cuda-zone (accessed
3.26.17).
Di Salvo, R., Pino, C., 2011. Image and video processing on
CUDA: state of the art and future directions. MACMESE
11, 60–66.
Epson, 2015. Epson iProjectionTM App [WWW
Document]. Epson IProjectionTM App. URL https://
epson.com/wireless-projector-app
Epson, n.d. Epson US [WWW Document]. Epson Wirel.
Proj. URL https://epson.com/how-wireless-projectors-
work
Farhadi-Niaki, F., Mehrvar, H., n.d. A Review on
“Implementation of Real-time Image and Video
Processing, and AR on Mobile Devices.”
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
FFmpeg [WWW Document], 2017. URL https://ffmpeg.
org/ (accessed 3.26.17).
Fu, X., Wu, X., Song, M., Chen, M., 2010. Research
on audio/video codec based on Android, in: Wireless
Communications Networking and Mobile Computing
(WiCOM), 2010 6th International Conference on. IEEE,
pp. 1–4.
GoToMeeting [WWW Document], n.d. . GoToMeeting.
URL https://www.gotomeeting.com
Iyer, A., Badgujar, A., Eaga, M., Iyer, R., 2013. Voice and
Video over Wireless LAN. Int. J. Sci. Res. Publ. 3, 1–5.
Join.me, n.d. SHARING IS CARING. AND SWAPPING.
AND EDITING. [WWW Document]. URL https://www.
join.me/screen-sharing
Kang, S.H., Zakhor, A., 2002. Packet scheduling algorithm
for wireless video streaming, in: International Packet
Video Workshop.
Katie Scott, 2008. Acer P5260i wireless projector.
Ketkar, O., Ninawe, P., Panchal, S., Sonar, Y., n.d. VIDEO
TRANSMISSION USING WIRELESS NETWORK.
Kulkarni, C., Grama, S., Suresh, P.G., Krishna, C.,
Antony, J., 2014. Surveillance Robot Using Arduino
Microcontroller, Android APIs and the Internet, in:
Proceedings of the 2014 First International Conference
on Systems Informatics, Modelling and Simulation. IEEE
Computer Society, pp. 107–111.
Lei, X., Jiang, X., Wang, C., 2012. Design and
implementation of streaming media processing software
based on RTMP, in: Image and Signal Processing (CISP),
2012 5th International Congress on. IEEE, pp. 192–196.
Lu, M.-H., 2009. Optimizing transmission for wireless
video streaming. Ozyegin University.
Manasa, J., Pramod, J.T., Jilani, S.A.K., Hussain, M.S.J.,
2015. Real Time Object Counting using Raspberry pi.
McCrohon, M., Lo, V., Dang, J., Johnston, C., 2001. Video
streaming of lectures via the Internet: An experience, in:
ASCILITE. pp. 397–405.
McGill, M., Williamson, J., Brewster, S.A., 2014.
Mirror, mirror, on the wall: collaborative screen-
10TH INTERNATIONAL RESEARCH CONFERENCE 1035
PROCEEDINGS
mirroring for small groups. ACM Press, pp. 87–94.
doi:10.1145/2602299.2602319
NEC, 2016. NEC MultiPresenter Stick [WWW Document].
URL http://www.projectorpoint.co.uk/meeting-room-
technology/presentation-systems/nec-multipresenter
OpenCV library [WWW Document], n.d. URL http://
opencv.org/ (accessed 3.26.17).
Optoma, n.d. Wireless [WWW Document]. URL https://
www.optoma.co.uk/projectortechnology/Wireless
Panasonic, n.d. Wireless Projector for iOS [WWW
Document]. URL http://panasonic.net/avc/projector/ios/
Peltotalo, J., Harju, J., Väätämöinen, L., Bouazizi, I., Curcio,
I.D.D., 2010. RTSP-based Mobile Peer-to-Peer Streaming
System. Int. J. Digit. Multimed. Broadcast. 2010, 1–15.
doi:10.1155/2010/470813
Periscope - Live Video - Android Apps on Google Play
[WWW Document], 2017. URL https://play.google.
com/store/apps/details?id=tv.periscope.android&hl=en
(accessed 4.28.17).
Pete Carey, 2015. Review: Intel’s computer-on-a-stick:
Awesome idea, but its day hasn’t arrived [WWW
Document]. URL https://phys.org/news/2015-06-intel-
computer-on-a-stick-awesome-idea-day.html (accessed
3.25.17).
Pulli, K., Baksheev, A., Kornyakov, K., Eruhimov, V., 2012.
Real-time computer vision with OpenCV. Commun.
ACM 55, 61–69.
Ramya, R., Thendral, G., ME, M.T.K., 2015. EFFICIENT
PARALLEL FRAMEWORK FOR ADAPTIVE VIDEO
STREAMING IN CLOUD SOCIAL TV. Asia Pac. J. Res.
Vol Issue XXIV.
screenleap, n.d. Screenleap [WWW Document]. URL
http://www.screenleap.com/
Seeling, P., Reisslein, M., Kulapala, B., 2004. Network
performance evaluation using frame size and quality
traces of single-layer and two-layer video: A tutorial. IEEE
Commun. Surv. Tutor. 6.
Shah, S., 2014. Real-time image processing on
low cost embedded computers. Techincal Rep. No
UCBEECS-2014–117.
1036 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Shan, Y., 2005. Cross-layer techniques for adaptive video
streaming over wireless networks. EURASIP J. Adv. Signal
Process. 2005, 871928.
Skype, n.d. Skype: Group screen sharing [WWW
Document]. Group Screen Shar. URL https://www.skype.
com/en/features/group-screen-sharing/
Sukthankar, R., Mullin, M.D., 2000. Automatic keystone
correction for camera-assisted presentation interfaces,
in: Advances in Multimodal Interfaces—ICMI 2000.
Springer, pp. 607–614.
The Evolution of Mobile Technologies, 2014.
Tran, D.A., Hua, K.A., Do, T., 2003. Zigzag: An efficient
peer-to-peer scheme for media streaming, in: INFOCOM
2003. Twenty-Second Annual Joint Conference of the
IEEE Computer and Communications. IEEE Societies.
IEEE, pp. 1283–1292.
Ujjainiya, L., Chakravarthi, M.K., 2006. Raspberry–Pi
Based Cost Effective Vehicle Collision Avoidance System
Using Image Processing.
Wan, F., Dai, Y., 2016. The Research of video format
conversion based on educational video-on-demand.
Wang, B., Kurose, J., Shenoy, P., Towsley, D., 2003. A model
for TCP-based video streaming. Univ. Mass. Tech. Rep.
Youtube Live vs Facebook Live compared to Online Video
Platforms [WWW Document], 2017. URL https://www.
dacast.com/blog/comparing-youtube-live-and-facebook-
live-with-pro-video-streaming/ (accessed 4.28.17).
Zhang, Q., Zhu, W., Zhang, Y.-Q., 2001. Resource
allocation for multimedia streaming over the Internet.
IEEE Trans. Multimed. 3, 339–355.
Zhu, X., Girod, B., 2007. Video streaming over wireless
networks, in: Signal Processing Conference, 2007 15th
European. IEEE, pp. 1462–1466.
 ISBN number: 978-955-0301-41-6
A REVIEW ON HOME AUTOMATION
WITH HUMAN BEHAVIOUR
MLNT Chandrasekara 1#, WPJ Premarathne1
1
Faculty of Computing, General Sir John Kothelawala Defence University, Rathmalana, Sri Lanka
# nimesh94@yahoo.com
Abstract - Smart homes have been viewed with increasing
interest by both home owners and the research community
in the past few years. One reason for this development
is by using modern automation technology such as
Arduino in the home or building, it promises considerable
savings of energy. Therefore simultaneously reducing the
operational costs of the building over its whole lifecycle.
At present one of major issue in Sri Lankan society is
high electricity consumption. As a solution for that we
are proposing a system which identifies human behaviour
to make decisions mainly for lighting. The integration of
these everyday systems will give the average homeowner
the control they desire within their home easily. This work
highlights various current systems with their pros and
cons and how to prevent those weaknesses.
Keywords - smart home, automation, arduino
I. INTRODUCTION
The concept called “home automation” has been there since
the late 1970s. (Cyril Jose and Malekian, 2015) But with the
advancement of technology in both software and hardware
fields, people’s expectations of how an automation should
work or the way that the services should be provided has
changed a lot during the course of time. So idea of “home
automation systems” were improved day by day. This work
highlights various flaws in existing home automation
systems. In this paper, it has addressed what is Home
automation and how it works under various technologies
with advantages and disadvantages of them. After that the
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
challenges in home automation systems from the point of
view of both the homeowner and a commercial user. This
work goes on to explain why home automation systems are
such attractive targets for an attacker and how the various
technologies handle it.
Under the first topic it has mainly considered about a brief
introduction about this review paper and what is the content
of the whole paper. Under the second topic “technical
review” it presents a brief explanation about various home
automation systems that exist and some concepts that are
planning to build. It also mention about the technologies
which were used to make these kind systems much more
efficient and user friendly to the user. In “technological
review” the main target was to address the technologies
that were used while the researches completing their
projects such as neural networks, programming boards
etc... The work concludes by explaining future directions
home automation System Research could take in much
more efficient and cost effective way.
II. SYSTEM REVIEW
Under this topic it has been descried about the existing
home automation systems and the systems which are
still in development stage or which are still concepts. If
we look at various home automation systems over the
course of time, developers have always tried to provide
efficient, convenient, and safe ways for home inhabitants
to access their homes. Irrespective of the change in user
10TH INTERNATIONAL RESEARCH CONFERENCE 1037
PROCEEDINGS
expectations, advancement of technology, or change of
time, the role of a home automation system has remained
the same.(Cyril Jose and Malekian, 2015) Each and every
system uses various kinds of technologies to implement
there system and there are advantages and disadvantages
of using those.
The paper called “Home Automation Systems - A Study”
(Palaniappan et al., 2015)by S. Palaniappan, N. Hariharan,
N. T. Kesh, S. Vidhyalakshimi briefly describe about
some home automation systems. Based on all the systems
surveyed with their pros and cons, this paper presents the
features to be added to an ideal system for home automation
with remote access. In this paper it shows comparison
between systems such as Bluetooth, GSM, mobile based,
Zigbee and Wi-Fi. The systems that have been considered
by the researches have mentioned about certain common
features. All of them are based under a same principle
which is those kind of systems use a basic underlying
communications technology. The pros and cons of the
system originate according to the underlying technology.
All of those systems have a controlled electrical structure
which is used to connect with the electrical peripherals.
They have mentioned about a common command system
that will be used to issue commands to the control circuits.
A major part in the system is done by the user interface.
This decides how the user will interact with the system
and extent of control the user applies while them using the
system. This effects the usability and the user friendliness
of the system. Most systems have security features as well.
It has added to ensure only authorized access to avoid
unauthorized accessing and unauthorized updating of
current data to protect confidentiality and integrity of the
system. To prevent that most of the systems use various
kinds of security mechanisms. At the same time these kind
of system should be available from all over the world to a
user and in real time. These are the main points that the
research considered from the overall study.
There’s another paper which was concerned about
above technologies. That paper is “An overview of home
automation systems” (Asadullah and Raza, 2016) by
M. Asadullah and A. Raza. In this paper different home
automation systems were surveyed with their strengths
and weaknesses.as an example Bluetooth based system is
a flexible and low cost medium, but that kind of system
is limited to a short range of Bluetooth wireless network.
System which are controlled by the voice recognition are
most suitable for elderly and handicapped people, because
they can control them by saying the name of appliances
1038 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
or by using a specific key word. Such systems are noise
sensitive and the problem is accuracy can be affected by
signal to noise ratio (SNR). So this kind of system is not
suitable for a noisy place. Another automation mechanism
used is ZigBee RF modules for the implementation of
wireless network. Inside these kind of wireless networks
the user has the full remote controlled access of home
appliances. In this paper a GSM based module is also
studied and according to this system user can control
and monitor the home appliances via text message from
his or her mobile phone. But these kind of systems are
much more slowly compared to other systems. Internet
of things based home automation system can only work
in the presence of internet. The rapid growth of IOT
devices brings concerns and benefits. The future of home
automation system requires to make homes smarter and
more convenient to make things easier to the user. As for
the future work it is suggested to develop image processing
based home automation system using the technologies
which discussed above. In such kind of automation
system, home appliances will be controlled by different
gestures which will be identified through the camera.
Moreover, home automation system could be developed by
using technologies such as interfacing biomedical signals
(Electromyography (EMG) signal) with computer, it will
offer opportunity to amputee to control appliances from
different arm gestures. It will be useful in area of robotics
for controlling robot through gesture for different tasks. In
addition, future work would be implementation of above
discussed home automation systems on a much more large
scale, such as factories, industries, offices etc.
So if we take those technologies individually, first we
can concern about the Bluetooth technology. The paper
called “Bluetooth Based Home Automation and Security
System Using ARM9” (Naresh et al., 2013, p. 9) by D.
Naresh, B. Chakradhar, and S. Krishnaveni, has clearly
stated about how that research used Bluetooth technology
to fulfil the need of home automation. This paper put
forwards the design of home automation and security
system using ARM7 LPC2148 board. Home appliances
are directly connected to the board and communication
is established between the board and ARM9 using a
Bluetooth device. According to the paper, that low cost
system was designed to improve the standard living in
home. Its remotely controlled by ARM9 provides help
and assistance especially to disabled and elderly. In order
to provide safety protection to the user, all the electrical
switches were replaced by the low voltage activating
switches. Moreover implementation of wireless Bluetooth
 ISBN number: 978-955-0301-41-6
connection in control board allows the system installation
in more simple way. The control board was directly
installed beside the electrical switches. So the switching
connection was controlled by relay. Furthermore, flexible
types of connections were designed as backup connections
to the system in a case of emergency. The connected GUIs
are fully synchronized with the ARM board. They show
the real-time switches status to the user. The system is
designed in a very much user-friendly way with a good
interface. As for the future work, they were planning to
implement the speech recognition in order to control via
voice control. In that current system all the voice signal
inputs to the ARM9 were transmitted to the Window GUI
for signal processing. These are the key features which were
mentioned on that paper. Due to the use of Bluetooth, the
weakness of that system was the range of the system was
concluded into a small area.
There was a paper called “Smart GSM based Home
Automation System” (Teymourzadeh et al., 2013) by R.
Teymourzadeh, SalahAddin Ahmed, Kok Wai Chan,
and Mok Vee Hoong, maily concerning about the
GSM based system. This research work investigates the
potential of “Full Home Control”, which is the aim of
the Home Automation Systems in near future. The study
and implementation of the home automation technology
using GSM used a modem to control home appliances
such as light, air conditional system, and security system
via SMS. The proposed system was mainly focused on the
functionality of the GSM protocol, which allows the user to
control the target system away from residential, using the
frequency bandwidths. For the development of the smart
GSM-based home automation system they have used the
concept of serial communication and AT-commands. So
home owners will be able to get a feedback status of each
and every home appliances under control whether it is
switched on or off remotely from their phones. So simply
the mobile phone acts as the remote in this system. That
proposed prototype was implemented and tested by the
researchers with maximum of four loads and shows the
accuracy of more than 98%. Finally with that kind of big
accuracy this system would have been an effective system.
Another research was concerned with the use of GSM
technology for their system. That was “Microcontroller
Based Home Security System with GSM Technology,”
(Hasan et al., 2015)by R. Hasan, M. M. Khan, A. Ashek,
and I. J. Rumpa. In this paper, design and implement of
a microcontroller based home security system with GSM
technology have been presented and analysed. Mainly
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
with two microcontrollers and other devices such as
LED, LCD display, Buzzer and GSM Module are used for
reliable operation of the proposed system. They added
some ways to input to the system. First way is that the
mobile phone is interfaced with microcontroller through
a Bluetooth device in order to control the system. Then,
a manual keypad is another way to lock or unlock the
system. A Compiler Code “Vision AVR” is used to design
a program that controls the system along with maintaining
all functions such as controlling and security. That paper
presented design and implementation of a smart home
security system based on microcontroller along with GSM
for user friendly application. That system was intelligent
enough to monitor environment. At the same time, the
user is informed about each and every security breach
using GSM network that provides a special opportunity
whenever the user stays at far away from home. However,
Android application was the most spectacular feature in
order to control the system through a Bluetooth device.
The system has provided the reliable operation within
reasonable cost and removes the system complexity. In
this work, traditional burglar alarm mode, LED lights and
LCD are the promising features used to ensure reliability.
The whole system was developed on a practical home
security system which requires significant effort to install
it. The system is also appropriate for commercial purposes
because of versatile ways of security, confidentiality and
controllability. At the latter part of that paper it concluded
with test results of practical circuit which show the
proper functionality and also validate the reliable security
compared with its reasonable cost.
Few researches concerned about using Wi-Fi technology for
their system. One of them is “Design and implementation
of a WIFI based home automation system,”(ElShafee and
Hamed, 2012) by A. ElShafee and K. A. Hamed. That
paper basically concerned about the design and prototype
implementation of new home automation system by using
WiFi technology as a network infrastructure connecting
its main parts. That proposed system consists of two main
components such as the server and the hardware interface
module. That paper proposed a low cost, safe, universally
accessible, auto-configurable, remotely controlled solution
with many more functions. The main approach stated in
the paper has accomplished the target to remotely control
home appliances by using the Wi-Fi as the medium to
connect system parts. Wi-Fi technology capable solution
has proved to be controlled remotely, provide home
security and is cost-effective by comparing to the previous
systems. In that paper the system design and the main
10TH INTERNATIONAL RESEARCH CONFERENCE 1039
PROCEEDINGS
architecture used were discussed briefly, and prototype
presented the basic level of home application control then
the remote monitoring was implemented according to
the design. Finally, the proposed system was better from
the scalability in the point of view by comparing to the
previous automation systems. Unlike most of available
automation system in the market, that system was scalable
because by using one server can manage many hardware
interface components as long as it is in the range of that
WiFi network coverage.
Another research mentioned about the wireless method
“Zigbee” as their technology. That paper was “Wireless
home automation system using zigbee,” (Arul, 2014)by
S. B. Arul. This paper contained the overall design of a
wireless home automation system that was constructed and
implemented. Main concept of that project on recognition
of voice commands and uses low-power RF ZigBee wireless
communication components which are quite cheap than
other technologies. The home automation system was
proposed to control all lights and electrical applications
in an ordinary home or office using voice commands. The
system was tested and confirmed. The main verification
tests included voice recognition response test which was
known as the “indoor ZigBee communication test”. That
test involved sending a total of 70 basic commands and the
rate of efficiency was 80.05%. Main target of that system
was elderly and disabled people. This research had worked
to achieve main four key features which are
• Low Power
• Interoperability
• Mesh Networking
• Robust
The system implements voice recognition unit using HM
2007. That system implemented the wireless network
using ZigBee modules for their efficiency and low power
usage. So the end test results were promising in that project
because of those reasons.
There was a research called “Intelligent Home Automation
System using BitVoicer,” (Krishna and Lavanya, 2017) by
I. Krishna and K. Lavanya which is also concerned about
technology in Zigbee. In most of developed countries,
many households have home automation systems which
notice their environment and act on the data to trigger
home devices. But these systems are not very popular in
developing nations with compared to developed nations
due to their high pricing. Also, almost all of these kind
of systems are hard coded so they just giving decisions
1040 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
based on the iterative conditions it recorded. The
proposed Zigbee based automation system and Wi-Fi
network are combined through a mutual home gateway.
That combination simply offers network interoperability,
a simple and a user friendly interface, and remote access
to the system. That proposed system was very useful
for elderly and physically disabled persons who mainly
needed the physical assistance for daily requirements.
It removes the limitation of wiring difficulties and also
the range of voice commands can be incresed by using
wireless routers and internet of things. On the off chance
the practical usefulness and the cost of that automation
decrease will enable compatibility with future technologies
for improved user support. It mainly stated that creation
capacity as well as innovation level of the system was going
to enhance increasingly. In Future, by using a technology
like Arduino, it can enable the control via central server
through smart mobile phones.
Another related system which was created “Self-configuring
home automation networks based on psychophysical
principles,” (Flessner and Frenken, 2016) by is J. Flessner
and M. Frenken. That paper says that the apparent
comfort of building environment may be improved by the
development of the interaction between users and home
automation control. A higher level of interaction is possibly
achievable by including the psychophysical condition of
the users. A methodical way for the design of an adaptive
automation system was presented. A fundamental part of
this method was the comparison between current guiding
principle and discoveries about the effect of environmental
factors on the user’s mental and physical condition. The
main application of this method was basic principle for the
design of building controls. According to the researcher
it could lead to the creation of healing environments. To
respond dynamically on changes of the user’s condition,
a rule base system had to be developed to create a self-
configuring automation system. As the future works, the
implementation of the idea for a selected subset such as
ventilation, heating or lighting had to be done to review
the feasibility of the proposed system. For that purpose,
the interface between user and automation control need to
be designed. One opportunity is the use of identification
mechanisms to dynamically react according to different
people inside the building environment. At the same time
according to the identification of the users, an adaptive
home automation system needs to respect different people
with various requirements. Therefore the process to
handle changing or oppositional requirements is required
according to the opinion of the specific researcher.
 ISBN number: 978-955-0301-41-6
According to the current systems and current technologies
which are mentioned above, it can be concluded to a single
table for the features of each and every system under the
system review.
Table 1. Current Systems
III. TECHNOLOGY REVIEW
Under the sub section of technology review it has mainly
considered about some papers which are deeply discussed
about some technologies related to the proposed system.
The first paper is “A tool for the analysis of energy systems
in Smart Cities,”(Fabbri et al., 2016) by G. Fabbri, C.
M. Medaglia, D. Sbordone, and B. Di Pietra. This paper
mainly considered about the analysing energy usage
and identifying the patterns within those records. Cities
represent three quarters of world energy usage and 80%
of CO2 emissions globally being the most dangerous of
any environmental policy challenge. To deal with this
continuous urban growth there should be new methods
to manage cities and make them more efficient in order
to prevent those kind of problems. The union between the
world of energy and the field of information technology,
will pave the way for a new bionetwork of services which
should enable both reduction of energy consumption
and improve the better quality of life. Due to the massive
development in the field of IT, it is converting all economic
sectors, including energy to an optimized way than the
past. In this paper the main issues related to smart grids
and their integration with the cities will be analysed
emphasizing the significance to set of simulation tools and
models aimed at facilitating feasibility analysis for smarter
energy technologies.(Fabbri et al., 2016) according to
that paper a simplified tool was under development to
introduce to the micro grid and distributed power systems
projects. The tool takes in account both conventional
and renewable technologies for energy and that was
designed to simulate as a micro grids or as a distributed
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
generation within a larger grid. According to that paper
the first results from that tool had been compared with
experimental results and by using these kind of data from
previous tests and used to perform simulations of various
systems.
There’s a paper which mainly considered about how to
identify the human behaviour for such kind of a system.
The paper is “(Ebeid et al., 2016),” by E. Ebeid, R. Heick, and
R. H. Jacobsen. A framework for deducing user behaviour
from smart meter data has been presented in that paper.
According to the researcher the framework separates
the disaggregation analyses from the natural language
data processing. It uses uniform interfaces to exchange
data seamlessly between its items. It relies on a shared
database which enables easy upgrades of the framework
components. Experiments have been performed and
the results have been evaluated for choosing the data
reconstruction and Load Disaggregation Algorithm
(LDA) algorithms. The experiments have shown the
complete framework’s data flow process starting from
aggregating the main meter data and ending by delivering
natural language reports. The experiments have validated
the feasibility of the framework. Future work will include
analysis of multiple users ‘ behaviour. Different smart
meter data will be utilized for improving the detection
accuracy of the LDA algorithms. At the same time the
researcher also concerned about the privacy of the user as
well.
The paper called “Home automation system with android
application,” (Azni et al., 2016) by M. N. Azni et al is
mainly considered about how to use android to create
such a system. This work demonstrates a simple home
automation system that permits the consumer to control
home applications via wireless medium such as Wi-Fi
using an android application. Electronic peripherals such
as Lights, A/C, electronic doors and fans could be used
in that proposed system. In this system, the controlling
and monitoring the appliances can accomplish by using
mobile based on Android application. The web interface
had framework known as Restful Api and function as to
control Raspberry Pi gpio using an http request. Besides
that, Android apps will exploit the services provided by
Restful Api for controlling gpio of Raspberry Pi. Both
the methods are using Restful Api but it is included
in the web interface, whereas for Android we need to
add their suitable functionalities. The server software
was run in a desktop PC. Raspberry Pi board was used
as the main board to connect the appliances thru input
10TH INTERNATIONAL RESEARCH CONFERENCE 1041
PROCEEDINGS
ports and output ports. The communication between
the smartphone and the Raspberry Pi board is based on
a wireless network such as WiFi. As the conclusion, that
system could be defined as useful application to everyone
and especially for disabled and elderly people. This system
also had been developed with a user friendly Interface to
allow more users will able to implement the system in a
much cheaper way. According to the researcher’s point of
view that project also can be developed by adding support
to IPhone (iOS) and Windows Mobile user because smart
phones users use other type of operating systems as well.
By covering all of these main types of mobile operating
system, it can increase the compatibility of the system.
Therefore, more users will be able to use the system and be
more beneficial to society.
IV. DISCUSSION
This paper gives a brief comparison of all the above
systems described. The systems that have been studied
have certain common features. All these systems use
a basic underlying communications technology. The
advantages and drawbacks of the system derive from this
underlying technology. All the systems have a control
circuitry that is used to interface with the electrical
appliances. There has to be a common command system
that will be used to issue commands to the control
circuits. The next important feature of the system is the
user interface. This determines how the user will interact
with the system and extent of control the user exerts over
the system. According to my knowledge the user interface
should come with a web application which has a linked
mobile application. So that people who are using all kinds
of operating systems can access the system without having
a problem. Such a system should also have the feature of
being easy to install and user friendly as well. Only then
can automated homes become commercially viable from
user to user. There should be a lot of thought put into the
design of the user interface for the project. This influences
the usability of the system. Most systems also have security
features to ensure only authorized access.
V. CONCLUSION
Future scope for the home automation systems involves
making homes even smarter than today. Homes can be
interfaced with various kinds of sensors such as motion
sensors, light sensors and temperature sensors and provide
automated toggling of devices based on conditions. More
1042 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
energy can be conserved by ensuring occupation of the
house before turning on devices and checking brightness
and turning off lights if not necessary. The system can be
integrated closely with home security solutions to allow
greater control and safety for home owners. The next stage
would be to extend this system to automate a large scale
environment, such as offices and factories.
Based on all the systems surveyed with their advantages
and drawbacks, this paper presents the features to be
possessed by an ideal system which was created for home
automation. By reviewing those research papers which
related to the area of home automation and related works
to that area. Finally the reader can get a brief idea about
what should he do and what shouldn’t do during the
implementation process. Plug and play capabilities might
be an added bonus for the system as a further development.
Ease of adding a new device to an automated house will
play an important role in taking forward the systems
commercially.
This paper mainly concerns about two major areas
which are home automation and report generating. It has
mentioned there are various options. But all these may
not provide a desired accuracy therefore depending on
the problem domain choosing the best option would be of
paramount importance for a successful result.
REFERENCE
Arul, S.B., 2014. Wireless home automation system using
zigbee. Int. J. Sci. Eng. Res. 5, 133.
Azni, M.N., Vellasami, L., Zianal, A.H., Mohammed, F.A.,
Daud, N.M., Vejasegaran, R., Basharudin, N.W., Jusoh, M.,
Azir, K., Kan, P.E., 2016. Home automation system with
android application, in: Electronic Design (ICED), 2016
3rd International Conference on. IEEE, pp. 299–303.
Cyril Jose, A., Malekian, R., 2015. Smart Home Automation
Security: A Literature Review. Smart Comput. Rev.
doi:10.6029/smartcr.2015.04.004
Delgado, A.R., Picking, R., Grout, V., 2006. Remote-
controlled home automation systems with different
network technologies.
Ebeid, E., Heick, R., Jacobsen, R.H., 2016. Presenting
user behavior from main meter data, in: Smart Grid
Communications (SmartGridComm), 2016 IEEE
International Conference on. IEEE, pp. 594–599.
 ISBN number: 978-955-0301-41-6
ElShafee, A., Hamed, K.A., 2012. Design and
implementation of a WIFI based home automation
system. World Acad. Sci. Eng. Technol. 68, 2177–2180.
Fabbri, G., Medaglia, C.M., Sbordone, D., Di Pietra, B.,
2016. A tool for the analysis of energy systems in Smart
Cities, in: Industrial Electronics (ISIE), 2016 IEEE 25th
International Symposium on. IEEE, pp. 968–973.
Flessner, J., Frenken, M., 2016. Self-configuring
home automation networks based on psychophysical
principles, in: Control, Electronics, Renewable Energy
and Communications (ICCEREC), 2016 International
Conference on. IEEE, pp. 105–109.
Frincu, M., Draghici, R., 2016. Towards a scalable cloud
enabled smart home automation architecture for demand
response, in: PES Innovative Smart Grid Technologies
Conference Europe (ISGT-Europe), 2016 IEEE. IEEE, pp.
1–6.
Gomez, C., Paradells, J., 2010. Wireless home automation
networks: A survey of architectures and technologies.
IEEE Commun. Mag. 48.
Haddon, L., 1995. Home automation: Research issues.
Hasan, R., Khan, M.M., Ashek, A., Rumpa, I.J., 2015.
Microcontroller Based Home Security System with
GSM Technology. Open J. Saf. Sci. Technol. 05, 55–62.
doi:10.4236/ojsst.2015.52007
Krishna, I., Lavanya, K., 2017. Intelligent Home
Automation System using BitVoicer, in: Intelligent Systems
and Control (ISCO), 2017 11th International Conference
on. IEEE, pp. 14–20.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Matlak, S., Bogdan, R., n.d. Reducing Energy
Consumption in Home Automation based on STM32F407
Microcontroller.
Naresh, D., Chakradhar, B., Krishnaveni, S., 2013.
Bluetooth Based Home Automation and Security System
Using ARM9. Int. J. Eng. Trends Technol. IJETT–Volume
4, 4052.
Noorwali, A., Rao, R., Shami, A., 2016. Wireless home area
networks in smart grids: modelling and delay analysis, in:
Smart Grid (SASG), 2016 Saudi Arabia. IEEE, pp. 1–7.
Palaniappan, S., Hariharan, N., Kesh, N.T., Vidhyalakshimi,
S., others, 2015. Home Automation Systems-A Study. Int.
J. Comput. Appl. 116.
Patchava, V., Kandala, H.B., Babu, P.R., 2015. A Smart
Home Automation technique with Raspberry Pi using
IoT, in: Smart Sensors and Systems (IC-SSS), International
Conference on. IEEE, pp. 1–4.
Soumya, S., Chavali, M., Gupta, S., Rao, N., 2016. Internet
of Things based home automation system, in: Recent
Trends in Electronics, Information & Communication
Technology (RTEICT), IEEE International Conference on.
IEEE, pp. 848–850.
Teymourzadeh, R., Salah Addin Ahmed, Kok Wai
Chan, Mok Vee Hoong, 2013. Smart GSM based Home
Automation System. IEEE, pp. 306–309. doi:10.1109/
SPC.2013.6735152
Vasilevskiy, A., Morin, B., Haugen, Ø., Evensen, P., 2016.
Agile development of home automation system with
ThingML, in: Industrial Informatics (INDIN), 2016 IEEE
14th International Conference on. IEEE, pp. 337–344.
10TH INTERNATIONAL RESEARCH CONFERENCE 1043
PROCEEDINGS
COMMUNICABLE DISEASE
SURVEILLANCE AND RESPONSE
SYSTEM
AAA Chinthaka1#, LRK Kahandagamage1,
DGS Amaranath1and STS De Silva1
1
Centre for Research and Development – Ministry of Defence
# aminda@crd.lk
Abstract - The latest studies estimate in Sri Lanka around
5 million people living in areas of dengue risk, over 32,000
infections so far in this year and more than 50 deaths.
Though government of Sri Lanka has spent millions,
precise system not yet been instigated. It is learned that
the Public Health Inspectors (PHI) are the key people to
be addressed to implement proper communicable disease
surveillance system. Present manual system retard the
entire process of reporting patients and breeding locations
of mosquitos and other relevant information and never
been adopted a centralized mechanism. Awareness
and the active participation of the community is a vital
factor to successes such a system. Hence this system is a
participatory Web-Mobile and GIS based system which
developed to address burden of dengue and other similar
epidemic disseises in Sri Lanka. This system enriched
with latest technologies in mobile applications, Web and
Geographic Information Systems. This system is capable
of monitor trends in the distribution and spread of dengue
over time geographically for early response also possible
to report dengue breeding locations and patient’s locations
with various related data using mobile applications further
automated alerting system has developed for PHIs and
related authorities based on community mapping. Drone
mapping is used to map dengue risk areas, Water logged
1044 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
places, malfunctioning drainage systems, garbage dumps,
and/or any potential endemic areas will be marked as
risk areas with the use of aerial photos and videos. Live
monitoring of the dengue risk areas using drone is a vital
factor hence this system get the live streams from the high
flying drones in-order to monitor present condition of the
risk areas. Hotspot analysis and various statistics reports
enhance the risks of individual’s surroundings.
Keywords - Epidemic diseases, Web GIS, Drone Mapping,
Mobile Applications
I. INTRODUCTION
The lack of planning, inadequate housing, water, sewage,
and waste management has created ideal conditions for
dengue viruses and other epidemic diseases. The latest
studies estimate in Sri Lanka around 5 million people
living in areas of risk, over 32,000 infections so far in this
year, and more than 50 deaths. Thus, ignored for many
years, only recently has the potential magnitude of the
dengue problem been acknowledged by policymakers
and funding agencies. Government of Sri Lanka has
spent millions to developed dengue reporting, analysis
and prediction systems, thus never been able to reduce
 ISBN number: 978-955-0301-41-6
the impact of the dengue due to burden of practical
applicability and lack of community participation. The
aim of the research is to develop web-GIS based system
to reduce the impact of the epidemic diseases and to
improve the efficiency of the Public Health Inspectors
(PHI) by introducing digital formats and also to increase
the participation of the community.
II. METHODOLOGY
This system consist of two major web map engines, in-
house (CRD) map engine and the ArcGIS server, images
and the live video from drone will be stored in the CRD
server while inputs from the mobile application store in
both CRD server and the ArcGIS server. Public Health
Inspectors (PHI) has to login with their user names and
password to the same mobile application which use by the
general public. If there is any dengue suspicion area general
public can report those areas using mobile application
which is highly user-friendly application. PHI’s can
report the risk areas as well as dengue patient locations.
Drone mapping use to map the dengue risk areas using
aerial photography and live stream for live monitoring
of risk areas. When dengue breeding location reported
by the community warning message will be sent to the
PHI and relevant government authorities using ArcGIS
Modelbuilder and Python scripts. Web application and
a mobile application developed to report dengue patient
who are admitted to the government or a private hospital
by digitizing current manual procedures. General public
can access to the web page and perform various analysis
depends on their requirements.
FIGURE 1 - METHODOLOGY
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
III. EXPERIMENTAL DESIGN
This research involved developing two different Android
applications to address the aforesaid need. One Android
application to report the dengue breeding locations,
container type, etc… and to send data to MySQL database
with the X, Y coordinates through HTTP header. The
other Android application developed to reports dengue
patient’s information.
Php timer script running on Apache server will access to
the store in My SQL and automatically store data in MS
SQL Server Express. Incident’s locations stored in MS SQL
Server Express will be accessed through ArcGIS 10.1 in-
order to analyse the risk and to send warning messages
to the inhabitants who would possibly be affected with
the calculated risk factors (High Risk Zone, Medium
Risk Zone and Low Risk Zone) through e-mails and SMS
and also SMS alert will be sent to the responsible PHI of
the dengue reported area. This entire process has been
automated using “Model Builder” in ArcGIS to increase
the speed of the process and to reduce human errors.
For assigning calculated risk for individuals Visual Basic
Script has been used and two separate python scripts have
been used within the ArcGIS “Model Builder” to send
emails and SMS. Furthermore the system will send detail
report to the relevant government organizations through
e-mail and SMS.
Figure 2 - Experimental design
10TH INTERNATIONAL RESEARCH CONFERENCE 1045
PROCEEDINGS
IV. RESULTS AND DISCUSSION
This system been tested in Dehiwala and Mount Lavinia
areas during past two years and conducted awareness
programs for the PHIs and other relevant authorities.
As per the interviews with the PHIs this system is a vital
requirement and reduced the complexity for their duties
Figure 3–Main page
In the main page of the web system users can view the
current threat of the dengue fever for individual district
according to the number of reported patients and breeding
locations. This thematic colour map changing automatically
whenever new record is submitted to the system and also
hotspot analysis can be seen which generated through the
ArcGIS Server. Dengue analysis based on the Geography
always provides better understanding, thus this system use
ArcGIS server to analyse and display data.
Figure 4–Statistical charts
Statistical analysis charts are developed by considering
breeding locations, patients and the rainfall which can be
use for the prediction and analysis.
Figure 5–Andriod inputs and temporal analysis
1046 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Android mobile application is able to send data directly to
the ArcGIS server which is one of the significant research
finding in this research. Users also can view the breeding
map and the temporal analysis of spreading dengue patient
reported locations.
Figure 6 - Drone mapping
Drone mapping been used to map dengue risk areas.
Water logged places, malfunctioning drainage systems,
garbage dumps, and/or any potential endemic area marked
as risk areas with the use of aerial photos and videos. Live
monitoring of the dengue risk areas using drone is a vital
factor hence this system get the live streams from the
high flying drones in-order to monitor present condition
of the risk areas and system displays the current position
of the drone over the Google map using X, Y coordinates
receiving by the GPS of the drone.
V. CONCLUSION
GIS play vital role in early warning systems, disaster
management process and in epidemic diseases outbreaks.
With the development of the new technology there are
many options to be adopted and customize into our
own systems. Selecting best and most suitable option is a
vital factor since warning systems and epidemic analysis
is dealing with human life and the valuable assets of the
country.
REFERENCE
Aloudat A and atina K. (2011). The application of location
based services in national emergency warning systems:
SMS, Cell broadcast services and beyond. Research Online
- University of Wollongon.
Phaiju A., Bej D, Pokharel S., Dons U. (2010). Establishing
Community Based Early Warning System. Kathmandu,
Nepal: Mercy Corps Nepal.
Rogers D and Tsirkunov V. (2011). Implementing Hazard
Early Warning Systems. Global Facility for Disaster
Reduction and Recovery (GFDRR).
 ISBN number: 978-955-0301-41-6
Anderson E. (2015). Android developers. Retrieved
February 12, 2015, from developer.android.com: http://
developer.android.com/about/index.html
Earl, K. (2013, May 10). How to Get Better GPS Accuracy
with Your Rugged Handheld. Retrieved May 23, 2015,
from blog.junipersys.com: http://blog.junipersys.com/
how-to-get-better-gps-accuracy/
Grasso, V. F. Early Warning Systems. United Nations
Environment Programme (UNEP).
Schryen G and Wex F. Risk Reduction in Natural Disaster
Management. Germany: University of Regensburg.
Opadeyi, J. (2009). Research on the Application of
Geographic Information Systems . CDEMA.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Spalding, V. (2014). Automating Spatial Processes.
Retrieved 04 21, 2015, from http://www.esri.com/: http://
www.esri.com/news/arcuser/0700/mbover.html
Richter S and Hammitzsch M. (2013). Development of
an Android App for notification and reporting of natural.
Potsdam, Germany: EGU General Assembly 2013.
Poole, I. (2014). Radio Electronics.com. Retrieved from
satellite Technology: GPS Accuracy, Errors & Precision:
http://www.radio-electronics.com/info/satellite/gps/
accuracy-errors-precision.php
Prevention, C. f. (2014). Centers for Disease Control
and Prevention. Retrieved from School Violence: Data
& Statistics: http://www.cdc.gov/violenceprevention/
youthviolence/schoolviolence/data_stats.html
10TH INTERNATIONAL RESEARCH CONFERENCE 1047
PROCEEDINGS ISBN number: 978-955-0301-41-6

DIGITAL CERTIFICATE MANAGEMENT

SYSTEM FOR eHEALTH AND
mHEALTH PRACTITIONERS IN
SRI LANKA TO SECURE
MEDICAL DATA
TMKK Jinasena1#, RGN Meegama1 and
RB Marasinghe2
1
Department of Computer Science, University of Sri Jayewardenepura,
2
Department of Medical Education and Health,
University of Sri Jayewardenepura,
# kasunkosala@yahoo.com

Abstract - eHealth and mHealth systems are getting more attended for the test and their responses were critically
popular today; yet, vulnerabilities are much higher when analysed. Compare to the conventional command line
the sensitive medical data being transferred through based method, 100% of user satisfaction has been gained
public networks. Therefore, it is essential to have a by the developed tool. In conclusion, it is a simple, free and
digital identification and authentication mechanism to open source software for the public to secure their digital
authenticate peers in a digital world. Especially, it will data.
help to avoid attacks such as man-in-the-middle attack.
Although the digital certificates can solve this issue, it Keywords - Computer Security, Digital Certificate, PKI,
has not been used by the general public yet to protect eHealth & mHealth.
their digital data. This is mainly due to their limited
knowledge in IT and the complexity of the process. Thus,
it is required to have a simple security tool to support
I. INTRODUCTION
encryption, digital signature, digital authentication,
Health and mHealth systems are getting more popular
and integrity verification. However, we have developed
today; yet, vulnerabilities are much higher when the
a digital certificate management system to facilitate all
sensitive medical data being transferred through public
these features including creating asymmetric key pairs,
networks. As the IBM survey in 2015, medical data has
generating, signing, chaining and revoking certificates,
the highest per record security cost. Moreover, in USA,
and signing and verifying digital contents. Because it is
medical identity theft is the most rising crime today. All
a Java based application, it is platform independent; thus
these emphasize the criticalness of security of medical
portable. In backend, it uses OpenSSL library. Moreover,
data. However, identity cards play an important role in
it is capable of managing present RSA based certificates
human society by providing a convenient way to prove
as well as the novel Elliptic Curve (EC) based certificates.
the identity of a person to others. In the same way, digital
Thus, it is more robust, future-proof and well-suited for
certificates provide a way to prove the identity of a person
mobile devices. However, a usability test was performed
or a device uniquely in the digital world. As more and more
to evaluate its usability, efficiency and the effectiveness. 47
devices getting connected to public networks with the
undergraduate and postgraduate students were voluntarily
advancement of Mobile technology, Internet, and Internet

1048 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
of Things (IoT), it has become a crucial requirement to
identify them and their activities uniquely. Especially,
it is essential to avoid man-in-the-middle attacks. At
present, digital certificates are being used to identify
mainly servers, not the clients. This is because of the
complexity of Issuing, distributing, renewing, revoking,
and validating certificates. As a result neither servers nor
the peers would be able to verify the real originator of
the message. Thus, it would be not possible to implement
security services such as non-repudiation. However, if
one obtained a digital certificate, he/she has to pay a
fee for the certificate. Therefore, the communities with
thousands of user like universities, hospitals, government
departments, agencies, etc. have to spend a huge amount
of money for this annually. If an organization wants to
identify its clients, it needs to have specialized people
to manage those certificates. Besides, organizations can
not have local authority on their identification system if
they use the global verification chain as external parties
can create valid certificates for them. Moreover, present
encryption mechanisms are not suited to secure real-
time video data in a mobile environment due to its time
complexity. Therefore, it is necessary to have a simple
tool to manage digital certificates within an organization
especially in a non-technical environment like health.
The objective of this research is to develop a simple but
novel and robust digital certificate management system
for eHealth and mHealth practitioners in Sri Lanka to
manage their medical data securely (Abdalla et al., 2000;
Al Nuaimi et al., n.d.; Andrews et al., n.d.; Armknecht et
al., n.d.; Brands, n.d.; Gerard et al., n.d.; Gui-hong et al.,
n.d.; Dahlman, 2014; Zhou & al, 2007)
II. BACKGROUND
When a client visits a web site which has a certificate
installed in the server, the web browser can verify the
validity of that certificate through its certificate authorities.
Especially, this is very helpful to avoid DNS attacks when
you are visiting sites like email, internet banking or
payment sites like PayPal. In fact, digital certificates help
you to identify any site before you enter your sensitive data
such as passwords or pin numbers to that site or before
you download software from it.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Figure 1: Certificate authority hierarchy
Thawte, GlobalSign, Combo, and DigiCert are some of
the root certificate authorities who provide this service.
However, there are number of intermediate certificate
authorities in the world who are extending the service
of those root certificate authorities. Practically it is not
possible to installed signatures of all certificate authorities
in a web browser. Thus, chain validation is being used. In
chain validation, browser repeats the validation process
till it gets the signature of a known trusted certificate
authority (Nash et al., n.d.; Hunter et al., n.d.; Koehler,
n.d.; Leavitt, n.d.). Figure 1 shows how the certificate
authority hierarchy is organized.
III. METHODOLOGY
A free and open source library for cryptography named
OpenSSL has been using in the backend to do the all
cryptographic works. Thus it will automatically fix the
bugs as you update the library. In order to make it platform
independent, Java has been chosen as the development
language. For the simplicity of the user, tasks have been
divided into tabs so that user can focus on a single task at
a time. Tabs have been organized in a way that it takes user
from one step to the other as in wizards. Further, defaults
values have been given to minimize the user inputs as well
as give insight about the value that required for that field.
Moreover, users can customize these values and save their
own default values for later use.
Main steps of the process can be summarized as key pair
generation, certificate sign request generation, signing
a certificate, viewing a certificate, verifying a certificate,
chaining certificate authorities, generating revoke list,
and generating renew request. Apart from that, message
digest generation, signing documents and verifying digital
signatures can also be performed.
10TH INTERNATIONAL RESEARCH CONFERENCE 1049
PROCEEDINGS
A. Creating a Digital Certificate
Figure 2 shows how the digital certificate is created by
putting CA’s signature for user’s details and his/her public
key.
Figure 2: CA and Signing Digital Certificates
B. Generating a Digital Signature
Figure 3: Signing a message
Figure 3 shows how the digital signature is generated using
the message digest and the private key of the sender. This
will provide a way for receiver to verify the integrity of the
message and to authenticate the sender of the message.
C. Verifying the Integrity and Sender
Figure 4 shows how that verification process and
authentication are happening.
Figure 4: Verifying a Signed Message
1050 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Finally, a usability study was performed with the help of
university students. First, they were given command line
OpenSSL and a list of commands to execute. Secondly, they
were asked to do the same task using the tool developed by
this research. Thirdly, they have been asked to rate their
experiences of both methods and finally, their feedbacks
were critically analysis using statistical methods.
Figure 5: Key Generation
IV. RESULTS AND DISCUSSION
Present global public key infrastructure is based on
RSA asymmetric cryptography system. However, in this
research we have used a novel and robust cryptography
method called Elliptic curve cryptography. Moreover, it is
efficient than the present RSA method as it can provide
the same security under 30 times less key size. Hence,
it is suitable for mobile devices too. Elliptic curves are
exponential. Besides, it is based on new mathematics on
finite field rather than the integer factorisation. Thus, it is
more robust than the present RSA method. Therefore, this
public key infrastructure is more robust, future-proof and
well-suited for mobile devices.
Figure 5 shows how the key pair and parameter files being
created. Figure 6 shows how to sign a certificate using a
root or an intermediate certificate. Besides, it allows you to
create self sign certificates as well as root and intermediate
certificates. Initially it was decided to carry out the usability
test with the help of medical students by considering their
higher IT knowledge compare to Doctors and other health
workers. However, in the pre-test, it was found that most
of the medical students were very uncomfortable with
the command line although they have a relatively good
knowledge in IT. Therefore, the computer science and IT
undergraduate and post graduate students were chosen
for the final test in order to compare user experiences of
both. 47 students were voluntarily attended on the test.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Most of them were familiar with the OpenSSL. Thus they
were able to do all the command line tasks successfully
with miner mistakes. Results show that 100% of the
testers were voted this as a convenient, efficient, and less
error prone tool for the public key management. In pre-
test, some of the experts favour command line tool rather
than this because it allows more freedom for customizing
commands. However, the objective of this research is to
empower general public with user friendly tool for public
key cryptography and infrastructure management. Thus,
the res
Gui-hong, L., Hua, Z. & Gui-zhi, L., n.d. Building a secure
web server based on OpenSSL and apache. In E-Business
and E-Government
Hunter, T.B., Weinstein, R.S. & Krupinski, E.A., n.d. State
medical licensure for telemedicine and teleradiology.
21(4), pp.315-18.
Koehler, S.C., n.d. Method and system for authenticating
digital certificates issued by an authentication hierarchy.
Google Patents. 00149.

Leavitt, N., n.d. Internet security under attack: The

undermining of digital certificates. 44(12), pp.17-20.
00052.

Nash, A., Duane, W. & Joseph, C., n.d. PKI: Implementing

and Managing E-security.
Zhou, Y. & al, e., 2007. Access control in wireless sensor
networks. Ad Hoc Networks, 5(1), pp.3-13.

Figure 6: Signing Certificates

REFERENCES

Abdalla, M., Shavitt, Y. & Wool, A., 2000. Key management

for restricted multicast using broadcast encryption.
Networking, IEEE/ACM Transactions on, 8(4), pp.443-54.

Al Nuaimi, N., AlShamsi, A., Mohamed, N. & Al-Jaroodi,

J., n.d. e-Health cloud implementation issues and efforts.
In Industrial Engineering and Operations Management (
Andrews, R.F., Williams, P. & Lin, J., n.d. Risk management
for public key management infrastructure using digital
certificates. Google Patents. 00128.
Armknecht, F. et al., n.d. An efficient implementation
of trusted channels based on OpenSSL. In Proceedings
of the 3rd Brands, S.A., n.d. Rethinking public key
infrastructures and digital certificates: building in privacy.
Mit Press. 00768.

Dahlman, E.a.M.G.a.P.S.a.e.a., 2014. 5G wireless access:

requirements and realization. IEEE Communications
Magazine, 12(52), pp.42--47.

Gerard, P. et al., n.d. Cybersecurity in radiology: access

of public hot spots and public Wi-Fi and prevention of
cybercrimes and HIPAA violations. 201(6), pp.1186-89.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1051
PROCEEDINGS ISBN number: 978-955-0301-41-6

BUILDING DIGITALLY INCLUSIVE DIFFERENTLY

ABLED COMMUNITY IN SRI LANKA:
AN INFORMATION AND
COMMUNICATION TECHNOLOGY (ICT)
POLICY FRAMEWORK
N Wedasinghe1#, N Sirisoma1, and APR Wicramarachchi2
1
General Sir John Kotelawela University, Ratmalana, Sri Lanka
2
University of Kelaniya, Sri Lanka
# niroshaw@hotmail.com

Abstract - Digital inclusion (DI) is an ability of individual I.INTRODUCTION

and groups to access and use information communication
technology. In the Sri Lankan context, it is a problem
among disabled people. This study investigates influencing
factors of digital inclusion in Sri Lankan disabled
community. Research methodology consists of a survey
done at Ranaviru apparels Yakkala, Sri Lanka in the phase
one of this study. It interviewed 180 disabled people who
work at this institute. Descriptive statistic method was
used for data analysis. According to results, current website
development issues, difficulties in software accessibility,
human perception on technology, financial difficulties
to purchase and use of equipment indicated as major
issues. A multidimensional framework is recommended
to overcome issues. Standardization of web and internet
facilities, concentration on user friendliness in software
development process, infrastructure development, social
and cultural changers, financial support for disabled
people and special training and education on technology
are recommended in the framework. This framework can
be considered as a tool for building digitally inclusive
differently abled community in Sri Lanka
Keywords - Disability Digital Divide, Digital inclusion,
A. Background to the Study
Accessibility issues for disabled people are a pertinent
issue take into consideration in the modern globalised
world. In Sri Lanka also this is a influencing topic because
many people suffered due to road accident and three
decade of civil war. In Sri Lankan civilian who was born
without any physical disability became disabled because
of civil war in the country. This adds a severe disadvantage
for the Nation. Due to the disability issue many people are
getting disadvantage of accessing equal opportunities in
the country. This is specially happen when they are using
new technologies in their office work and maintaining
social relationships and accessing public infrastructure.
County policy makers are considering overcoming these
issues and ensuring that all citizens are enabling access to
new digital technologies.
Objective of this study is to propose a framework to
overcome issues which influence for digital inclusion
among Sri Lankan differently abled community.
This research basically divided into three different
research questions. First question is to finds out what are

1052 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

the main factors influencing for digital inclusion among Digital Divide
differently abled community. Second question is to find The exclusion of people with disability from the online
out how to overcome the above identified problems and world has been referred to as one important component
issues. Third question is to finds out how to validate the of the “digital divide” ( Dobransky & Hargittai, 2006;
propose solutions. Warschauer, 2003).People with disabilities are considered
likely to be able to make especially fruitful use of the
Outcome of the research propose a framework and online world to help overcome disabling barriers they face
validate method to overcome the digital inclusion issues caused by societal attitudes, organisation, and structuring
in Sri Lanka. which mean that their differences, for instance physical,
sensory, intellectual or psychological impairments, are
II. LITERATURE REVIEW not adequately considered and so they are discriminated
against. For example: a person with considerable physical
impairments can pursue online education meaning there
Definitions
is less need to leave home; a person with a significant
Research on digital inequality studies how different
visual impairment can gain access to documents by
social groups access technologies and how this access
downloading them and converting the text to speech; a
contributes to offline advantages and disadvantages
person with a learning disability can socialise and make or
(Chen, 2013). The use of technology to communicate
maintain friendships from home. Sadly though, disabled
has become an essential and socially acceptable aspect of
people, because of poverty, lack of social support or other
most people’s lives and it is becoming increasingly difficult
reasons, frequently lack the means to get online and if
to distinguish between the “digital world” and the “real
they can, may not be adequately equipped or supported (
world” ( Helsper, 2008; Ritchie & Blanck, 2003). Hence,
Chadwick et al., 2013; Hoppestad, 2013).
Digital Inclusion is an increasingly important social issue,
reflecting imperatives, opportunities, and considerations
According to the Roger et.al(anon) Digital divide is a
about human rights, equity, issues of identity, language,
latest evocative term that refers to differences in access to
social participation, community and civic engagement,
uses of information technology that are correlated with
and opportunity pertaining to the digital world ( Castells,
income, race and ethnicity, gender, age, place of residence,
1997; Warschauer, 2003).
and other measures of socioe-conomic status. According
to them some people have the most powerful computers,
Disability
the best telephone service and fastest Internet service, as
According to the Charloton (1998) and Driedger (1989)
well as a wealth of content and training relevant to their
study disability is complex, dynamic, multidimensional,
lives. Another group of people do not have access to the
and contested. Over recent decades, the disabled people’s
newest and best computers, the most reliable telephone
movement together with world report on disability
service or the fastest or most convenient Internet services.
numerous researchers from the social and health sciences–
The difference between these two groups is called the
have identified the role of social and physical barriers in
Digital Divide.
disability (Barnes, 1991). More recent thinking around
disability indicates that the identification of deficits should
Digital inclusion
be integral to the identification of necessary support
Crandall and Fisher (2009) suggest digital inclusion is
people require to overcome these challenges and that
the rallying cry of the twenty-first century. They claim
people’s strengths should also be highlighted ( Schalock et
that digital inclusion goes beyond access to computers
al., 2010).
and the internet for all, regardless of physical, cognitive,
or financial ability; it means technological literacy and
According to the Sen(2009) Disability is a development
the ability to access relevant online content and services.
issue, because of its bidirectional link to poverty: disability
Hache and Cullen (2009) extend the definition by arguing
may increase the risk of poverty, and poverty may
that digital inclusion is the process of democratization of
increase the risk of disability. This disadvantage group of
access to Information and Communication Technology
community also face some difficulties of access computers,
(ICT) in order to allow for the inclusion of the marginalized
mobiles and Internet. It causes for many reasons such as
in society. They claim that digital inclusion should be seen
poverty, lack of awareness, language issues, Social and
as a wagon to social inclusion that ensures individuals
ethical background and lack of Interest.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1053
PROCEEDINGS ISBN number: 978-955-0301-41-6

and disadvantaged groups have access to the skills to use were conducted. Participants for this study were selected
ICTs, further indicating these individuals will be able again from different Age, Type of Work, Gender, Living
to participate in and benefit from electronic-mediated, arrangement, Employment status, Management Level and
growing knowledge within an information society. Income.

III. METHODOLOGY Both of these initial study result focuses to model the
solution framework and propose validation method in
this study.
This research examines the factor need to consider digital
inclusion among differently abled community. Therefore C. Data Analysis Framework
literature survey has conducted in the initial stage of
the research. Then it is finding out how to overcome the Table 1: Different type of digital inclusion
problems and finally it discusses how to implement the among differently abled community in
solutions and validate them. The methods used to collect Sri Lanka
and analyse data are discussed below.
Type of Accessi- Variable Description
Notation
A. Research Strategy bility
B1 Basic Telephone Use
This research consists of basically a literature study. The Basic Web Access – Information
core construction of this model are adapted from the B2 Search and Email use
theory of Technology Acceptance Model(TAM).Data B3 Basic Computer Use
Sample collected for this study include with literature
I1 Intermediate Office application Use
behind the disability digital divide in the globe, social
I2 Intermediate Multimedia use
construction of disability, the development of information
technology and internet related technology through a I3 Intermediate Screen reader Software use
literature survey and global and Sri Lankan corporate I4 Intermediate Social media use
policies related to information technology and web A1 Advance IT Administration
accessibility through literature and Interviews had with
A2 Advance S/W Design Development
public and private sector organization. In order to identify related
real requirement of differently able community in relation
A3 Advance High-tech S/W Use (Eg. CAD/
to Information Technology, interviews were conducted CAM Operations)
with a range of disabled people who effected from Sri
Lankan civil war. In the second survey has been more focuses on in-depth
analysis of the factors influence for digital inclusion. The
B. Data Collection main factors divided into three main dimensions. They are
perception on technology, technological factors and other
Data collected in this study via an observation and inhibiting factors. They are given in the below Table 2.
interviews and literature study. Researcher has model the
solution basically initial survey conducted at Sri Lanka Table 2: Factors influencing for digital
Army CLI, unit Panagoda. Structured interviews were inclusion
conducted. This study interviewed 314 differently abled
people using a non-probability sampling method such as Perception on
Technology
Technical problems and
issues
Other inhibiting factors

convenience sampling. Age group of the participants was Income(PI) web standard issues(TW) Cost of Software Technology(OC)
between 20 to 50 years old. Participants for this study were Education(PE) Software incompatibility/ Lack of Awareness(OLA)
user friendliness (TS)
selected from different communities, disabled categories
and various geographical locations. Data is collected in Desire for Lack of Direction(OLD)
learning(PL)
under the different levels of technology usage under the
Entertain- Lack of Skills(OLS)
three themes such as basic, intermediate and advance. ment(PEN)
In the second phase of this study conducted at the Culture(PC) Lack of resources(OLR)
Ranaviru Apparel Yakkala Branch. Study conducted Lac of Training(OLT)
with 180 disabled employees. Structured interviews

1054 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Collected data were analysed based on qualitative
approach according to thirteen themes. Analysis
techniques used in this research were co-relation on the
above factors on disability digital divide. Factors analysed
based on the three main viewpoints including perception
on technology, Technical problems and issues and other
Inhibit factors.
In order to identify the factors affecting to digital inclusion
factors classify under three viewpoints. The first theme is
perception on technologies. Under this five main factors
were taken into consideration. Such as person income,
education background, personnel interest on learning
new technologies, use of technology to watching films,
listing music and other entertainment activities. Then,
it considers how the person culture effects to disability
digital divide.
Next dimension of the factor analysis focuses on Technical
problems and issues. Basically it discusses web standard
issues and software incompatibility or the friendliness of
the software.
Finally, it concerns the other inhibiting factors which
causes for digital inclusion such as cost of software
technology. Awareness of the technology, Use only some
basic technologies without proper direction. Some people
have proper direction but due to disability they suffer with
have lack of skills. Next it considers the resources issues.
Most of the disabled users are facing with some financial
difficulties. Therefore, most of the assistive technologies
cannot be purchased.
According to the descriptive statistics three main areas
were taken into consideration such as current ICT
knowledge, Relevance of IT for job related activities
and Interest of ICT for future needs. Result of the study
indicated that the level of Current ICT Knowledge is high
among the age 25 to 28. Reason for this could be that
younger generation is using more smart phones and they
are likes to incorporate new technologies. Adaption level
is high among this community.
English language issues among this community can be
considered as a significant issue. 90 percentage of the
participant had English language issues and they are
working in Sinhala language. Therefore, majority are
having language barriers to use technology. Therefore,
most of them are in the medium level . Most of the
computer users are using office package in day today life
at the same time some participants in young age like to use
innovative technology. But employees who are closer to
retirement age do not like to learn new technologies. Their
ambition is to have their pension and stay at home. But
they are interested in giving IT education to their children
rather than learning themselves.
On the other hand according the findings social media
usage is 0% among the community. Restriction of social
media with in the office hours and office computer are
causes for this result. Most of them do not have a computer
at home and as a result less usage of social media among
the community.

IV. RESULT AND DISCUSSION

Descriptive statistics of the sample, the general information
of differently abled community (age, experience in
technology), and the other details are reported in Table 3.

Table 3: Use of Technology

Description Respondent
Telephone Use
Web Access
Computer Use
Office application Use
Multimedia use
Screen reader Software use
Social media use
IT Administration
S/W Design Development
related
High-tech S/W Use
Percentage
252 98 % Figure 01 : Technology inclusion
150 58 %
Factors influencing for digital inclusion
155 60 %
131 51 % In the next survey conducted at the Ranaviru Apparel
28 11 % Yakkala mainly focuses the factors that are influencing
4 0% for digital inclusion among the disabled community.
10 0% Descriptive statistics are discussed in the Table 4.
2 0%
1 0% In this study different type of disabled participant were
interviewed. It was included with 53 people with having
0 0% serious difficulties in eyes, 05 hearing problems, 06 people
with upper Limbs and 06 with lower limbs. In addition to

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1055
PROCEEDINGS ISBN number: 978-955-0301-41-6

that 107 people with different complications of disabilities Basically total responses valve divided by 5 based on the
were involved. All together 177 participants were taken in five point lickert scale. Then it is equally divided into 36,
to this research. 72,108,140 and grater.

Table 4: Descriptive statistics of factors If response rate for a given Factor (F) is less than or equal
influencing for digital inclusion 36 then points given to this factor is 1. If F greater than 37
or less than or equal 72 then points given to this factor is 2.
Age 18-25 25-35 36-45 >45 Then If F greater than 73 or less than or equal 108 points
given to this factor is 3. If F greater than 109 or less than or
23 90 58 6 equal 140 points then given to this factor is 4. If F greater
than 141 then this factor is assign by 5 point.
Type of Work Admin Operational
19 158
Language Sinhala Tamil English
Findings are given in the Table 5.
177 - -
Gender Male Female Factor relate to Very im- Im- 50-50 Less Not im-
Digital Inclusion portant(5) port- (3) import- portant(1)
175 2 ant(4) ant(2)

Leaving Ar- With Patents 112 Income(PI) 3

rangements
Shared accommodation 7 Education(PE) 3
with friends
Desire for learn- 3
With Partner 41 ing(PL)

Living alone 1
Entertain- 3
In a rehabilitation centre 15 ment(PEN)
Culture(PC) 5
Healing Home for War 1
heroes
web standard 2
Employment Full Time 177 issues(TW)
Status

Software incom- 3
patibility/user
friendliness (TS)
Middle 55
Cost of Software 1
Low 122 Technology(OC)

Income Lack of Aware- 1

Less than Rs. 15,000 0
ness(OLA)
Rs. 15,000 – Rs. 25,000 5
Lack of Direc- 2
Rs. 25,001 – Rs. 35,000 71 tion(OLD)
Rs. 35,001 – Rs. 45,000 89 Lack of 3
Skills(OLS)
Rs. 45,001 – Rs. 60,000 12
Lack of re- 3
➢ 60,001
0 sources(OLR)
Lack of Train- 5
ing(OLT)

According to the analysis, factors that are influencing

digital inclusion among disabled community are Based on the above finding it is indicated that different
categorised into five point likert scale. According to the factors are affecting in different level.
responses given basically categorised the response rates in
to 5 categories according to the following method.

1056 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Overcoming Digital inclusion factors
The second research question is to find out a suitable
method to solve the issues. Based on the literature and
survey findings, solution framework describe below figure
2.
Figure 2: digital inclusion solution framework
1. Infrastructure Consideration
Smart objects connected through radio signals, Bluetooth
and wireless connectivity and Internet of Things (IOT)
base Solutions. In addition to that one main factor that
needs to improve the digital inclusion among the disabled
community is the telecommunication infrastructure
changers.
When the roads are newly constructed, implementing
obstacle detection will help blind people to detect obstacle.
Object interconnected and improves the usability. Web
base solutions for disabled are need higher bandwidth
special on accessing computers and internet. Community
centres are far away and difficult to travel for disable
people.
2. Technology Consideration
Promote low cost technologies are one main solution to
overcome the digital inclusion. Invest to development on
low cost solution and promote research and development
on low cost technology solutions
Promote Language friendly Applications. Develop and test
the language base solutions such as text to voice converter
in Sinhala and Tamil language.
Mobile technology considerations are another finding
to make disable people incisive on digital technologies.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Follow the mobile accessibility guidelines. Design mobile
accessibility guidelines (Ex. BBC guidelines for mobile
development for disabled)
Computer technology considerations are another
main factor. Computer technology considerations
divide according to the barriers faced by the disabled
community. These barriers can be grouped into three
functional categories: barriers to providing computer
input, interpreting output, and reading supporting
documentation.
3. Training and Development
Special Design IT training for middle age people will be
helpful for digital inclusion. Implementing a disabled
training module is another solution. Special Design IT
training for old people needed to consider. Specific user
oriented training for old disable users
4. Web Standards
Promote W3C guide lines among web developers. Identify
the problems of current implementations
Awareness on the W3Cguidelines
5. Government intervention and policy
implementation
Technology barriers. Study the technological requirement
of disabled community and identify the policy issues
related to technology
6. Social, Cultural and economic consideration
Social and cultural influences directly correlate with
digital inclusion. Influencing by Media could be helpful
to make disabled people digitally inclusive. In order to
socially interaction differently abled people need to be
digitally inclusive. Promote social inter connection with
relatives, friend and communicate with similar groups via
social media
7. Financial support for disabled community
Financial difficulties among the disabled community
make them digitally exclusive. Special loan system to buy
smart and computer devises and provide special low cost
solutions will make them more digitally inclusive.
10TH INTERNATIONAL RESEARCH CONFERENCE 1057
PROCEEDINGS
V. CONCLUSION AND
RECOMMENDATION
According to the above two surveys conducted for this
research, indicated that digital exclusion is an issue in
Sri Lankan differently abled community. Therefore, it is
recommended a solution framework for policymakers
to consider. It is recommended to validate this solution
frame work. Recommended validation are given in the
following Table 6.
Table 6: Validation framework
Solution to overcome disability digital
divide
When the roads are newly constructed,
implementing obstacle detection will
help blind people to detect obstacle.
Object interconnected and improve the
usability
Web base solutions for disabled are
need higher bandwidth special on
accessing computers and internet. Com-
munity centers are far away and difficult
to travel for disable people
Invest to development on low cost solu-
tion and promote research and develop-
ment on low cost technology solutions
Develop and test the language base
solutions .Text to voice converter in Sin-
hala and Tamil language.
Follow the mobile accessibility guide-
lines
Design mobile accessibility guidelines
(Eg. BBC guidelines for mobile develop-
ment for disabled)
Computer technology considerations
divide according to the barriers faced by
the disabled community.
These barriers can be grouped into
three functional categories: barriers to
providing computer input, interpreting
output, and reading supporting docu-
mentation.
Implementing a disabled training
module
Specific user oriented training for old
disable users
Identify the problems of current imple-
mentations
Awareness on the W3Cguidelines
1058 10TH INTERNATIONAL RESEARCH CONFERENCE
Validation method
Study the case studies on different
implementations has been done in
other countries.
Conducing a survey to identify the
technology adoption between disable
people who are accessing high
speed bandwidth and not
Conduct a survey to identify the
technology adaption differences
between the people who are using
low cost technologies and high cost
technologies.
Get the feedback of the user satis-
faction to prove the factor
Conducting a comparative analysis
to identify the effectiveness of the
guidelines for disable people. (Pre
implementation and post implemen-
tation will be validated)
Design an assessment evaluation
forms to get the user satisfaction of
implementations
Survey result to identify the user sat-
isfaction of the training and usability
Survey result to identify the user sat-
isfaction of the training and usability
Survey on disabled user satisfaction
on before and after implementation
of W3C standards for a web site
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Study the technological requirement of
disabled community and identify the
Propose sample policy changers by
analysis of other case studies using in
policy issues related to technology
other countries.
Influencing by Media Develop a video’s to present the im-
portance of IT for disabled community.
Conducting a survey to measure the
improvement of usage by the technolo-
gy after awareness through media
Promote social inter connection with Create a social media group and
relatives, friend and communicate observe the connectivity among the
with similar groups via social media disabled community
Special loan system to buy smart and Survey on financially capable and
computer devises incapable disabled community usage.
Special low cost solutions Monitor and assess the improvement
of the technology adoption after pro-
viding financial support
REFERENCES
Crandall, M., Fisher, K. E. (2009). Digital Inclusion:
Measuring the Impact of Information and Community
Technology. Medford, New Jersey: Information Today,
Inc.
Chen, W. (2013). The implications of social capital for the
digital divides in America. The Information Society, 29,
13–25. doi:10.1080/01972243.2012.739265
Chadwick, D. D., Quinn, S., Fullwood, C., & Chen-Wilson,
J. (2015). Public perceptions of the risks and benefits of
being online for people with an intellectual disability. 13th
Biannual Research Conference of the Nordic Network on
Disability Research, Bergen, Norway, 6–8 May, 2015.
Dobransky, K., & Hargittai, E. (2006). The disability divide
in internet access and use. Information, Communnication
and Society, (Special Issue: Disability, Identity, and
Interdependence: ICTs and New Social Forms), 9(3),
313–334.
Helsper, E. (2008) Digital inclusion: an analysis of social
disadvantage and the information society. Department for
Communities and Local Government, London, UK.
Hoppestad, B. S. (2013). Current perspective
regarding adults with intellectual and developmental
disabilities accessing computer technology. Disability &
Rehabilitation: Assistive Technology, 8(3), 190–194.
Ritchie, H., & Blank, P. (2003). The promise of the Internet
for disability: A study of on-line services and web site
accessibility at centers for independent living. Behavioral
Sciences & the Law, 21(1), 5–26.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Roger G. Noll, Older-Aguilar. D, Rosston G and Richard Sen A.(2009) The idea of justice. Cambridge, The Belknap
R. R(Anon), the digital divide: definitions, measurement, Press of Harvard University Press
and policy issues [online] Http://faculty.chicagobooth.
edu/austan.goolsbee/teaching/digdiv.pdf Warschauer, M. (2003). Technology and social inclusion:
Rethinking the digital divide. Cambridge, MA: MIT Press.
Schalock, R. L., Borthwick-Duffy, S. A., Bradley, M. et al.
(2010). Intellectual disability: definition, classification,
and systems of supports (11th ed). Washington: American
Association on Intellectual and Developmental Disabilities

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1059
PROCEEDINGS ISBN number: 978-955-0301-41-6

AN IMAGE PROCESSING APPLICATION

FOR DIAGNOSING ACUTE
LYMPHOBLASTIC LEUKEMIA (ALL)
JKC Shyalika1#, PPNV Kumara1, and
Darshana Kottahachchi2
Faculty of Computing,
1

1
General Sir John Kotelawala Defence University, Sri Lanka
s
Faculty of Allied Health Sciences,
2
General Sir John Kotelawala Defence University, Sri Lanka
# chathurangijks@gmail.com

Abstract - Leukemia, simply called “Blood cancer”
is a fatal disease where the white blood cells (WBC)
increases in bone marrow and peripheral blood. Acute
lymphoblastic leukemia (ALL) is one of the most
common types of leukemia aroused by accumulation and
overproduction of immature and cancerous cells known
as lymphoblasts. Presently, the diagnosis of ALL includes
performing a full blood count, blood picture, bone marrow
biopsy, cytochemical stain, immunophenotyping and
cytogenetics. These techniques are highly tedious, costly,
requires expertise of hematologists and available only in
few hospitals. Therefore, as an alternative, use of image
processing to diagnose ALL would become an effective
solution. Although, several research groups have employed
image processing to identify ALL, recognition and
splitting of overlapping red blood cells (RBC) with WBC
has yet been a challenging issue. This paper is about an
application which includes an image processing algorithm
to diagnose ALL while attempting to solve the above
mentioned issue of overlapping cells. The algorithm is also
extended to detect the quality devastation in blood films in
terms of storing them for prolonged period. The inputs for
this application are microscopic peripheral blood images
of ALL patients obtained from Department of Pathology
Clinic at Faculty of Medicine, University of Colombo.
Then, image processing techniques; image enhancement,
segmentation, feature extraction and classification are
performed. For the detection and diagnosis of leukemia,
segmentation using morphological operations in OpenCV
Python and classification using K-Nearest Neighbour
and Support vector machine implementations has been
proposed in this research. It is observed that the proposed
algorithm has led to a high accuracy in diagnosing ALL.
The system also includes a PHP based web application
that serves hematologists, doctors and patients to log in
to their specific user accounts and make records, insert
details and view diagnosing reports.
Keywords - Acute Lymphoblastic Leukemia, Image
processing, Segmentation and Feature extraction,
Classification
I. INTRODUCTION
Blood is one of the most important materials of the
human body as it is the principle agent that make
humans live. Human blood consists of two major parts;
plasma and cells. The peripheral blood cells consist of
three main components; Red blood cells (RBC), White
blood cells (WBC) and platelets. There are five types of
WBCs; Neutrophils (40-75%), Lymphocytes (20-45%),

1060 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Eosinophils (1-8%), Monocytes (0-10%), Basophils (0.5-
1%). Leukemia, simply called “Blood cancer” in which
usually the number of WBC increase in the bone marrow
and peripheral blood. These leukemic cells (usually
immature) replace the normal blood cells and causes
malfunction of the bone marrow and peripheral blood.
Furthermore, excess amount of these cells travels to
other sites such as liver, spleen to maintain normal cells
production. Later, the leukemia cells also invade other
organs causing them to malfunction (Hoffbrand et al,
2004).
There are two main types of leukemia according to the
morphology of cells in the bone marrow. They termed as
acute & chronic Leukemia. Acute leukemia involves the
rapid overgrowth of very immature blood cells whereas
chronic leukemia involves the overgrowth of somewhat
mature blood cells in the bone marrow compared to acute
type. In the French-American-British (FAB) classification
(Bennett et. al, 1976), acute leukemia is further categorized
into two groups based on the white blood cell from
which the malignancy originates from. They are acute
lymphoblastic leukemia (ALL) is caused by abnormal
lymphoid cells, and acute myeloid leukemia (AML) is
caused by abnormal myeloid cells in the bone marrow
(Hoffbrand et al, 2004). The predominant abnormal cells
in the ALL are lymphoblasts.
The detection, identification and classification of leukemia
normally follow series of steps. They are full blood count,
examination of blood picture, bone marrow & trephine
biopsy, immunophenotyping and cytogenetics (Hughes-
Jones et. al, 2004; Hoffbrand and Lewis 1995; Purohit, 2000;
Cui et. al, 2004). The whole process takes about 3-4 days
and also needs a well-trained experienced staff. However,
early diagnosis of leukemia contributes to early treatment
and proper management of patients. Furthermore,
manual detection procedure stated above is a highly
tedious task that involves the effort of hematologists and
other supporting staff as it is intensively slow, costly, time
consuming. Even though advanced techniques are being
used there may be errors especially diagnosing subtypes.
Image processing and data mining fields have provided
fast, cost effective and accurate solutions in fields such
as medical image management, image data mining,
bioimaging, neuroimaging and virtual reality in medical
visualization (Scholl et al., 2011; Hegadi, 2010). Researches
have been conducted for the detection and counting of
red blood cells, white blood cells and to diagnose diseases
like anaemia, malaria and deficiency of vitamin B12 using
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
blood images (Vaghela et al., 2015). Furthermore, Image
Processing techniques are the used in detecting cancer
cells (Patil and Jain, 2014). Image processing is also been
used in diagnosing leukemic cells in blood samples.
Techniques such as Image acquisition, pre-processing,
segmentation, Feature extraction and classification have
been used in the diagnosis.
II. LITERATURE REVIEW
Presently, a considerable contribution has been done
by researchers in the aim of ALL detection using image
processing. The common flow of the image processing
techniques that is used in diagnosis can be illustrated by
below figure.
Figure 1. Flow of ALL diagnosis
The researches done so far are different from the
segmentation methods and classification methods they
have used. A review done on image processing techniques
have been elaborated in this section.
A. Segmentation methods
1) Watershed segmentation: “Watershed” refers to a
ridge that splits areas drained by different river systems.
Watershed lines are defined on the nodes, edges, hybrid
lines on both nodes and edges and in continuous domain.
Watershed segmentation is an easy method for the
detection of WBC but requires best quality images in
order to achieve a better accuracy (MAHAJA et al., 2014).
2) Fuzzy C Means Clustering (FCM): This data clustering
technique groups a dataset into n clusters with all data
points in that dataset belong to each and every cluster to
a certain degree. FCM result is much accurate and it’s able
to measuring nucleus boundaries with shape, colour and
texture, but it’s difficult in classification of lymphoblast in
to its sub types through this segmentation (MAHAJA et
al., 2014; Viswanathan, 2015).
10TH INTERNATIONAL RESEARCH CONFERENCE 1061
PROCEEDINGS
3) Fuzzy K-Means Clustering in L*A*B* colour space:
K-Means method is a least squares partitioning method
and it divides a collection of objects to K groups of
clusters. It considers each object have a location in the
space and finds partitions in the image such that objects
within each cluster close to each other as likely, and as far
from the objects in other clusters as possible. This method
is not applicable on incremental data and it cannot give
classification with labelled data (Vaghela et al., 2015).
Mohapatra and the colleagues have used fuzzy based
blood image segmentation for separate out leucocytes
from other blood components (Mohapatra et al., 2011).
4) Otsu’s method: This is a thresholding method and
it’s the easiest and fastest method used in segmentation.
Thresholding is based on a clip-level named a threshold
value used in converting a grayscale image into a binary
image. Fabio Scotti has used Otsu’s method in nucleus
and cytoplasm selection in lymphoblasts and lymphocytes
(Scotti, 2005). Their experiments have showed a good
performance of this method in separating the nucleus
from the cytoplasm.
5) Shadowed C-means clustering (SCM): SCM is a
method of partitive clustering developed in the framework
of shadowed sets. Unlike rough clustering, in SCM, the
choice of threshold parameter is fully automated and
the number of clusters is optimized in terms of various
validity indices (Mitra et al., 2010). Shadowed clustering
can handle overlapping among clusters efficiently and also
it can model uncertainty in class boundaries (Mohapatra
et al., 2014). The algorithm is robust in the presence of
outliers too. However, fuzzy c-means clustering have
problems with high dimensional data sets and a large
number of prototypes (Oliveira and Pedrycz, 2007).
6) HSI Colour based Segmentation: HSI (Hue, Saturation,
Intensity) is a common colour model used in image
segmentation. HSI colour model has a good capability of
representing the colours of human perception (H. D.Cheng
et al., 2006). Nor Hazlyna and the team have conducted a
research for ALL detection based on segmentation using
HSI and RGB colour space (Nor Hazlyna et al., 2010).
The results have shown that the proposed segmentation
technique based on HSI has successfully segmented
the acute leukemia images while preserving significant
features and removing background noise. Singhal and
Singh (Singhal and Singh, 2014) and Halim and his
colleagues (Halim et al., 2011) are some research groups
who have used HSI colour based segmentation in ALL
1062 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
diagnosis. They have used HSI colour based segmentation
as it provides better performance than RGB colour
segmentation.
7) K-means clustering: K-means clustering is an
unsupervised learning algorithm which involves two
simple processes as relegating the given data set and
classifying the colligated data sets to the centroid nearest
to them. K-means clustering segmentation have been
used in identifying the leukemia sub types (Fatma and
Sharma, 2014; Ashwini Rejintal and Aswini.N, 2016) and
in AML screening systems (Agaian et al., 2014). K-means
clustering does not give classification with labelled data
and also not applicable on incremental data (Vaghela et
al., 2015).
8) Morphological Operations (shape-based): Segmentation
using morphological operations is a technique considering
the processing of geometrical structure based on set
hypothesis, topology, lattice hypothesis and arbitrary
functions etc. This is the most successful segmentation
method that has been used so far. Through this method it
is very easy for detecting white cells, overlapping of cells
and shape of cells. Thus this is based on statistics so can
get approximate results.
Bhattacharjee and Saini (Bhattacharjee and Saini, 2015),
Vaghela and the group (Vaghela et al., 2015) and Raje and
Rangole (Raje and Rangole, 2014) are some researchers
who have used morphological based image segmentation
in leukemia diagnosis. They have discovered that the
morphological operators used for the extraction of
features have resulted in high segmentation accuracy.
Segmentation using morphological operations has been
used in morphological classification of Leucocytes by
microscopic images (Scotti, 2005; Piuri and Scotti, 2004).
There, the researchers have focused on reducing the
problem of identification and classification of WBC types
in microscope images using morphological operations.
B. Classification Methods
Classification is in charge of assigning to the unknown
test vector which is a label from one of the known classes
(Rege and Dr.Gawli, 2015). Mostly used classifiers are as
follows.
1) Support Vector Machine (SVM): SVM is a discriminative
classifier that is formally defined by a separating hyper
plane. When labelled training data is given (supervised
 ISBN number: 978-955-0301-41-6
learning), the algorithm outputs an optimal hyperplane
which categorizes new examples. Patel and Mishra (Patel
and Mishra, 2015) is a research group who presented
an automatic approach for leukemia detection using
microscopic images. Colour, geometric, shape and
statistical features have been analysed and classified
under the SVM classifier in the intention of grouping the
normal and abnormal cells. SVM has been used to classify
leukemia types too. A three-layered framework consists
of feature extraction, coding, and classification for the
detection of leukemia from blood smear images has been
proposed by Faivdullah and his colleagues (Faivdullah et
al., 2015) leukemia types. They have employed a one-vs-all
technique to convert SVM, which is a binary classifier in
to a multi-class classifier.
2) Artificial Neural Network (ANN): This is a statistical
learning algorithm defined by an interconnected set of
nodes that are similar to the network of neurons found
in brain. ANNs are capable of pattern recognition and
machine learning, thus is mainly used in generating
and estimating the output from a large number of input
data set (Bhattacharjee and Saini, 2015). Mohapatra and
the colleagues (Mohapatra et al., 2012) have engaged in
another project in Lymphocyte image segmentation using
Functional Link Neural Architecture for ALL detection
(Mohapatra et al., 2012). Fatma and Sharma (Fatma and
Sharma, 2014) have tried on a system to identify and
classify sub types of acute leukemia using neural network.
3) CART (Classification and Regression Trees): CART
(Classification and Regression Trees) statistical method
has been used in automatic leukemia diagnosis in
investigating the classification power of cell markers
extracted in segmentation (Serbouti et al., 1991). This
method generates classification tree diagrams with
complete splitting information at each node and then
produces a classification matrix, splitting cost and
probability matrix for both the learning sample and the
cross validation. The classification trees can be saved and
used in classifying unknown specimens. Serbouti and the
research team has employed CART in their research done
in automatic leukemia diagnosis (Serbouti et al., 1991).
4) K-Nearest Neighbour (KNN): This is considered to
be the best classifier in the family of nonparametric
method with a good scalability. In leukemia detection
kNN=1 is considered to classify between blast cells and
normal lymphocytic cells (Bhattacharjee and Saini, 2015).
Bhattacharjee and Saini (Bhattacharjee and Saini, 2015)
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
in their research in diagnosing ALL have discovered that
KNN is the best classifier that produced high specificity
and also have the lowest computational complexity which
has produced a specificity of 95.23%.
5) Ensemble of Classifiers (EOC): Ensemble methods
are machine learning algorithms that construct a set of
classifiers and then classify new data points by taking a
weighted vote of their predictions (Dietterich, 2011). EOC
improves of the performance of individual classifiers. The
ultimate goal of classification result integration algorithms
is to generate more certain, precise and accurate system
results. But EOC possess some limitations also such as
increased storage, increase the number of computations
and decreased comprehensibility. EOC is been an
efficient classification model used in leukemia diagnosis
so far. An ensemble classifier system for early diagnosis
of ALL has been developed by Mohapatra and group in
2014 (Mohapatra et al., 2014). As the results they have
obtained more accuracy in EOC in comparison with other
classifiers employed. Scotti and Piuri (Piuri and Scotti,
2004) have used ensemble of classifiers on their research
done in Morphological Classification of Blood Leucocytes
by Microscope Images. The classification accuracy has
been tested and a proper classifier has been chosen from a
set of candidates of different classifiers.
III. METHODOLOGY AND
EXPERIMENTAL DESIGN
An automated Acute Lymphoblastic Leukemia diagnosing
application would be a useful tool in diagnosing of Acute
Lymphoblastic Leukemia in blood samples efficiently and
accurately. The basic method for diagnosis can be divided
into following steps.
Figure 2. Basic diagram of the system
10TH INTERNATIONAL RESEARCH CONFERENCE 1063
PROCEEDINGS
The input for the system is the Leishman’s stained blood
slide image, and the lymphocytes in the image are cropped
and individuated manually. Firstly in the pre-processing
module, the image acquisition noise and background non-
uniformities are removed. Secondly image segmentation is
performed using a proper segmentation technique. This is
done using three consecutive steps of background removal,
separating the lymphocytic cell and separating the nucleus
region which has been described in this paper. In the
feature extraction module, various morphological features
are been sorted differently using the segmented regions
of the lymphocytic cell and the nucleus. Combining the
features of both cell and nucleus, some new features are
also calculated. In the lymphocyte classification module
the tested cells are labelled as blast or normal.
A. Image Acquisition
The inputs for this automation process are microscopic
images obtained from peripheral blood films which stained
by Leishman’s that has been obtained from Department
of Pathology Clinic at Faculty of Medicine, University
of Colombo. All the obtained images are affected from
B-ALL precursor which is a major type of ALL. The
images are captured from two different camera sources
as Huawei GR5 2017 smartphone camera and Canon
camera in the same lightning conditions, resolution and
magnification. The slides are placed under a MicroTech
XSZ-N207 microscope in 100 magnification. 75 of the
chosen images are taken into the experiment. Microscopic
images acquired from the ALL-IDB (Acute Lymphoblastic
Leukemia Image Database) are used as standard. 50
images chosen from ALL-IDB2 database are used.
Figure 3. Block diagram of the proposed algorithm
1064 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
B. Image pre-processing
Pre-processing is essential as normal images consist of
excessive staining and shadows. Image enhancement,
which is used to bring out the image details that are
obscured is the main task of this stage. Following three
main tasks are performed in this stage.
1) Fast NlM eans Denoising Coloured technique: This is
done to remove noise and excess blurriness.
2) Edge enhancement: Done in order to sharpen the image
by cleaning the cell/cell segments in the boundary of the
image.
3) RGB splitting: The RGB image is split in to three
channels; green, red and blue in order to identify the
red blood cells and white blood cells separately.
4) Removing the green channel: Green channels are mostly
sensitive to red blood cells. Thus, it is removed from the
image in this step.
C. Image segmentation
Segmentation process partitions an image to its
constituent segments or objects known as pixels. This
locates objects and boundaries (curves and lines etc.) of
images and modifies the representation of an image into
somewhat that is more meaningful and easier to analyse.
This is a crucial step as the following feature extraction
and classification results are much related with the result
of the segmentation module. In this stage the following
three steps are under gone.
1) Background removal: In this stage, canny filter is first
used to reconstruct the border of the cells present in the
image. Then morphological operation ‘dilation’ is done
using a prepared structuring element. Then ‘closing’
is done. Combing the images obtained from dilation
and closing, a new image is obtained. Next, threshold
to Zero, Inverted thresholding is performed to the
image obtained from pre-processing. Then the resulted
image is combined with the new image obtained from
morphological segmentation and the background is
now removed.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

2) Isolating the lymphocytic cell: In the resulted image, the E. Image classification
largest contour area is considered to be the area of the
cell region. The image is then subjected to a combination For classification, K-Nearest Neighbour classifier is used
of binary thresholding and Otsu’s thresholding and a
binary image of the cell is produced. The total blood
cell’s binary image is now ready for feature extraction.

3) Isolating the nucleus: In this step, firstly the intensity

of the original cropped blood image is increased such
that only the nucleus will be visible in the image. Then
thresholding is done in order to separate the nucleus.
Here a combination of binary thresholding and Otsu’s
thresholding is done. Then the nucleus region is Figure 4. ALL Diagnosing Window
segmented by subjecting the image to the background
removal step described earlier. Then the segmented
nucleus is converted to binary and it is now ready for
IV. RESULTS
feature extraction.
There are two main objectives that this research is focused
on. One is to develop an algorithm to compare the image
D. Feature extraction
qualities of purely isolated cells and overlapping cells.
Second one is to detect the quality devastation in blood
In feature extraction, the acquired data from the image
films in terms of storing them for prolonged period.
is transformed and labelled to a particular set of features,
The proposed algorithm was implemented using python
which is going to be used for further classification. The
programming language using the OpenCV package for
binary equivalent images produced by the segmentation
python. Following figures depicts a design of user interface
technique of blood cell and cell nucleus are used to extract
and the results that were obtained in the segmentation,
those morphological features. Using the extracted features
feature extraction and classification stages separately.
of blood cells and nucleus, some combined features
also have been calculated. Some of the features that are
explored are given below

Table 1. Parameters Obtained In

Feature Extraction
Feature Parameters Extracted
Colour features Mean colour values
Geometric features Perimeter, Radius, Area,
Compactness, Rectangularity,
Convexity, Symmetry, Concavity,
Elongation, Eccentricity, Solidity
etc. Figure 5. Segmentation results of the cell
Texture features Entropy, Energy, Correlation,
Homogeneity etc. Figure 5 elaborates the results of a lymphocyte image used
Statistical features Skewness, Variance, Mean, Gra- for segmentation stage of the cell. In the figure firstly,
dient matrix etc. original image is shown. The pre-processed image shows
the result of the steps FastNlMeansDenoisingColoured
technique, Edge enhancement, RGB Splitting and
removing the green channel respectively. Next canny,
dilation, closing and combination of dilated and closed
images, thresholding for pre-processed image and binary
image of the cell has been shown.

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1065
PROCEEDINGS
Figure 6. Segmentation results of the nucleus
Figure 6 elaborates the results of the same lymphocyte
image used for segmentation stage of the nucleus. In the
figure firstly, original image is shown. Then the thresholding
mask image shows the result of the combination of the
binary and Otsu’s segmentation performed. Next canny,
dilation, closing and combination of dilated and closed
images, thresholding for pre-processed image and binary
image of the cell has been shown.
In the feature extraction module, 12 geometric features
with regard to the cell and nucleus had been extracted
separately. The features extracted are area, perimeter,
circularity, diameter, roundness, compactness, nucleus
to cell ratio, cytoplasm area, cytoplasm to nucleus ratio,
convexity and solidity. The features had been saved in
a csv file. Those features extracted with respect to a
one lymphocytic cell are depicted in Figure 7. In the
classification stage, the features have been fed to the
K-Nearest Neighbour classifier. Out of the 75 images used,
35 images have been used for the training the classifier and
40 for testing. The label 1 has been outputted for detected
acute leukemic cells and label 0 has been outputted for
healthy cells respectively as shown in Figure 7.
Figure 6. Segmentation results of the nucleus
1066 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
The proposed algorithm is been tested and developed
to detect features of overlapping cells and to detect the
quality devastation that occurs in old blood films when
they are kept for 3-6 months.
V. CONCLUSION AND
FURTHER WORK
The developing system will be extended to have good
results in automatic diagnosis of the disease in the acquired
human blood samples. The proposed algorithm can also
be further develop to detect the granules and intra cellular
components inside the cell. As per the statistical data
published in future spreading of a cancer like leukemia
in the world, automation procedures to detect leukemia
has become an urgent need. Governments, especially
in a developing country like Sri Lanka would find these
automatic leukemia detection systems as cost effective
solutions to implement in hospitals.
ACKNOWLEDGMENT
Authors would like to acknowledge the supervisors Mr.
PPNV Kumara and Dr. Darshana Kottahachchi for their
massive guidance, valuable advices, immense assistance
and support throughout the research. Special thanks goes
to all the lecturers of Faculty of Computing and staff of
Department of Medical Laboratory Sciences, Faculty of
Allied Health Sciences, whose dedicated support for the
project was intensely appreciated.
REFERENCES
Agaian, S., Madhukar, M., Chronopoulos, A.T., 2014.
Automated screening system for acute myelogenous
leukemia detection in blood microscopic images. IEEE
Syst. J. 8, 995–1004.
Ashwini Rejintal, Aswini.N, 2016. Leukemia Cancer Cell
Detection using Image Processing.
Bhattacharjee, R., Saini, L.M., 2015. Robust technique
for the detection of Acute Lymphoblastic Leukemia, in:
2015 IEEE Power, Communication and Information
Technology Conference (PCITC). IEEE, pp. 657–662.
 ISBN number: 978-955-0301-41-6
Dietterich, T.G., 2011. Ensemble Methods in Machine
Learning.
Faivdullah, L., Azahar, F., Htike, Z.Z., Naing, W.Y.N., 2015.
Leukemia Detection from Blood Smears. J. Med. Bioeng.
4, 488–491. doi:10.12720/jomb.4.6.488-491
Fatma, M., Sharma, J., 2014. Identification and classification
of acute leukemia using neural network, in: Medical
Imaging, M-Health and Emerging Communication
Systems (MedCom), 2014 International Conference on.
IEEE, pp. 142–145.
Halim, N.A., Mashor, M.Y., Nasir, A.A., Mokhtar, N.R.,
Rosline, H., 2011. Nucleus segmentation technique for
acute leukemia, in: Signal Processing and Its Applications
(CSPA), 2011 IEEE 7th International Colloquium on.
IEEE, pp. 192–197.
H. D.Cheng, X. H.Jiang, Y.Sun, Jing Li Wang, 2006. Color
Image Segmentation: Advances & Prospects.
Hegadi, R.S., 2010. Image Processing: Research
Opportunities and Challenges. NSRC Dec.
MAHAJA, S., GOLAIT, S.S., MESHRAM, A., JICHLKAN,
N., 2014. REVIEW: DETECTION OF TYPES OF ACUTE
LEUKEMIA.
Mitra, S., Pedrycz, W., Barman, B., 2010. Shadowed
c-means: Integrating fuzzy and rough clustering. Pattern
Recognit. 43, 1282–1291. doi:10.1016/j.patcog.2009.09.029
Mohapatra, S., Patra, D., Kumar, S., Satpathy, S., 2012.
Lymphocyte image segmentation using functional link
neural architecture for acute leukemia detection. Biomed.
Eng. Lett. 2, 100–110. doi:10.1007/s13534-012-0056-9
Mohapatra, S., Patra, D., Satpathy, S., 2014. An ensemble
classifier system for early diagnosis of acute lymphoblastic
leukemia in blood microscopic images. Neural Comput.
Appl. 24, 1887–1904. doi:10.1007/s00521-013-1438-3
Mohapatra, S., Samanta, S.S., Patra, D., Satpathi, S., 2011.
Fuzzy based blood image segmentation for automated
leukemia detection, in: Devices and Communications
(ICDeCom), 2011 International Conference on. IEEE, pp.
1–5.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Nor Hazlyna, H., Mashor, M.Y., Mokhtar, N.R., Aimi
Salihah, A.N., Hassan, R., Raof, R.A.A., Osman, M.K.,
2010. Comparison of acute leukemia Image segmentation
using HSI and RGB color space. IEEE, pp. 749–752.
doi:10.1109/ISSPA.2010.5605410
Oliveira, J.V. de, Pedrycz, W. (Eds.), 2007. Advances in
fuzzy clustering and its applications. Wiley, Chichester.
Patel, N., Mishra, A., 2015. Automated Leukaemia
Detection Using Microscopic Images. Procedia Comput.
Sci. 58, 635–642. doi:10.1016/j.procs.2015.08.082
Patil, B.G., Jain, S.N., 2014. Cancer Cells Detection Using
Digital Image Processing Methods. Int. J. Latest Trends
Eng. Technol. IJLTET 3.
Piuri, V., Scotti, F., 2004. Morphological classification of
blood leucocytes by microscope images, in: Computational
Intelligence for Measurement Systems and Applications,
2004. CIMSA. 2004 IEEE International Conference on.
IEEE, pp. 103–108.
Raje, C., Rangole, J., 2014. Detection of Leukemia
in microscopic images using image processing, in:
Communications and Signal Processing (ICCSP), 2014
International Conference on. IEEE, pp. 255–259.
Scholl, I., Aach, T., Deserno, T.M., Kuhlen, T., 2011.
Challenges of medical image processing. Comput. Sci. -
Res. Dev. 26, 5–13. doi:10.1007/s00450-010-0146-9
Scotti, F., 2005. Automatic morphological analysis for acute
leukemia identification in peripheral blood microscope
images, in: 2005 IEEE International Conference on
Computational Intelligence for Measurement Systems and
Applications. pp. 96–101.
Serbouti, S., Duhamel, A., Harms, H., Gunzer, U., Aus,
U.M., Mary, J.Y., Beuscart, R., 1991. Image segmentation
and classification methods to detect leukemias, in:
Engineering in Medicine and Biology Society, 1991.
Vol. 13: 1991., Proceedings of the Annual International
Conference of the IEEE. IEEE, pp. 260–261.
Singhal, V., Singh, P., 2014. Local Binary Pattern for
automatic detection of Acute Lymphoblastic Leukemia,
in: Communications (NCC), 2014 Twentieth National
Conference on. IEEE, pp. 1–5.
10TH INTERNATIONAL RESEARCH CONFERENCE 1067
PROCEEDINGS
Vaghela, H.P., Modi, H., Pandya, M., Pandya, M., 2015.
Leukemia Detection using Digital Image Processing
Techniques. IJAIS–ISSN 2249-0868 Vol. 10.
Viswanathan, P., 2015. Fuzzy C Means Detection of
Leukemia Based on Morphological Contour Segmentation.
Procedia Comput. Sci. 58, 84–90. doi:10.1016/j.
procs.2015.08.017
Hoffbrand, A.V. and Lewis, S.D. (1995) Postgraduate
Haematology (Second edition).
1068 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Bennett, J.M., Catovsky, D., Daniel, M.T., Flandrin,
G., Galton, D.A., Gralnic, H.R., Sultan, C. (1976) Br J
Hematol; 33: 451-8.
Purohit, S.S. (2000) Biotechnology Fundamentals and
Applications (Third Edition).
Cui, J., wang, J., He, K., Jin B., Wang, H., Li, W., Kang, L.,
Hu., Li, M. H., Yu, M., Shen, B., Wang, G. and Zhang, X.
(2004) Proteomic analysis of Human Acute Leukemia
Cells: Insight into Their Classification, Clinical Cancer
Research Vol. 10, 6887 – 6896, October 15.
Hughes-Jones, N.C., Wickramasinghe, S.N., Hatton, C.
(2004) Lecture notes on Haematology (Seventh Edition).
 ISBN number: 978-955-0301-41-6
CONCEPTUAL APPROACH TOWARDS
STATEFUL COMPUTATION
OFFLOADING IN RESOURCE
CONSTRAINT ANDROID DEVICES
P Vekneswaran1# and NR Dissanayake1
1
Informatics Institute of Technology, Colombo , Sri Lanka
# prathieshna.2016281@iit.ac.lk
Abstract - Cyber Foraging is an approach to resolve the
performance and resource limitations of small portable
hand-held devices, through offloading heavy processes
to nearby stationary devices called Surrogate Devices,
which have more computation capabilities. The way the
Cyber Foraging is mostly tackled by invoking a remote
method in the Surrogate Device – which contains the
application related information as a pre-requisite – at the
time of local invocation, in a state-less manner. This limits
the possibilities where the Cyber Foraging can be used in
state-full context, which can provide maintenance of state
which will help the developer to transmit state of objects
used inside the offload candidate so it won’t differ from
local execution and remote execution. In order to use in
a stateful manner, researchers have used a virtualisation
approach, which is a resource consuming approach. As an
alternative, we propose a conceptual solution to transmit
state from mobile platform to the surrogate platform,
without having to deploy a mobile OS virtual machine
into the surrogate environment. The concept will address
all the drawbacks of virtualisation and provide the similar
benefits at a lower resource cost in the Surrogate end.
Keywords - Cyber Foraging, Distributed Computing,
State-full Offloading
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
I. INTRODUCTION
This section gives a brief background to the domain
cyber foraging, specifying the problem we are focusing
on and the motivation towards proposing a conceptual
approach towards stateful computation offloading and the
methodology used for the research. Followed by existing
work, proposed approaches, limitations, conclusions and
future work.
a. Background
1) Cyber Foraging: Since then there has been various
attempts to fulfil Mark Weiser’s vision (Weiser, 1991);
however, these mobile devices are coupled with some
limitations like limited processing power and battery
standby. Cyber Foraging, which was a term coined by M.
Satyanarayanan, is an attempt to offload process from a
portable resource constraint platforms of mobile devices
to a stationary resource rich platform, called a surrogate.
Following his approach there have been various attempts
towards a perfect cyber foraging solution, which is both
efficient and developer friendly.
2) Offloading Approaches used in Cyber Foraging and
their Limitations: There are two computation offloading
approaches available for cyber foraging: 1) State-less
computation offloading and 2) State-full computation
offloading.
10TH INTERNATIONAL RESEARCH CONFERENCE 1069
PROCEEDINGS
State-less offloading is where the surrogate is not aware of
the state of the application or objects in the heap memory.
Surrogate will execute the offloaded method in a method
level where the parameters are passed and method will be
executed in a static context. Most of the cyber foraging
approaches do not maintain state during the offload, as a
consequence granularity of the off loadable component is
limited to the method.
There are only very few significant approaches, which
support state-full offloading. They use virtualisation,
where the mobile operating system is replicated in a remote
location and the mobile device and the surrogate share
the application state among each other, which is costly
on the surrogate end, making it unattractive towards the
surrogate owner. This also creates a dependency between
surrogate and the mobile device.
In order to avoid these adverse impacts, we propose a
state-full offloading approach, which does not create
unnecessary dependencies and does not involve
Virtualisation.
b. Methodology
The existing approaches and their specifications were
identified from a literature survey. Almost 57 preliminary
researches has been conducted in the domain of enabling
cyber foraging (Lewis & Lago, 2015). The survey published
by Lewis & Lago covers majority of the approaches at
the time of their writing. It was helpful in revisiting the
approaches, and also recent approaches had been surveyed
and considered within the scope of this research.
We took the liberty of referring our previous work in the
domain, targeting towards a better practical solution into
consideration, as empirical research. We had experimented
with a state-less architecture, and we are currently focusing
on supporting state-full offloading without virtualisation,
which was one of the major limitation addressed in our
earlier proposed architecture. (P Vekneswaran, 2016)
II. EXISTING WORK
Even though there are significant amount of researches
done in the domain, there is only a handful of approaches
that maintains state while offloading; and all of them are
virtualisation based solutions where maintaining the state
is vital to react upon each invocation.
1070 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
a. AMCO
The mobile application developer declaratively specifies
the suspected energy consumption hotspots in a mobile
application. Based on this input, AMCO then automatically
transforms the application to enable it to offload parts of
its functionality to the cloud. The offloading is exceedingly
versatile, being driven by a runtime framework that
powerfully decides both the state-to-offload and its
exchange instrument in light of the execution condition
set up. In addition, the system continuously improves with
a feedback-loop mechanism.
To mark hotspot components, AMCO provides a Java
annotation @OffloadingCandidate; this information can
also be specified through an XML configuration file. Based
on this input, an analysis engine first checks whether the
specified component can be offloaded as well as any of its
sub-components (i.e. successors in the call graph). The
engine additionally ascertains the program state, to be
exchanged between the remote and local segments that
would should be exchanged to offload the execution of
both the whole segment or of any of its sub-segments. A
bytecode enhancer at that point creates the checkpoints
that spare and re-establish the ascertained state for the
whole hotspot segments and additionally for each of its
subcomponents (Kwon & Tilevich, 2013).
It is easier to develop cyber foraging enabled application
on top of AMCO because it is just a matter of annotating
the offload candidates or migrating the entire processes
to the cloud. But it also adverse impacts such as cost of
sharing the entire state, tracking state modifications also
offloading everything to the could not be beneficial at all
times.
b. ROAM
ROAM is a JAVA application framework that can assist
developers to build Java Based multi-platform applications,
which run on heterogeneous devices; and allows a user to
move/migrate a running application among heterogeneous
devices without any significant effort by the user.
In nutshell, the ROAM agent on the source device first
negotiates with the ROAM agent on the target device.
The negotiation involves exchanges of the target device
capabilities needed by each application component, and
the code base URL where the ROAMlet component byte
code can be downloaded from. Based on the exchanged
 ISBN number: 978-955-0301-41-6
information, the ROAM agent decides the appropriate
adaptation strategy for each component. The ROAM agent
on the target device downloads the necessary ROAMlet
class byte code from the HTTP server for all application
components that will be instantiated on the target device.
The ROAMlet on the source device serializes its execution
state and sends it to the ROAM agent on the target
device. The ROAM agent may perform execution state
transformation if an application component is transformed
or dynamically instantiated. The ROAM agent instantiates
the ROAMlet on the target device. (Hao-hua Chu, 2003)
ROAM doesn’t target computational offloading directly
but does focus on task migration aspect in depth. What
ROAM is trying to achieve is pretty far fetch because it
is attempting to tackle pervasive computing first rather
than Cyber Foraging itself. Server components must
be manually written so according to the device context
the code/service will be retrieved from the server. Even
though it mentions about the heterogeneous devices, they
are targeting Java based operating systems all run on JVM.
So, transforming state is straight forward.
C. Clone-Cloud and Cloudlet
Cloud computing alludes to a style of figuring where online
assets and applications are accessed by clients through
a web browser, yet the real software and data are put on
remote servers. As the client’s point of view, the client does
not have to buy, oversee or maintain the technology “in
the cloud”. Intel’s thought is to take the mobile device that
has lower performance and clone the whole arrangement
of information and applications onto the cloud, at that
point sync the two. It is then likewise conceivable to dole
out graphics and processing power to the task in the cloud
and give an apparently elite with full PC-like experience
to the present smartphone or netbook with no change to
the client experience and interface. (Chun, Ihm, Maniatis,
Naik, & & Patti, 2011)
Clone-Cloud and Cloudlet approaches don’t fall too far
away from each other. Cloudlet is a concept to deploy an
in-box cloud in house, which is self-managed and does not
need any professional attention. A cloudlet is deployed for
few users to share the computing resource via the high-
speed LAN network, rather than using the distant public
cloud on the Internet in order to shorten the latency for
real time application.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Even though it’s easier to develop just like AMCO, Clone-
Cloud requires heavy modifications to the Android Device
kernel to support this infrastructure.
D. Collaborative Applications with Mobile Cloud
This contains an architecture and a paradigm for developing
collaborative applications with minor modifications to
today’s mobile and cloud computing infrastructures.
This approach is focusing on cloning the kernel requests
and responses in the cloud to maintain state. Also has an
offload advisor to evaluate the effectiveness of offloading
and make the choice to offload or not. (Yu-Shuo Chang,
2012)
This also requires hosting and has the same drawbacks as
of Clone Cloud and Cloudlet.
III. PROPOSED APPROACH
In order to enable less developer effort in cyber foraging
enabled mobile applications development, we think that
the assistance provided by the development framework
to the developers is an important fact. (P Vekneswaran,
2016) Aspects (JBoss, 2016) reduce the build process
immensely, by reducing the need to write separate code
for the off loadable components, which will be discussed
in the following section. Also, we made improvements
for achieving computation offloading, to support state
transmission to other VMs.
a. Aspect based Annotation
Candidate methods, the resource intensive components of
the code, recommended for offloading can be scattered all
over the application, thus the application has to invoke the
decision-making engine every time before such method
is invoked. Knowledge of those offload method calls is
simply irrelevant to the business logic in the class. In such
scenario, we propose to use Aspect Orientation in mobile
application code, to identify the compute intensive tasks
all around the application, and while using annotations
won’t interfere with the business logic or the object
orientated design and architecture of the application. (P
Vekneswaran, 2016)
The annotation is the key of our proposed approach, which
helps the developer to specify the identified intense tasks
in the application to state, which affects the execution. In
10TH INTERNATIONAL RESEARCH CONFERENCE 1071
PROCEEDINGS
this approach, there are already predefined advices, which
need to be injected during the compilation phase. Here
the advice is the code that is injected into the class file;
typically, which needs to be inserted before, after, or instead
of the target method. When the method is marked using
the “offloadMethod” annotation, AspectJ will generate the
boilerplates that is necessary for the annotated method.
In our approach, we suggest annotating the potential
offloading candidate methods in classes, using the
“offloadMethod” along with the name of the objects
which states need to be transferred to the Surrogate using
Aspect orientation. Just using one annotation, identifying
different methods with different behaviours is not
possible. Therefore, it is necessary to be able to annotate
a single method using multiple annotations as a part of
the framework we propose; so, the developer can create his
custom defined aspects through creating new annotation
interfaces using default templates provided.
b. Development Process and Limitations
When using the annotations, the developer should identify
the methods and objects that is referenced inside the
context that consists of the compute intensive components
that has be considered by the decision-making engine
during the runtime, weather it needs to be offloaded or
not. This requires the developer to have a basic knowledge
of the application source, states of the objects that is
required for offloading, and the flow of execution.
Identified intensive methods cannot reside inside Android
Activities. These methods should not attempt to read
hardware specific sensor data inside their scope, which
will cause errors during the runtime. Alternative is to
read the sensory data and pass it as a state. The intensive
components should be added to a separate java class for
it to be executed correctly in the surrogate environment.
If the source is already separated this step can be ignored.
These candidate methods should be marked by the
developer by adding annotations along with the object
details, which the state needs to be preserved. During the
build time boilerplate codes will be added by the AspectJ
runtime, which is included in the framework. There are
predefined aspects, which the developer can use, or if the
developer is looking for some distinct characteristics, he
can modify the aspects accordingly. Afterwards developer
can build the application and produce the APK, which is
ready to be installed in the mobile devices.
1072 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
The surrogate service should be running at the time when
the application is about to execute the compute intensive
tasks. According to configuration, the mobile application
will find the surrogate device and attempt to connect to
the service and offload. The surrogate service will then
check if the mobile application is already in its repository
of packages. If not, it will pull the source from the mobile
device. As the components are sent automatically to the
surrogate, there is no need for the developer to manually
deploy the components explicitly. This is a one-time
process, and afterwards any mobile device running the
similar application can offload to that particular surrogate
device, without pulling the source to the surrogate. Once
the source is loaded, the surrogate will execute the task
copying the objects and their states to its heap and return
the result back to the mobile device with the updated heap
state. If it is not beneficial to offload, the mobile device
will execute the task normally without offloading to the
surrogate.
IV. EVALUATION
The proposed solution will have the following advantages
over the existing development approaches.
It eliminates the need for writing two separate codes to
do the same task in the mobile and surrogate to preserve
state, or use virtualisation as an alternative to enable cyber
foraging in new applications. Also, no need to modify the
existing source code, just need to annotate the offload
candidate methods and meta information required to
process the information. This will cut down the time
required by the developer to enable cyber foraging in
mobile applications. The Aspect based approach will
allow the developer to plug and play his own logic in the
decision-making engine, communication protocol etc.
according to his requirements giving him/her full freedom
to customize the outcome. When building the application,
the AspectJ runtime will handle the generation of the
necessary boilerplate codes, which will further cut
down the developer cost. In comparison to the existing
approaches that is discussed in Section 3, the proposed
solution in this paper has majority of the development
process automated and easily configurable through the
build scripts. This approach will also give developers,
enough customisation options as well. Also without
hosting the entire mobile operating system the lightweight
service will compute the JAVA components.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

V. CONCLUSION AND Hao-hua Chu, H. S. (2003). Roam, a seamless application

FUTURE WORK
Even though cyber foraging can be easily achieved
through virtualization, there are other factors such as
functional overhead, which have adverse impact on the
energy consumption and performance. We can conclude
that the offload technique we propose is more effective in a
preliminary level, and reducing the time consumption for
the development by automating most of the build process
with the help of AspectJ and Android Development
Environment. At the moment doesn’t support non-java
based platforms. Also, it will not support rapid state
changes at the moment.
We expect to further improve the proposed approach
and introduce a framework, allowing more common
features like integration to the available IDEs, which
will reduce the developer effort furthermore, supporting
rapid development. Future of this research will extend to
supporting the android code such as activities, fragments
rather than being only able to offload the JAVA components
of the application.
REFERENCES
Balan, Krishna, R., Gergle, D., Satyanarayanan, M., &
Herbsleb., J. (2007). Simplifying cyber foraging for mobile
devices. (pp. 272-285). ACM.
framework. The Journal of Systems and Software .
JBoss. (2016). Chapter 1. What Is Aspect-Oriented
Programming? Retrieved March 30, 2016, from http://
docs.jboss.org/aop/1.0/aspect-framework/userguide/en/
html/what.html
Kwon, Y.-W., & Tilevich, E. (2013). Reducing the Energy
Consumption of Mobile Applications Behind the Scenes.
IEEE.
Lewis, G. A., & Lago, P. (2015). A Catalog of Architectural
Tactics for Cyber-Foraging. ACM.
P Vekneswaran, D. N. (2016). Annotation based Offload
Automation Approach for Cyber Foraging Frameworks.
9th International Reserch Conference. KDU.
Satyanarayanan, M. (2001). Pervasive computing: Vision
and challenges. Personal Communications, IEEE 8.4 (pp.
10-17). IEEE.
Weiser, M. (1991). The computer for the 21st century. (pp.
94-104). Scientific american 265.3.
Yu-Shuo Chang, S.-H. H. (2012). Developing Collaborative
Applications with Mobile Cloud. Journal of Internet
Services and Information Security.

Balan, R., Flinn, J., Satyanarayanan, M., Sinnamohideen,

S., & Yang, H.-I. (2002). The Case for Cyber Foraging.
Proceedings of the 10th workshop on ACM SIGOPS
European workshop (pp. 87-92). ACM.

Chun, B. G., Ihm, S., Maniatis, P., Naik, M., & & Patti,
A. (2011). Clonecloud: elastic execution between mobile
device and cloud. (pp. 301-314). Proceedings of the sixth
conference on Computer systems.

Flinn, Jason, Park, S. Y., & Satyanarayanan, M. (2002).

Balancing performance, energy, and quality in pervasive
computing. (pp. 217-226). IEEE.

Gartner. (2016). Gartner Says Worldwide Smartphone

Sales Grew 9.7 Percent in Fourth Quarter of 2015.
Retrieved April 16, 2016, from http://www.gartner.com/
newsroom/id/3215217

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1073
PROCEEDINGS
TOWARDS EMAIL-BASED WORKFLOW
AUTOMATION
HD Rajapakse1# and
NR Dissanayake1
1
Department of Computing, Informatics Institute of Technology, Colombo, Sri Lanka
# dilee.dev@gmail.com
Abstract - Modern day organisations and individuals use
emails as a communication media and there exist workflows
correlated to the email communication such as responding,
forwarding and more complex workflows such as content
analysis and attachment handling. These workflows are
highly user-dependent due to various requirements such
as uniqueness of users’ fields of occupation. There exist
commercial tools, which are explicitly designed to fulfil
the purpose of email based workflow automation and
management; however, these tools are associated with
email marketing campaign management thus, providing
a set of marketing-centric features. If we can understand
the diverse requirements of the users, it may help in
providing a better solution towards effective email based
workflow automation. Having analysed aforementioned
limitations of prevailing tools, we conducted a survey
towards understanding the requirements of the users well,
towards suggesting an email-based workflow automation
system. We expect to utilize this knowledge to develop
a more customizable solution to satisfy the diverse user
requirements via enhanced decision-making abilities.
Keywords - Email, User requirements, Survey, Workflows,
Workflow Automation
I. INTRODUCTION
There exist several email workflow automation tools and
Platforms, which have different capabilities of automating
processes related to email communication. The majority
1074 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
of these applications are designed to be marketing-centric
thus, providing a set of features, which are highly converged
towards email marketing campaign management and
automation. This itself is a limitation of those platforms
since they are not suitable for prevailing association with
non-marketing intentions. Even though a few of the
aforementioned platforms such as GSuite (Google 2016)
and IFTTT (IFTTT Inc. 2016) look promising towards
customisable email-based workflow automation, there are
certain limitations, which possibly obstruct the usage of
them.
Platforms like GSuite are more focused on the business
process management even though it provides very
preliminary features related to Email-based Workflows
such as auto-responding and clustering Emails. The IFTTT
platform has a comparatively larger number of automation
options in that case. IFTTT’s keyword-based automation
is a fundamental decision-making technique, which has
less flexibility towards handling complex scenarios such as
when there are multiple criteria affecting an automation-
related decision.
The actual user requirements can be much complex and
diverse; therefore, the available platforms may not cater
the features to fulfil them. When user requirements for
automation are complex unlike a simple keyword-based
condition and vary due to various reasons such as user’s
field of occupation, the complexity of the automated
workflow increases. Therefore, enhanced decision-making
abilities and customizability to handle various user
 ISBN number: 978-955-0301-41-6
requirements are needed to satisfy the diversity of the user
requirements. In the direction of doing so, it is important
to understand the nature of actual user requirements and
the limitations of available solutions.
We conducted a cross-sectional survey to derive a sound
understanding of the user requirements in the context of
the email-based workflows and the automation of them
by organising and structuring them in a systematic way.
This study contains results and the analysis of the cross-
sectional survey carried out among the target audience.
Then we review the existing work with the context of the
identified requirements.
The methodology followed in this study is given in section
II. Section III provides the background of the problem
domain along with related subject domains, discussed in
detail to provide an overall understanding of the problem
addressed by this study. After that, in section IV the results
and the analysis of the cross-sectional survey is presented
in detail. Section V reviews the existing work within the
context of the analysis stated in the previous section.
Finally, the study is concluded with an explanation of the
conclusions derived, and also specifying the future work.
II.METHODOLOGY
The methodology used for the survey is specified under
two sections: Literature survey and cross sectional survey.
a. Literature survey
The identified domains related to this study are
workflows, workflow management and automation, email
communication, and decision making concepts. Literature
in aforementioned domains was identified and critically
analysed through a Literature Survey. It was conducted to
derive an understanding of the background of the study,
the approaches, which have been already proposed and
have been implemented by researchers, and to understand
the potential, which existing solutions hold towards
fulfilling the needs related to the email based workflow
automation.
There was a considerable amount of research papers
related to the domains of this study found on the internet
libraries. Google Scholar was the primary search engine
used to search for academic publications available on the
internet and, IEEE Explore Library was used to download
conference and journal publications related to domains.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
The knowledge gained is utilised in discussing the
background in section III and existing work in section V.
b. Cross-sectional Survey
The target population for the cross-sectional survey was
the general public who are using emails and involved in
several fields of occupation. The survey was carried out in
the form of an online questionnaire designed and published
using Google Forms. The online questionnaire was sent to
known contacts via social media websites (Facebook and
Twitter) and was also sent to a randomly selected set of
email addresses, which were gathered from colleagues
working in different industries. The ‘Simple Random
Sampling technique’ was used to select 63 candidates out
of total 70 responses, which were collected within a period
of 14 days. The Successful Response Rate (SRR) of the
survey can be calculated using the equation (1).
The calculated successful response rate suggests that the
survey was successful.
The gathered data was exported to an Excel spreadsheet
using Google Sheets, and statistical information such as
averages, percentages as well as graphical models such
as charts were generated using Microsoft Excel’s inbuilt
formulas and wizards.
III. BACKGROUND
This section discusses the background of the emails,
workflows and workflow automation, and email based
automation gained through the literature survey, to have
a strong understanding towards designing the cross-
sectional survey.
a. Email Communication
Email or electronic mail communication has become a
vital method of communication in the present day and
many organisations, as well as individuals, use email to
collaborate and share information (Zhang et al. 2006). The
idea of Electronic Mail was originated in late 1960’s (Gibbs
2016) and has currently emerged as a swift and reliable way
10TH INTERNATIONAL RESEARCH CONFERENCE 1075
PROCEEDINGS
of communication, which operates through the internet.
The email communication mechanism is based on the
client-server architecture. In a typical scenario, an email
will be sent to a server using a client application, and it will
be forwarded to the corresponding recipient by the Email
server. Originally, email messages were solely based on the
ASCII text format and eventually was prolonged to use
Multipurpose Internet Mail Extensions (MIME), which
allowed to send multimedia content attached to messages
and text character sets other than ASCII (Bonaventure
n.d.). An email message has a simple anatomy. It consists
of a header and a body. The body contains the content of
the message, and the header contains various information
which helps to encode and route the message across the
network.
b. Workflows, Workflow Automation, and Management
A workflow could be a single operation or a sequence of
operations conducted by an individual, a group of people
or an organisation (Georgakopoulos et al. 1995). In early
days, processes in an organisation were solely controlled
by humans who manipulated physical objects, and with
the introduction of information technology, processes
above were ported to information systems and also
automated partially or entirely (Georgakopoulos et al.
1995). The notion of the automation and management
of workflows was originated about three decades ago
(Ploom et al. 2013), and it has been evolving throughout
years. There are tools and platforms designed specifically
for business workflow automation such as Xiaochun’s
Real Estate Office Automation System (Xiaochun 2009),
Wu & Huang’s COAS System (Wu & Huang 2011), and
the Platform Based Automation System of Ploom et al.
(2013). Also, there are many industry-level Business
Process Management tools and platforms such as WSO2
Business Process Server (WS02 n.d.), Sensus BPM
(Sensus n.d.), ProcessGene BPM (Processgene n.d.), and
KISSFLOW (KiSSFLOW 2014). These tools are solely
based on workflow management and automation and are
specifically designed to deploy and utilise in a small to
large-scale business environments.
As the Workflow Management Coalition (WfMC)
(Workflow Management Coalition 1999) states, a workflow
management system provides the automation of a business
process by managing the array of activities. According to
Lawrence, a workflow management system is a generic
software tool which permits for the interpretation,
execution, delegation and control of workflows (Lawrence
1997). Processes need to be initially captured to manage
1076 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
them in a workflow management system, and each process
needs to be understood with the cooperation of a human.
This process is called ‘Process Modelling’, and it is a crucial
part in the design of workflow management systems and
has been pleaded by numerous authors using distinct
approaches (Flores-Badillo & López-Mellado 2008).
Repetto et. al and Dumas & Hofstede have used a UML-
based modelling approach (Repetto et al. 2003; Dumas
& Hoftstede 2001), and Eyal & Haller have used applied
temporal and transactional logics.
c. Email Automation
The term ‘email automation’ fundamentally refers to the
automation of business workflows related to emails. It
involves the automation of sending corresponding emails
to recipients based on factors such as time, customer
engagements and statistical factors.
Email-based workflow automation systems such as Adobe
Campaign (Adobe n.d.), Act-on’s Marketing Platform
(Act-on n.d.), Infusionsoft Automation Tool (Infusionsoft
n.d.), and MailChimp (Mailchimp n.d.) are widely used
in modern day business environments to fulfil their
marketing needs. However, all the aforementioned Email
automation tools are developed with a marketing-centric
perspective; hence, making it a limitation of them since
they cannot serve as a general automation tool, which
can automate non-marketing workflows. However, there
exists an insignificant number of platforms/tools, which
are not marketing-centric and has the ability to fulfil the
purpose of email-based workflow automation such as
IFTTT (IFTTT Inc. 2016) and GSuite (Google 2016).
These systems are reviewed in section V.
IV. RESULTS OF THE
CROSS-SECTIONAL SURVEY
The fundamental goal of conducting a data survey was
to obtain an overall understanding of the usage of email
communication among the target audience as well as to
identify associated email-related workflows. It was needed
to gather data from a variety of fields of occupations to
derive a thorough understanding of the diversity of user
requirements in the context of workflow automation.
Another goal of conducting the survey was to verify the
need for an automation system for email-based workflows
by examining the usage of existing automation systems
among the target audience.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

1) Field of Occupation: A question was asked to identify As results reflect, the most significant department, which
the current field of occupation of the email users. The has a higher usage of emails is the IT Department. The
goal of asking this question was to understand the possible cause for that will be because the survey was
categorization of the email-related tasks/workflows based distributed mostly among people who are working in the
on their profession since the purpose of a task/workflow, IT field.
which the user tend to automate would vastly vary based Above two questions share a common goal, which is the
on their field of occupation. The responses as a percentage categorisation of the workflows based on users’ role in
have been presented in Table 1. an organisational environment. These results affect the
The results reflect the distribution among the fields, customizability of the system since the workflows, which
which were defined hence, the identification of the variety are to be automated by it vastly vary based on the users’
of tasks/workflows based on the area of occupation is requirements.
achieved. 3) Average time spent on emails: A question was asked to
derive an idea of the average time a respondent spend on
Table 1. Field of occupation of the respondents reading and writing emails.
Field of Occupation Percentage
Table 3. The amount of time a respondent
Legal Occupations 1.60%
spends on reading and writing emails on
Education, Training, and Library 24.60%
Occupations a regular work day
Architecture and Engineering 24.60%
Duration Percentage
Occupations
About an hour 34.40%
Business and Financial Op- 19.70%
erations, Sales and Related From 2 to 3 hours 14.80%
Occupations
From 4 to 6 hours 6.60%
Arts, Design, Entertainment, 1.60%
More than 6 hours 8.20%
Sports and Media Occupations
Management, Office and Admin- 19.70%
istrative Support Occupations The results shown in Table 3 reflects that the majority of
the users spend around an hour, and also a considerable
Other 8.20%
percentage (~30%) of the users spend more than two
2) Employed Department of respondents: A question was hours per day. Spending hours on processing emails
included to identify the section of the organisation, which affects the productivity of an employee in a greater way,
the respondent is contributing. The aim was to categorise and that justifies the need for an automation tool/platform
the tasks/workflows based on the section a respondent to automate portions of work.
belongs to in an organisation. Figure 2 reflects the results 4) The importance of email usage: The next question was
as percentages. asked to derive an idea of how important emails are in
respondent’s daily work. Table 4 shows the results of the
Table 2. The Department of the organisation, responses.
which the respondent belongs to
Department Percentage
Table 4. The role of Emails for respondent’s
Accounting 4.90% work duties and responsibilities
Administrative 3.30%
Opinion Percentage
Customer Service 9.80%
Email makes me more effective. I 46.0%
Engineering 18% can also use other communication
Marketing 16.40% methods
Operations 6.60% Email is critical. I cannot operate at all 46.0%
without Email
Human Resources 8.20%
Sales 8.20% Email is optional. I mainly use other 7%
communication methods (chat, phone,
Legal 3.30% fax, personal assistant)
IT 34.40% Other (Different Opinions) 1.60%
Other 4.90%

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1077
PROCEEDINGS ISBN number: 978-955-0301-41-6

The results clearly reflect that the communication through Criteria to review existing work were identified after
emails plays a vital role for those surveyed day-to-day the analysis of the data gathered through the cross-
work responsibilities and obligations for the majority sectional survey. The criteria were selected based on
of respondents hence, supports the proposal of the user requirements identified, and existing work will be
automation system to be based on emails. comparatively reviewed against them. The criteria are as
follows.
5) Respondent Opinion: The final question was to get
Criterion A - The support for email-based workflows is a
the respondent’s opinion about the need for a system as
proposed and the results are shown in Figure 5 reflects key factor, which needs to be fulfilled by the
that there is a clear necessity for an automation system for reviewed approaches.
Criterion B - The customizability towards userm require-
email based workflows.
ments is an essential factor since the vivid-
ness of them was high a identified through
Table 5. Respondent’s opinion about the
the survey.
necessity of the proposed system Criterion C - The extensibility is another key factor, which
needs to be satisfied by existing approaches
Opinion Percentage
up to a great extent. Complex workflows such
Yes 83.60% as classification of emails and email content
No 16.40% analysis, which were identified through the
survey suggest that the automation approach
Ultimately, the respondents were asked, which workflows should be able to integrate with external tools
related to emails would they like to automate and the and platforms.
responses were summarised as follows. Criterion D - The usability of the system is an essential
factor since it is prominent to have a learning
1. Classification of emails (complaint/issue/informative) curve, which is not steep for the user since it
2. Identify instructions in the email body could affect the productivity of an individual
3. Drafting documents based on the content or an organisation.
4. Summarising the content Criterion E - The deployment and maintenance cost plays
5. Categorising emails into folders
a major role in any system. As identified
6. Update systems based on the email content (system
through the survey, the automation system
configuration through email)
should be able to cater single users as well as
7. Scheduling appointments, calendar entries based on
large-scale organisations.
the content
8. Saving attachments A. Workflow Automation and Management
9. Notify others based on the content Approaches
10. Prioritise emails This section describes existing workflow management and
11. Keyword-triggered actions automation approaches, which are not limited to email-
12. Send generic replies based on the email type based workflows but also to other business processes such
13. Predict the response as assembly line processes, quality assurance processes
and maintenance processes.
V. EXISTING WORK 1) Xiaochun (2009): Xiaochun proposes a Workflow
Automation and Management System for Real Estate
This section contains a review of the similar work
Office Automation. He proposes a system to automate
identified in the Literature Survey. Approaches, which are
the execution of Business Processes through different
related to workflow automation and management, as well
Departments of the office. The implementation model
as Email Automation and Management, are succinctly
based on email transmission involves the processing
discussed in this section to derive an understanding of
of workflows based on person-to-person email
pros and cons of both traditional and modern trends in
communication. The disadvantages he mentions are
the domain of Workflow Automation.
the dynamic nature of these sorts of workflows and the

1078 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
difficulties of grasping them. He further implies that the
difficulty to achieve automation of these types of workflows
because of reasons such as the concentration of archiving
and the high network traffic caused. He implies that the
use of such a system does not suit for small scale Office
Automation because of the higher development cost, the
complexity of the implementation and the maintenance
difficulties of the system.
2) Wu & Huang (2011): Wu & Huang (2011) propose a
design for Office Automation Systems based on Workflows.
The Company Office Automation System (COAS) they
propose is a framework which manages data about the
uniformity of resources in the office using a centralised
database.
B. Email Automation Tools and Platforms
This section discusses existing approaches which are
designed for email campaign management and automation
as well as general automation tools which can be utilised to
automate email-based workflows.
1) IFTTT: ‘IF This Then That’ is a web platform which
provides automation capabilities across various online
services such as Gmail, Google Drive, Social Media websites
and even IoT Device Management Platforms. IFTTT
uses a concept called ‘Recipes’ to automate workflows by
connecting two web services. A Recipe is simply a rule
which consists of a trigger and a corresponding action, and
the trigger always being a specific keyword. This platform
can be used to automate Workflows related to Emails by
connecting an Email provider such as Gmail to any other
supported online service. However, this platform can only
connect two services at a time using a Recipe, and it is
limited to only a supported number of services.
2) MailChimp: This platform is designed specifically for
Email Marketing Campaign Automation and Management
and can integrate with a variety of other web-based services
such as Google Analytics and Salesforce. MailChimp
provides a significant number of pre-defined Email
templates which can be used to design Email messages
based on user requirements. MailChimp also provides
Marketing Automation features such as auto-responders,
periodic reminders and category-based follow-ups.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
3) Adobe Campaign: This service is a portion of the
Adobe Marketing Cloud and provides a large number of
Marketing Automation Solutions. Adobe Campaign is a
highly scalable platform and provides dynamic interfaces
for real-time user interactions.
4) GSuite: GSuite is a collection of apps provided by the
Google cloud platform. The apps include Gmail, Google
Docs, Google Calendar and Google Drive and the
platform provides the ability to connect them to form and
automate business workflows. However, this platform is
limited to the aforementioned applications hence, lacks
the customizability.
A contextual comparison of existing approaches against
identified criteria is presented in Table 6. The meanings of
the notation used are as follows.
• (-) - Low
• (-+) - Medium
• (+) - High
• NA - Not Applicable
VI. CONCLUSION AND
FUTURE WORK
This paper has presented an overall study on subject
domains of workflow automation, email usage among the
general public and email-based workflow automation.
Data gathered from the conducted data survey reflected
the vividness and complexity of email-related workflows
and supported the necessity of an email-based workflow
automation system, which has more customizability with
enhanced decision-making abilities was essential.
Existing approaches related to workflow and email
automation were reviewed and contextually compared
based on criteria: the support for email-based workflows,
customizability, extensibility, usability and the cost.
10TH INTERNATIONAL RESEARCH CONFERENCE 1079
PROCEEDINGS ISBN number: 978-955-0301-41-6

Table 6. Contextual comparison of existing Gibbs, S., 2016. How did email grow from messages
between academics to a global epidemic? The
workflow automation approaches Guardian. Available at:https://www.theguardian.com/
Xiaochun Wu & Huang IFTTT Mailchimp Adobe GSuite
technology/2016/mar/07/email-ray-tomlinson-history.
(2009) (2011) Campaign
Criterion A -+ - + + + + Google, 2016. G Suite – Gmail, Drive, Docs and More.
Criterion B - - -+ -+ -+ - Available at: https://gsuite.google.com/.
Criterion C - - -+ - - - IFTTT Inc., 2016. Learn how IFTTT works. Available at:
Criterion D NA NA + + -+ -+ https://ifttt.com/.
Criterion E + + - - -+ -+
Infusionsoft, Small Business Marketing Automation
The review along with the survey revealed that there Software. Available at: https://www.infusionsoft.com/
exists no email-based workflow automation system, software/features/marketing-automation.
which has enough customizability to tackle personalised
user requirements and, which fulfils the aforementioned KiSSFLOW, 2014. Business Process Management Tool &
criteria up to a satisfactory level. It was further identified Workflow Software. Available at: http://kissflow.com/.
that existing systems lacks decision-making abilities, Lawrence, P., 1997. Workflow Handbook, Joh Wiley and
which could handle sophisticated automation scenarios. Sons, New York.
We anticipate utilising the knowledge gained through this
paper to introduce an email-based workflow automation Mailchimp, Email Marketing Platform. Available at:
system, which provides more customizability towards https://mailchimp.com/.
user requirements and has enhanced decision-making
capabilities. Ploom, T., Glaser, A. & Scheit, S., 2013. Platform based
approach for automation of workflows in a system of
REFERENCES systems. c2013 IEEE 7th International Symposium on
the Maintenance and Evolution of Service-Oriented and
Act-on, Marketing Automation Software - Marketing Cloud-Based Systems, MESOCA 2013, pp.12–21.
Platform. Available at: https://www.act-on.com/.
Processgene, BPM Software | Processgene. Available at:
Adobe, Cross-channel campaign management, marketing http://processgene.com/solutions/bpm-software/.
automation software. Available at: http://www.adobe.com/
marketing-cloud/campaign-management.html. Repetto, M., Paolucci, M. & Boccalatte, A., 2003. A design
tool to develop agent-based workflow management
Bonaventure, O., Electronic Mail — Computer systems. Proc. of the 4th AI• IA/TABOO Joint Workshop“
Networking : Principles, Protocols and Practice. Available From Objects to Agents”: Intelligent Systems and Pervasive
at: http://cnp3book.info.ucl.ac.be/protocols/email.html. Computing, (January), pp.10–11. Available at: http://
citeseerx.ist.psu.edu/viewdoc/d
Dumas, M. & Hoftstede, A., 2001. UML Activity
Diagrams as a workflow specification language. UML 2001 Sensus, Sensus Process Management Software. Available
Conference. at:http://www.sensus-processmanagement.com/bpm-
software/.
Flores-Badillo, M. & López-Mellado, E., 2008. Mobile
agent based automation of distributed workflow processes. Workflow Management Coalition, 1999. Workflow
2008 IEEE International Conference on System of Systems Management Coalition – Terminology & Glossary.
Engineering, SoSE 2008. Technical Report, Available at: http://www.wfmc.org/.
WS02, WSO2 Business Process Server. Available at: http://
Georgakopoulos, D., Hornick, M. & Sheth, A., 1995. wso2.com/products/business-process-server/.
An Overview of Workflow Management : From Process
Modelling to Workflow Automation Infrastructure. , 153,
pp.119–153.
1080 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Wu, W. & Huang, Y., 2011. The analysis and design of
office automation system based on workflow. Proceedings
of 2011 International Conference on Electronic and
Mechanical Engineering and Information Technology,
EMEIT 2011, 1, pp.223–225.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Xiaochun, F., 2009. Workflow Automation and
Management in the Real Estate Office Automation System.
2009 Third International Symposium on Intelligent
Information Technology Application Workshops,
pp.322–324. Available at: http://ieeexplore.ieee.org/
document/5419429/.
Zhang, D. et al., 2006. Modeling Interactions from Email
Communication. Multimedia and Expo, 2006 IEEE
International Conference, pp.2037–2040.
10TH INTERNATIONAL RESEARCH CONFERENCE 1081
PROCEEDINGS
AUTOMATED GARBAGE COLLECTION
ALERT SYSTEM
UVADE silva 1, LD Ranatunge1, KA Weeraman1,
GSADE silva1 and ADAI Gunasekara1
Faculty of Computing, General Sir John Kotelawala Defence University,
Ratmalana, Sri Lanka
# vinura.asel999@gmail.com
Abstract - Waste collection has become one of the major
issue in today’s world. In a country like Sri Lanka there
are no proper facilities for final disposal produced by
households and industries. And also, relevant authorities
do not pay much attention to this matter. There are
different types of waste collection and management
methods all around the world. But in Sri Lanka there is
no proper computer based waste collection system has
been developed yet. In this study a methodology for
the optimization of the waste collection and transport
system, based on Geographic Information System (GIS)
was developed in this system, we introduce a mobile
application to manage garbage in various areas. Through
this mobile application, we can direct connect to the
Municipal Council and inform the exact location of the
garbage dump. In this system, we mainly focus on garbage
dumps in the streets and main roads. By using this mobile
application both people and municipal council can save
time and extra money that they spend to collect waste in
country like Sri Lanka.
Keywords - GIS, Solid waste disposal, Location tracking
I. INTRODUCTION
With the increase in population garbage has become one
of the major problem in a country like Sri Lanka. There are
no proper way to disposal the garbage in our country. So
people used to put garbage beside the roads. Daily life in
countries like Sri Lanka can generate several kilograms of
1082 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
solid waste from houses, factories, companies and various
places. And with the development of Sri Lanka people are
moving to cities for their own benefits, so that the amount
of garbage increasing day by day. Because of these reasons
garbage problem has become a huge problem for people.
There are different types of waste collection and
management methods all around the world. But in Sri
Lanka there is no proper computer based waste collection
system has been developed yet.
So in this project we decided to implement a mobile
application to develop a computer based garbage collection
system. Using Information & communication technology
this problem can solve successfully and also improve the
environment friendliness.By using this “CleanUp” mobile
application anyone can inform the locations of garbage
dumps.
Anyone can this mobile application from anywhere by
typing their name and their mobile number. And then he/
she can track the garbage location and send it to the nearest
municipal council. Then user will get a confirm message
whether the tracked location and other information’s (date
and time, username, user mobile number) are send or not
to the municipal council.
A. Objectives
With the aim of developing an Automated Garbage
Collection Alert System, we have identified the following
objectives.
 ISBN number: 978-955-0301-41-6
• To critically study the current issues in garbage
collectionin Sri Lanka
• To critically study the existing computer based solution
for garbage collection
• To do a study software engineering life cycle
• To conduct a detailed study on current mobile
applications, database and web technologies
• To develop a prototype for the system
• Design and develop Automated Garbage Collection
Alert System
• Evaluate the new Automated Garbage Collection Alert
System
B. Hypothesis
One of the major issue that we are facing today is garbage
collection. The methods that we are using to collect garbage
today in Sri Lanka is not effective. And there is no proper
Information Communication and Technology (ICT) based
system to solve this matter. Technology has become a very
important thing to our day today life and it is beginning to
reshape our standard way of living. Internet is one of the
information resources that today people are used. People
save a lot of time and cash and find the information by
giving little effort within the comfort of the home. In near
future Information and Communication technologies
(ICTs) will apply directly to urban problems to enable safer,
healthier cities. The hypothesis for the creating CleanApp
mobile application is to develop a computer based garbage
collection system. In this system we mainly focus on to
remove the the garbage that collect beside roads. Using
Information & communication technology this problem
can solve successfully and also improve the environment
friendliness.
C. Software solution for Garbage Collection Alert
System
The solution of the Garbage collection alert system it is
necessary to use new technological methodology for
the system. According to that requirement system has
developed by using C# and using SQL database to run on
windows operating system. The mobile application has
developed by using Java and Android studio.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
D.Resource Requirements
• The mobile application would develop as it is compatible
with any other mobile phones which runs android, IOS
and windows.
• The web server will be developed as it is compatible with
any computer which runs Windows 7, Windows 8.1 and
Windows 10.
• The mobile application will be implement by using Java
and the android studio.
• The administrative interface will be developing by using
the C# language with ASP.net framework.
• The database will be implement by using db forge studio
for SQL server
II. LITERATURE REVIEW
Through the review, it will be thought about regarding the
procedure and available practices in automated garbage
collection
[1] Jose M. Gutierrez et al (2015) has presented a waste
collection solution based on providing intelligence to
trashcans, by using an IOT prototype embedded with
sensors, which can read, collect, and transmit trash
volume data over the Internet. They have described how
an integrated cyber physical system design, based on the
combination of different disciplines in engineering, and
taking advantage of municipal wireless access networks
can lead to smart ways of improving the management of
cities
[2] Adebayo P. Idowu et al. (2012) has developed a web
based GIS waste disposal management system with the
aim of achieving an effective waste management system
and a spatial view of waste collection locations in any local
government are in Nigeria. With this waste management
system, the location of all the waste collection tanks in any
location will be, monitored, managed and maintained.
The use of this system will ease the job of the waste
management unit of the local government areas in Nigeria
in achieving a clean environment and mitigate the spread
of epidemic in a way to ensure safety of all and sundry.
[3] Ashish Sam Geo, et al., have developed a new approach
for a systematic development in managing solid waste.
Garbage location is identified by using the GPS device
installed and the coordinates are send to the corresponding
mobile with GPRS module.
10TH INTERNATIONAL RESEARCH CONFERENCE 1083
PROCEEDINGS
In this system, they have focused on to develop an
electronic monitoring system for solid waste management.
The system will have the facility to send SMS to the workers
and supervisors. To develop a web based GUI so that the
system can be accessed from anywhere and information
can be viewed by different group of people.
III. METHODOLOGY AND
EXPERIMENTAL DESIGN
A. Approach
In the Garbage collection alert system process first user
should login to the Mobile application by providing user’s
name and phone number to the system. Then the mobile
application will track the coordinates (longitude and
latitude) of the location where user is in (where the garbage
dump is located). Then those information; User name,
User mobile Number, Date and time and coordinates of
the tracked garbage dump will send to the web server
that is located in Municipal Council. After taking the
information from the mobile application those details will
be displayed on the municipal council’s web server. The
location of the garbage will be display on a map according
to the coordinates that sent from the mobile application.
Then the assign people will be send to that particular
places to collect that garbage. And the web application will
send a notification for the user whether the information
send successfully or not and give a compliment for the
sender for his service Each information that users send
will stores in SQL server database successfully and for the
work with the application it required continuous internet
connection.
B. Technology Adopted
It is very important to use acceptable tools so as to develop
productive system. Use on any inappropriate tools can
solely ends up in develop a system with unnecessary errors
and faults and use of those badly chosen technologies
additionally can ends up in crashed when the new system
implementation. Badly chosen technologies which can
be extremely advanced and complicated will enable
manufacturing a system with a top quality, however these
technologies may result in develop a system that spend
lots of time and resources so as to perform a task that
is anticipated by the system. It is very important to use
applicable programming language and the other necessary
tools in order to develop a productive system. Therefore,
1084 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
these technologies and tools can help to develop the
system among a minimum development time the most
objective of developing this type of an application is to
produce the users more efficient work system instead of
doing manual approach. Because of that we should use the
most applicable tools available in the market to develop
the system. Technological considerations - followed
during the development of the system Efficiency and
Performance Re-usability and flexibility object oriented
development support so according to the Automated
Garbage Collection Alert System java and android studio
used to develop the mobile application. C# language is
used, SQL Server 2012 as database for this project. This
chapter includes the details about the technologies that we
are going to use to develop Automated Garbage Collection
Alert System.
Web Application
The programming language that is going to apply as
the developing language for the system development
turned into significantly trusted accuracy, performance.
When considering all these technologies which can
be associated with the Garbage collection system the
proposed system can be applied a web based technology.
The .NET Framework consists of the common language
runtime and the .NET Framework class library. The
common language runtime is the foundation of the .NET
Framework. You can think of the runtime as an agent that
manages code at execution time, providing core services
such as memory management, thread management, while
also enforcing strict type safety and other forms of code
accuracy that promote security and robustness. In fact, the
concept of code management is a fundamental principle
of the runtime. Code that targets the runtime is known
as managed code, while code that does not target the
runtime is known as unmanaged code. The class library
is a comprehensive, object-oriented collection of reusable
types that you can use to develop applications ranging
from traditional command-line or graphical user interface
(GUI) applications to applications based on the latest
innovations provided by ASP.NET, such as Web Forms
and XML Web services. For example, ASP.NET hosts the
runtime to provide a scalable, server-side environment
for managed code. ASP.NET works directly with the
runtime to enable ASP.NET applications and XML Web
services. C# programs run on the .NET Framework, it
runs on a virtual execution system called the common
language runtime (CLR) and a combined set of class
libraries. The CLR is the implementation by Microsoft of
 ISBN number: 978-955-0301-41-6
the common language infrastructure (CLI) and it helps to
create execution and development environments in which
languages and libraries work together without any flaw.
C# source code is compiled into an intermediate language
(IL) that conforms to the CLI specification. The IL code
and resources such as bitmaps and strings are stored on
disk in an executable file called an assembly. Assembly is
typically creating with an .exe or .dll extension. It contains
a manifest that provides information about the assembly
such as types, version, culture, and security requirements.
When the C# program is executed the assembly is loaded
into the CLR. Based on the information in the manifest
CLR might take various. When the security requirements
are met, the CLR performs just in time (JIT) compilation
and convert the IL code to native machine code. CLR
also provides other services such as automatic garbage
collection, exception handling, and resource management.
Following diagram illustrates the compile time and
run-time relationships of C# source code files and the
.NET Framework. The following illustration shows the
relationship of the common language runtime and the
class library to our applications and to the overall system.
The illustration also shows how managed code operates
within a larger architecture.
Figure 1. Interface of the web application
Mobile Application
Android gives you the freedom to put into effect your own
device specifications and drivers. The hardware abstraction
layer (Hal) presents a well-known approach for developing
software hooks among the android platform stack and
your hardware. The android working machine is likewise
open source, so you can make a contribution your own
interfaces and improvements. Android is an open source,
Linux-based software stack created for a wide array of
devices and form factors. The following diagram shows the
major components of the Android platform.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Figure 2. Interface of the mobile app
Android run time
For gadgets going for walks android version 5. Zero
(API level 21) or higher, every app runs in its very own
technique and with its own instance of the android run
time (artwork). Art is written to run multiple virtual
machines on low-reminiscence devices by using executing
DEX files, a byte code format designed especially for
android that is optimized for minimum reminiscence
footprint. Construct tool chains, including jack, assemble
java resources into DEX byte code that can run at the
android platform. A number of the predominant features
of art encompass the subsequent in advance-of-time and
just-in-time compilation optimized garbage collection
higher debugging guide, such as a dedicated sampling
profiler, exact diagnostic exceptions and crash reporting,
and the capability to set watch points to screen particular
fields previous to android model five.0 (API degree 21),
dalvik was the android run time. In case your app runs
well on artwork, then it need to work on dalvik as properly,
however the opposite may not be real. Android additionally
consists of a hard and fast of core run time libraries that
offer most of the capability of the java programming
language, which includes some java 8 language functions
that the java API framework uses.
Database Selection
Consistent with the above eventualities most of the
structures are used the square database to keep facts. It
seems it is simple to control and perform. So, the database
put in force on the server have to able to supplying
efficiencies operations. Consequently, the proposed system
decided on the Microsoft SQL server 2014 as server.
SQL server is the inspiration of Microsoft’s data base
platform, delivering challenge critical performance with
in-reminiscence technology and quicker insights on any
10TH INTERNATIONAL RESEARCH CONFERENCE 1085
PROCEEDINGS
information, whether or not on-premises or in the cloud,
and also Microsoft SQL Server is an application used to
create computer databases for the Microsoft Windows
family of server operating systems. Microsoft SQL Server
provides an environment used to produce databases that
can be accessed from workstations, the Internet, or other
media too. Database management or DBMS, store user’s
data and enables them to transform the
25 | P a g e information into statistics. Those systems
allow users to create, replace and extract facts from
their database. A database is an established collection of
information. Facts refer to the characteristics of human
beings, things and activities. Square server stores every
statistic item in its very own fields. In square server, the
fields related to a particular character, thing or occasion
are bundled collectively to shape a single complete unit of
records, known as a document. Each record is made up of
some of fields. No two fields in a record will have the equal
area name. Throughout an SQL server database design
project, the evaluation of your project wishes identifies all
of the fields or attributes of interest. If your commercial
enterprise desires trade through the years, you outline any
extra fields or alternate the classification of present fields.
Figure 3. Interface of the Database
IV. EVALUATION
In this, we describe evaluation of our approach and the
developed system while evaluating the objectives achieved
how the project deviated from its original specifications
and the circumstance identified during the time period of
the project. This chapter will give the idea of the measure
that have been taken to handle the problem occurred
and knowledge which have been gathered by supplying
solutions for such issues.
A formative evaluation a method for judging the worth
of a program while the program activities are forming
(in progress). This evaluation is done with the user
requirements or the function requirements.
1086 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Summative evaluation refers to the assessment of
participants where the focus is on the outcome of a
program. It is done with a high-fidelity prototype to assess
the achievement of product more progressive.
V. CONCLUSION
The results and outcomes generated in relative to the
specificity of the problem domain are enlarged into wider
concepts depending on logical assumptions. This chapter
aims to clearly emphasize the outcomes and findings of
the project and to determine way of these outcomes and
findings can be matched in different contexts that are
similar to the problems which are solved by the developed
Garbage collection and alert system. In the rest of the
chapter a total summary of the development of the system
is given. Furthermore, future enhancements for the
developed System have suggested finding out ways to give
in addition features to the system and using it outside the
business subject in use.
The aim of this project was to develop a Garbage collection
alert system as a solution of disposal of garbage in Sri Lanka.
The development team implemented this system in order
to determine its ability to satisfy the entire functional and
non-functional requirement with special qualities such as
flexibility, reliability efficiency and etc., to overcome the
drawbacks identified in the system. The study found out
that it is feasible to use the language ASP.NET in C#, SQL
Server 2012 as database and java in android studio used
to develop the mobile application to develop the project.
It’s is a mobile application and web-enabled project so this
mobile application offers user to install the application
and enter data. This is very helpful for the user to enter
the desired information through so much simplicity.
The user is mainly more concerned about the validity of
the data, whatever he is entering. In Web server admin
provided the option of monitoring the records entered
earlier. Data storage and retrieval will become faster and
easier to maintain because data is stored in a systematic
manner and in a database. Decision making process
would be greatly enhanced because of faster processing
of information since data collection from information
available on computer takes much less time than manual
system.
This system allows to get information about the garbage
dumps in the relevant city. This gives efficient and cost
effective. Mobile application can be access by defined
 ISBN number: 978-955-0301-41-6
user categories by verifying their username and telephone
number and web server can only be access by the admin by
verifying the username and password. Client machines can
be Windows xp, Windows 7, Windows 9 or Windows 10.
Server computer should have operating system Windows
xp, Windows 7, Windows 9 or Windows 10 and should be
installed Visual Studio 2012, SQL Server 2012 and tool set.
ACKNOWLEDGMENT
It would not have been conceivable without the generous
support and help of numerous Individuals who were
there throughout the project. I might want to expand my
thanks of every one of them. First and foremost, I would
like to express sincere gratitude to the supervisor of this
project Mr. Asela Gunasekara & Mrs. WPJ Premarathne
for all inspiration and direction to do this report and
kind support and assistance given throughout the report
writing. The courage he gave by assessing during our report
preparation and providing quick response, are very much
valued. I might want to express my unique appreciation
and thanks to industry persons Mr. Sidath Gajanayake and
Mr. Dinuka Jayawardane to help us by giving information
about process of the work flow and everybody who were
bolster us for the achievement of our project.
REFERENCES
Adebayo P. Idowu, Emmanuel R. Adagunodo, Olapeju A.
Esimai, & Tosin C. Olapade, “Development of A Web based
GIS Waste Disposal Management System for Nigeria” I.J.
Information Engineering and Electronic Business, 2012, 3,
40-48
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Ashish Sam Geo, Durga M., and D. Baskaran, “Intelligent
Garbage Location Monitoring using Remote Server and
Sensor Networks,” Int. J. Electron. Commun. Comput.
Eng., vol. 5, no. 2.
Jose M. Gutierreza *, Michael Jensenb , Morten Heniusa
and Tahir Riaz, “Smart Waste Collection System Based on
Location Intelligence”, Procedia Computer Science 61 (
2015 ) 120 – 127
S. D. T. Kelly, N. K. Suryadevara, and S. C. Mukhopadhyay,
“Towards the implementation of IoT for environmental
condition monitoring in homes,” IEEE Sensors Journal,
vol. 13, no. 10,pp. 3846–3853, 2013.
K. Gama, L. Touseau, and D. Donsez, “Combining
heterogeneous service technologies for building an
internet of things middleware,” Comp
L. Foschini, T. Taleb, A. Corradi, and D. Bottazzi, “M2M-
based metropolitan platform for IMS-enabled road traffic
management in IoT,” IEEE Communications Magazine,
vol. 49, no. 11, pp. 50–57, 2011.
H. Andrianto, Pemrograman Mikrokontroler AVR
Atmega16,menggunakan Bahasa C (codevisionAVR),
Bandung: Informatika,2008.
L. Wardhana, Belajar Sendiri Mikrokontroler AVR Seri
ATMega8535, Yogyakarta: Andi Offset, 2006.
Myke Predko,”Programming and customizing the ARM
7 microcontroller”, McGraw-Hill Professional.
D.Roy Choudhoury and Shail B. Jain,”Linear Integrated
Circuits”, New Age International Publishers, 2009
10TH INTERNATIONAL RESEARCH CONFERENCE 1087
PROCEEDINGS ISBN number: 978-955-0301-41-6

NATIONAL FORENSIC DNA DATABASE

MANAGEMENT SYSTEM
FOR CRIMINAL INVESTIGATIONS
NCD Arambawela1#, PPNV Kumara2,
CP Waduge1 and A Manamperi1
1
Faculty of Computing,
1
General Sir John Kotelawala Defence University, Sri Lanka
2
Faculty of Medicine, University of Kelaniya, Sri Lanka
# ncdarambawela92@gmail.com

Abstract - A DNA profile is a genetic fingerprint which
is unique to each person and it is used as a powerful
evidence to identify individuals in criminology. A DNA
database holds DNA profiles taken from individuals
and crime scenes and helps in quick identification of
criminals that leads to improve the efficiency of the
judiciary system of a country. The current process of
using DNA evidence in court cases in Sri Lanka is more
time consuming since the widely used electrophoresis gel
image analysing technique requires expert scientists and
often has human errors because of the anomalies that can
occur in the faint gel images. Although there are novel
advanced technologies and equipment based on capillary
electrophoresis, the enormous cost of implementing and
licensing them is unaffordable to third world developing
countries like Sri Lanka. Therefore, this research paper
proposes a fully automated and cost effective methodology
to re-engineer the current manual process by integrating
recent advancements in the computer science field such
as image processing and machine learning techniques
to spare human being from voluminous and tedious
image analysing and to provide accuracy and rapid speed
without human errors. The computerized system will get
the scanned electrophoresis gel image as the user input
and it will enhance the quality of the gel image by using
image processing techniques. After that it will generate
the DNA profile and save it as a number pattern to the
database for speedy retrieval. The system provides matches
linking an individual to a crime scene or a crime scene to
another crime scene. Furthermore, the system facilitates
family relationship identification based on machine
learning techniques. It ensures the security by integrating
fingerprint authentication. Establishing a National
Forensic DNA Database Management System in Sri Lanka
will benefit in identifying criminals and excluding the
innocent efficiently. More importantly, it will support to
prevent criminals from having many opportunities to
commit mass murders, rapes and robberies and to identify
potential victims.
Keywords - DNA profiling, DNA database, Electrophoresis
gel images, Image processing and Machine learning

1088 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
I. INTRODUCTION
DNA typing is an accurate scientific method that uses to
identify individuals from the differences in their DNA
(Deoxyribose Nucleic Acid) structure. It produces a
pattern which is termed as a DNA profile. An individual’s
DNA profile is a genetic fingerprint which is unique to each
person with the exception of identical twins that have the
same DNA content. But the probability of having the same
DNA profile is less than one in a billion (Great Britain et
al., 2015). Therefore, DNA profiles are used as a powerful
evidence to identify individuals in forensic science.
Every cell in a person’s body contains the same DNA
content. Therefore, DNA profiling can be performed by
using a very little amount of biological specimens using
blood, sperms, muscle, salvia, bones, teeth or even sweat.
The nucleus is the commanding center of a cell and
it houses the DNA that codes genetic information
responsible for all cellular functions. Several DNA
molecules that comprise genes are called as chromosomes.
In a human cell, there are a total of 46 chromosomes; 23
chromosomes are inherited from the mother and the other
23 chromosomes are inherited from the father. There are
repeating units of the same type of DNA sequence in some
chromosomal regions and the number of those repeating
units in individuals can vary. Hence those repeated DNA
sequences which are known as Short Tandem Repeat
(STR) or STR markers are used in human identification
in criminology. Tetra-nucleotide repeats are mostly used
in DNA typing. DNA profiling is done by counting the
number of times each repeating unit occurs within a
specific area on the chromosome.
Electrophoresis gel image analysis is the most widely
used technique that is used for DNA fingerprinting.
Electrophoresis is an electrochemical separation method
that makes DNA molecules to migrate through a specific
substrate such as polyacrylamide gel under the influence
of an electrical current. The output of this electrophoresis
process is called as an electrophoresis gel image and that
scanned image is analysed by the expert scientists for
DNA profiling.
In forensic science, DNA typing is used for two main
purposes; to identify individuals from biological samples
and to determine familial relationships. Since a DNA
profile has recognized as a powerful evidence in forensic
casework, the concept of national forensic DNA database
was introduced to the field of criminology. Hence national
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
DNA databases have been established in several countries
such as United Kingdom, Netherlands, Australia, Germany,
Finland, Norway, etc. and many other countries are
developing DNA database systems. Researches highlight
that DNA databanks of criminal offenders benefit in quick
identification of criminals and help to reduce the crime
rates of a country.
Although the existing legal framework in Sri Lanka
provides the legal acceptance for DNA evidence in court
cases, still a national DNA database is not available. In
Sri Lanka DNA profiling was used for the first time in a
criminal case when six family members were murdered in
Hokandara. Aftermath over 1400 criminal cases and over
3000 cases of disputed parentage have had the advantage
of DNA evidence (“Genetech,” n.d.). Some significant
incidents such as Sarath Ambepitiya murder case, Royal
Park murder case and Seya Sadewmi’s murder case can be
defined as landmark cases in terms of using DNA typing
as expert evidence.
The current process of using DNA profiles as an evidence
in court cases in Sri Lanka is more time consuming and
not cost effective. It requires expert scientists to match
profiles. Sometimes DNA samples that are taken from
crime scenes are full of dust, sault, etc. The electrophoresis
gel images of those samples contain anomalies, hence it is
very difficult to analyse those faint images. Security issues
also have arisen regarding the current process. Nowadays,
most countries are adopted to a novel DNA analysing
technique which is called as capillary electrophoresis.
The huge cost for purchasing, implementing, maintaining
and annual licensing of the equipment based on capillary
electrophoresis method is unaffordable to third world
developing countries like Sri Lanka.
Therefore, this research paper suggests to deploy recent
advancements in computer science field to re-engineer
the existing electrophoresis gel image analysing process.
Image processing techniques can be used to eliminate the
anomalies that can occur in electrophoresis gel images. All
the DNA profiles of individuals and crime scenes are stored
as a number pattern along with the personal information
in the database for further manipulation. Thereby, it will
save the time taken for investigations. Machine learning
techniques will be integrated to the system to enhance
the efficiency of the family relationship matching process.
Web technology is combined with the proposed solution
to speed up the existing process. It will gain the advantage
of using biometric authentication technique to ensure the
security of the sensitive data that handles in the system.
10TH INTERNATIONAL RESEARCH CONFERENCE 1089
PROCEEDINGS ISBN number: 978-955-0301-41-6

The aim of this research is to enhance the efficiency, Various image processing techniques had been used in a
accuracy and security of the current DNA analysing three steps algorithm to analyse DNA gel images. Firstly an
procedure in forensic casework in Sri Lanka by re- enhanced fuzzy c-means algorithm is designed for image
engineering it through integrating appropriate newest segmentation to extract the helpful information from the
technologies introduced in the computer science field. DNA gel images and exclude the unnecessary background
that includes blurred noise. The next step named as lane
detection uses Gaussian function to estimate and detect
II. LITERATURE REVIEW the location of each lane on the gel images. For the final
step renewing lost bands and eliminating repetitive bands
The literature review emphasizes an appraisal of the similar
are applied in order to ascertain each band more accurately
existing systems. The objective of the literature review is
(Lee et al., 2011).
to analyse the focused problems, adopted technologies,
proposed solutions and research gaps in previous similar
Literature emphasises image analysis techniques and
researches. Furthermore, it focuses on evaluating the
pattern recognition techniques can be used to obtain
applicability of previous studies into the problem domain
quantitative and qualitative information from gel images.
of this research.
The background of most gel images varies because of the
presence of noise and some bands of gel images are not
A research highlights a four steps algorithm that addresses
aligned horizontally because of non-uniform migration
the bottleneck for further development and reproducibility
called as ‘smile’ on gel. The goal of image enhancement
issues of manual or semiautomatic DNA profiling
is to obtain accurate quantification from distorted gel
processes. The four steps include in this algorithm are
images. Top-hat transform technique has been used for
automatic thresholding, shifting and filtering, detecting
background removal. The clustering analysis addresses the
and annotating gel bands and data processing. Automatic
issue of measuring similarity and dissimilarity between
thresholding is engaged to equalize the grayscale levels of
two samples based on the distance between them. The
electrophoresis gel image background without affecting
K-means algorithm, ISODATA algorithm and vector
the size of DNA bands. The purpose of the second step
quantization are used as clustering methods based on
is to shift the minimum level of the gel image to zero and
discriminant analysis (Ye et al., 1999).
to remove as much noise as possible using top hat filter.
An indirect method consists of top-hat filtering and
The “GELect tool” is an appropriate software for DNA
bottom-hat filtering has been used in the third step to
diagnosis from 1D electrophoresis gel images. The
improve the quality of gel images. In the data processing
workflow of GELect comprises three main procedures;
step object detection technique has been used to detect all
lane segmentation, DNA band extraction and band
the DNA bands and to compute quantitative information
genotyping. This tool efficiently segments lanes from the
(Kaabouch et al., 2007).
gel image by detecting curved lanes automatically and it
constructs a band model by performing band registration
A method of detection of DNA fingerprint to identify
against a reference band. GELect tool has used the band
family relationship through the use of image processing
classification technique to perform genotyping from DNA
based on medical knowledge is presented in literature.
gel images (Intarapanich et al., 2015).
In this method, when a DNA testing performer inputs a
DNA fingerprint to a commercial program, the DNA gel
A research done by Maxwell and William provides
image is improved by image enhancement as the first step.
an overview of the applicability of machine learning
For the next step the software converts the DNA image
approaches for analysing genome sequencing data sets. It
to a binary code, reduces small noisy spots and increases
presents challenges and considerations in the application
the quantity and the size of noisy spots of the size of
of supervised, unsupervised and semi-supervised machine
the binary image to be 45*26 pixels. In the third step it
learning methods as well as generative and discriminative
is correlated by template matching to identify the same
modelling approaches. Although the generative modelling
DNA positions in terms of mother, father and child. As
gives more compelling benefits than discriminative
the final step a complete relationship or no relationship is
modelling, discriminative modelling achieves more
verified in terms of 10 positions of similarity (Kiattisin and
performance than generative modelling (Libbrecht and
Leelasantitham, 2008).
Noble, 2015).

1090 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
cussed researches emphasise, the image quality of
electrophoresis gel images plays a significant role in
automatic analysis of DNA signatures. Existing researches
highlight the filtering techniques still require considerable
human intervention and pre-assumptions to define
appropriate values for thresholding due to the diversity
and quality of images. Moreover, those researches
underlines that some techniques are not suitable for DNA
electrophoresis gel images because they alter the size of the
DNA bands.
Most of the existing algorithms consider the data
structure of DNA sequences as tree, string and graph.
Some researches highlight that it can limit the efficiency of
retrieval and the speed of processing.
Some existing systems have used only the image
processing techniques according to the addressed problem
domains. Some systems have used signal processing
techniques or pattern recognition techniques combine
with image processing techniques to achieve better
performance in DNA sequence analysis. Some researches
emphasise signal processing techniques has received a
great attention in analysing numeric sequences compared
to other technologies such as pattern recognition. Very
few researches have considered of applying supervised
machine learning techniques such as Artificial Neural
Networks in analysis of DNA sequences.
Even though previously discussed DNA image analysing
systems are dealing with sensitive data, those systems did
not consider about applying robust security mechanisms
to those systems since they are not combining with
national DNA databases and crime analysis.
This research focuses on addressing the above identified
research gaps to deploy a more accurate, efficient and
secured IT solution in forensic science in order to speed
up the traditional DNA profiling process in Sri Lanka.
III. METHODOLOGY AND
EXPERIMENTAL DESIGN
A. Data Gathering
Qualitative and quantitative data that required for carrying
out the research were gathered through interviews and
document reviews. An interview was conducted with the
Head of Department of the Molecular Medicine Unit of
the Faculty of Medicine, University of Kelaniya to acquire
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
the medical knowledge that needs for the project. Another
interview was done with the Technical officer of the DNA
laboratory of the Molecular Medicine Unit of University
of Kelaniya to gather information of the electrophoresis
gel image analysing process. An interview was conducted
with the Deputy Government Analyst of the Government
Analyst Department to gather data about the existing
procedure of analysing DNA sequences for court cases.
Some DNA electrophoresis gel images, reports and excel
sheets of DNA profiles were reviewed during the data
gathering process.
B. Data Analysis
The data which was gathered during the data collection
process was analysed in this phase to define the
problem and to identify the limitations with the existing
process. Suggestions of the users to improve the current
process could be identified through the collected data.
Consuming much time and cost, low security, limited
analysing capabilities due to unclear images are some of
the drawbacks with the manual procedure which could
be identified through data analysis. Moreover, the data
analysis phase highlighted that the users are also willing
to adopt to an automated system to speed up the DNA
analysing process. And also this phase emphasized that
re-engineering the existing process for better functionality
is more important to enhance the efficiency of crime
investigations in Sri Lanka.
C. Approach
Users of this system are authorized police officers, jailers
and laboratory staff. Inputs for the system are scanned
DNA electrophoresis gel images and the personal details
of offenders or suspects. Outputs from the system are
basically a report that indicates the matching probability
of the profiles, an email of the report and a SMS alert.
The system receives inputs and executes user requests to
generate DNA reports and display the output through an
email.
D. Technology adopted
The developed system consists with a web application and
a standalone application. The web application has been
developed using Bootstrap UI design framework and
programming languages such as HTML, CSS, JavaScript
and PHP. The standalone application was designed
using Swing framework. Java programming language
10TH INTERNATIONAL RESEARCH CONFERENCE 1091
PROCEEDINGS
with object oriented programming concepts was used to
programme the functions of the standalone application.
Image enhancement process was designed through image
processing filters and functions in MatLab. An Artificial
Neural Network that is known as a supervised machine
learning model which identifies the patterns based on a
previously trained dataset, has been integrated in order
to develop the family relationship matching process. A
MySql database has designed to store data. Desktops,
Scanners and fingerprint scanners are some of the
hardware technologies deployed in the developed system.
E. Design
The overall architecture of the system can be defined based
on three main layers; client layer, application layer and
database and server layer. Architectural perspective of the
developed system is shown in figure 1.
1) Client Layer: The client layer provides the access to
the users of the system. There are two main user levels
(authorized police/jail officers and authorized DNA
laboratory staff) in this system and access levels are varies
from one user level to the other. Therefore the authorized
users will be predefined.
Figure 1. System Architecture
2) Application Layer: The application layer consists with
a web application and a standalone application. The web
application is designed for the jailers and police officers.
It can be further described through the below discussed
modules.
i. Login module
This module verifies the username and password that are
entered by the users and gives the access to the system for
the predefined authorized users. It has given an option
1092 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
to recover username and password if the username or
password is forgotten.
ii. Insert suspects module
In this module police officers can enter the personal
details of the suspects whose biological samples have been
submitted for DNA profiling.
iii. Insert offenders module
Jailers can insert the personal details of the offenders
whose biological samples have been submitted for DNA
profiling.
iv. Update suspects module
When the suspects are termed as offenders, relevant details
of them such as imprisonment date, number of years of
imprisonment, etc. can be inserted to the system through
this module.
v. Update offenders
This module facilitates jailers to update the information of
offenders when necessary.
vi. Delete offenders
When the offenders die, the details and DNA profiles of
them can be deleted through this module.
The desktop application allows authorized DNA laboratory
staff to perform the functions through below described
modules.
i. Login module
This module allows authorized users to access the system
by providing the fingerprint scan.
ii. One-to-one profile matching module
In this module user can browse and insert the scanned
electrophoresis gel image. The “image enhancement”
option has been provided to improve the quality of the
image. Then the DNA profile can be generated by clicking
the “generate profile” option. After that generated profile
will be saved to the database and then it goes for a search
against the database to find a full match or a family match.
A report will be generated indicating the DNA profile,
matching probability and other relevant details. The
workflow of this module is shown in figure 2
 ISBN number: 978-955-0301-41-6
Figure 2. One-to-one matching workflow
iii. Family relationship matching module
Scanned electrophoresis gel image will be browsed and
inserted to enhance the quality of the image. Then the
DNA profiles of mother, father and child will be generated
respectively. Those profiles will be saved to the database
and find whether there is a relationship in terms of father,
mother and child. After that a report will be generated.
Figure 3 shows the workflow of this module.
Figure 3. Family relationship matching
iv. Email sending module
The generated reports will be send to the relevant personals
through emails in this module.
v. SMS sending module
This module allows to send SMS alerts automatically to
inform about the emails.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
3) Server and Database Layer: This layer is responsible
for managing the server and the database of the system.
The server will provide the web connectivity to the web
application and the database will store the data that are
entered to the system by two applications and the stored
data will be provided to the users when necessary.
IV. RESULTS AND DISCUSSION
The above identified filtering techniques such as top-hat
filtering, bottom-hat filtering, Fourier transformation, etc.
that described in previous researches could not applied
directly to the addressed research domain. Therefore, a
combination of image processing filters, functions and
algorithms were used for the image enhancement process
of the system to improve the quality of electrophoresis gel
images without altering the size of the DNA bands.
The speed of retrieval and manipulation of data could be
upgraded by representing the DNA profiles in number
patterns rather than symbolizing them as tree, string or
graph.
Moreover, using machine learning techniques along with
image processing techniques in DNA profile matching
provides a better combination to analyse DNA sequences.
It gives more accurate and efficient results than other
combinations which are described in the previous studies
that discussed in the literature review section.
More importantly, security breaches of standalone
application could be eradicated successfully by using
fingerprint authentication mechanism rather than using
traditional username and password mechanism.
Implementation of the web application that allows police
officers and jailers to insert personal details from different
places helps to spare laboratory staff from tedious data
entering. The distribution of works benefits in enhancing
the efficiency of the system. Email sending facility and
SMS gateway supports to reduce the paper work.
More efficient and accurate electrophoresis gel image
analysing program for DNA profiling and a secured
solution for establishing a national forensic DNA database
could be provided through this research by addressing the
research gaps found in the previously proposed solutions
in literature.
10TH INTERNATIONAL RESEARCH CONFERENCE 1093
PROCEEDINGS
V. CONCLUSION AND
FURTHER WORK
In this research work, the author could develop an
efficient and secured National Forensic DNA Database
Management System by integrating novel advancements in
computer science field in order to speed up the traditional
electrophoresis gel image analysing process. It will help in
quick and accurate identification of criminals that leads
to speed up the crime investigation process which plays a
major role in an efficient judiciary system of a country. The
newest system will support not only to prevent criminals
from having many opportunities to commit crimes rapidly,
but also to identify potential victims of serial murders,
rapes and robberies.
In the near future, the database will populate with DNA
profiles day by day. Researches has estimated that the
stored DNA profiles of the DNA databases are increased
by about 75% in each year (Great Britain et al., 2015).
This rapidly increasing data may provide bottlenecks for
speedy retrieval and manipulation of data. Therefore,
the author infers to integrate Big Data theories and Data
Warehousing technologies to handle the large amount of
data for further work.
DNA samples that are taken form crime scenes such
as gang rapes provides a DNA profile which is called
as a mixed DNA profile. There is no proper automatic
technique to analyse those mixed DNA profiles. Finding
a novel accurate and efficient computerized solution for
mixed DNA profile analysing will be an interesting future
direction that will help leads to a great upheaval in forensic
science and computer science fields.
REFERENCES
Abeykoon A., Yapa, R.D., Sooriyapathirana S., (2013)
An Automated System for Gel image Analysis Using
Image Processing Technologies and Signal Processing
Technologies.
Bailey D.G., Christie B.C., (1994) Processing of DNA
and protein electrophoresis gels by image analysis, in:
Proceedings of the Second New Zealand Conference on
Image and Vision Computing, Palmerston North. pp. 1–2.
Genetech [WWW Document], n.d. . Genetech. URL
http://www.genetechsrilanka.com (accessed 7.1.17).
1094 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Great Britain, Home Office, National DNA Database
(Great Britain), Strategy Board, 2015. National DNA
Database Strategy Board annual report 2014/15. Stationery
Office, London.
Herisson J., Payen G., Gherbi R., (2007) A 3D pattern
matching algorithm for DNA sequences. Bioinformatics
23, 680–686. doi:10.1093/bioinformatics/btl669
Intarapanich A., Kaewkamnerd S., Shaw P.J., (2015)
Automatic DNA diagnosis for 1D gel electrophoresis
images using bio-image processing technique. BMC
Genomics 16, S15.
Kaabouch N., Schultz R.R., Milavetz B., (2007) An
analysis system for DNA gel electrophoresis images
based on automatic thresholding an enhancement, in:
Electro/Information Technology, 2007 IEEE International
Conference on. IEEE, pp. 26–31.
Kazhiyur-Mannar R., Smiraglia D.J., Plass C., (2006)
Contour area filtering of two-dimensional electrophoresis
images. Med. Image Anal. 10, 353–365.
Kiattisin S., Leelasantitham A., (2008) A Detection of
DNA Fingerprint Using Image Processing Based on
Medical Knowledge. UTCC Eng. Res. Pap.
Koprowski R., Wróbel Z., Korzyńska A., (2013) Automatic
analysis of 2D polyacrylamide gels in the diagnosis of
DNA polymorphisms. Biomed. Eng. Online 12, 68.
Lee J.D., Huang C.H., Wang N.W., (2011) Automatic DNA
sequencing for electrophoresis gels using image processing
algorithms. J. Biomed. Sci. Eng. 04, 523–528. doi:10.4236/
jbise.2011.48067.
Leung M.K.K., Delong A., Alipanahi B., (2016)
Machine Learning in Genomic Medicine: A Review of
Computational Problems and Data Sets. Proc. IEEE 104,
176–197. doi:10.1109/JPROC.2015.2494198.
Libbrecht M.W., Noble W.S., (2015) Machine learning
applications in genetics and genomics. Nat. Rev. Genet.
16, 321–332. doi:10.1038/nrg3920.
Liew A.W.C., Yan H., Yang M., (2005) Pattern recognition
techniques for the emerging field of bioinformatics: A
review. Pattern Recognit. 38, 2055–2073. doi:10.1016/j.
patcog.2005.02.019.
 ISBN number: 978-955-0301-41-6
Moreira B., Sousa A., Mendonça A.M., (2013) Automatic
Lane Segmentation in TLC Images Using the Continuous
Wavelet Transform. Comput. Math. Methods Med. 2013,
1–19. doi:10.1155/2013/218415.
Ye X., Suen C.Y., Cheriet M., (1999) A recent development
in image analysis of electrophoresis gels, in: Vision
Interface. pp. 19–21.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
ACKNOWLEDGEMET
Author would like to acknowledge the supervisor Mr.
P.P.N.V. Kumara and co-supervisor Ms. C.P. Waduge for
their motivation and inspiration that triggered throughout
the research. The author is highly indebted to Prof. A.
Manamperi for the guidance as well as the massive support
regarding the research in innumerable ways. Author
would also like to thank the technical officer and all the
staff of DNA laboratory of Faculty of Medicine, University
of Kelaniya, whose steadfast support of this research was
deeply appreciated.
10TH INTERNATIONAL RESEARCH CONFERENCE 1095
PROCEEDINGS ISBN number: 978-955-0301-41-6

ANALYSIS OF SYSTEMATIC DATA

MINING APPROACHES FOR ACHIEVING
COMPETITIVE ADVANTAGE BY
MONITORING SOCIAL MEDIA
MMF Naja1, MII Mohamed2
1
South Eastern University of Sri Lanka, Sri Lanka
2
IFS R & D International Private Limited,
Colombo - 06, Sri Lanka
# mmfnaja@seu.ac.lk

Abstract - This research paper mainly focuses at clarifying a way to analyse unstructured text content on social media
how and why social media monitoring is carried out to sites. The results shows the value of social media and
achieve competitive advantage. A systematic literature the advantages of text mining as an effective way to gain
review was used to find current insights on the methods business value from social media data available in plenty.
used, and so illustrate not only the benefits but also the Moreover over, some recommendations are also given in
difficulties that may arise during the monitoring process. this paper that aid business firms to develop competitive
Companies use social media to improve their business. strategy.
More and more companies nowadays use social media
tools such as Facebook, Twitter, linked in, you tube, Keywords - Sentiment Analysis, Text Mining, Social
Google+ etc. not only for the purpose of service providing, Network Sites, Opinion Mining, Big Data Analysis
but also to get in contact with customers.. Resulting
this, a large amount of user-generated content is freely
available on social media sites which is being utilized by
I. INTRODUCTION
business stakeholders. To increase competitive advantage
Social media have extremely changed our lives and how
and critically evaluate the competitive environment of
we interact with one another and our surrounding world.
businesses, companies are in need of monitoring and
Recent survey shows that most number of people use
analysing not only the customer-generated content in the
social media applications for various types of reasons such
company’s social media sites, but also the information
as connecting with new friends, contacting old friends,
on their competitors’ social media sites as well as the
exchanging information, and for making friends. As a
customers. In an effort to aid firms understand how
result, many business firms are adopting social media to
to perform a social media monitoring for achieving
get into this growing trend in order to achieve business
competitive advantage and convert data from social media
values such as marketing management, competitor
into knowledge that will aid decision making. This review
benchmarking, influencer identification and outreach,
paper also describes an in-detail how text mining provides
market research and reputation management.

1096 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

Social media monitoring supports typical activities such
as customer care, sales, branding (advertising, marketing,
and content delivery), and support, product development
and innovation. Social media tools that are widely used
nowadays have generated a collection of textual data,
which contain hidden knowledge about businesses to
leverage for attaining a competitive edge. In particular,
firms can go deep into the variety of amount of social media
data to discover and detect new knowledge (e.g., brand
popularity) and interesting patterns, get to know what
their competitors are doing and how the industries are
changing their trends, and use the findings and improved
understanding, to achieve competitive advantage against
their competitors (Xing & Zhan, 2015).
II. CRITICAL EVALUATION OF
LITERATURE
A.What makes social media?
however, remained shutting, mainly because of a lack of
knowledge and an unclear understanding of how social
media could be advantageous. They were trying to get to
know what social media is and why should the business
firms care about it. Unlike traditional media, that provides
a one-way experience (in which media broadcast some
news about the business for public consumption), social
media provides a two-way experience of interaction where
consumers can interact with companies. Customers in
touch with social media, unlike customers of traditional
media, can interact immediately and directly with either
the author or the originator of the proffered information.
The cross communication and interaction that social
media makes possible is the reason what makes social
media as a tool for changing the strategies in the business
that aids towards achieving business targets.

Social media platforms commonly characterized by the

profile-based communication and networking of people
via the Internet. The focus is addition to disclosing the own
personality on a user profile especially the collaborative
exchange within a larger community. Under the core
aspect of the interest-governed networking People thus
find, for example, together in thematically appropriate
groups and get the opportunity to talk about specific
issues. An important feature of this type of communication
within social media networks is the joint design of content
through the use of text as well as the aid of image, audio
and video content. Moreover, the aspect of networking Figure 1 Famous Social Media Sites and
implies the construction of secondary contacts on primary Usage of them
contacts that would come hardly possible without social
media platforms.

B. Why should business firms care about it?

Within the last one decade, social media has taken the
full control of the world. Facebook being the king of
social networking sites, it holds more than 750 million
active people using it. Even though, the majority of the
business firms did not immediately join the social media
revolution. Instead, they were focusing on some other
modes that aided them in
Figure 2 Leading Social Networking
developing and improving their businesses. But when Platform Users
the first group of companies joined, some of the rest of
the people anticipated the importance of it. Many others,

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1097
PROCEEDINGS ISBN number: 978-955-0301-41-6

Figure 1 and 2, elaborates the demand and usage of social

media among the people. That would be an ideal example
to show the extent to which people could be reached
through social media.

C. How Social Media aids to Competitive Advantage

Michael Porter’s Value Chain Analysis

One way to explore about the impact of social media within

a business firm is to know whether the organization will be
able to attain competitive advantage with its customers.

In this review article, we will use the concept of competitive

advantage as proposed by Porter in his work in 1985.
Michael Porter has described how companies can develop
sustainable competitive advantage in his publication in
1985. Porter describes three different ways that firms can
use for competitive advantage. They are Cost leadership,
Differentiation and Focus. Moreover, according to Porter’s
model, a firm’s Value Chain composed of all the activities
that are required for the production of goods and services.
These activities can be major milestone activities that
add value directly to the produced goods or services and
Figure 3 Usage of social media in Porter’s
provide support or secondary activities that add indirect
generic value chain
value to end product.
Out of all the generic value chain activities, sales and
A business firm can find out the ways of gaining marketing makes maximum out of social media. Fashion
competitive advantage by analysing its value chain. This retailers like Myntra.com and food sites like Foodpanda.
analysis is based on finding the impact of both primary in regularly use social media channels to run marketing
and secondary activities on ay things of Cost Leadership, campaigns and offer discounts (Singla, M.L & Durga,
Differentiation Advantage or Focus. Based on this analysis, A , 2015). In fact marketing campaigns does not only
the activities identified can be improved so that it leads focus on offering discounts. Many organizations segment
towards achieving competitive advantage or removing the users and target them on social media with various
competitive disadvantage. offerings as well as to collect feedback. Users who show
more interest with these brands their offerings become
We use the same logic to social media. In order to utilize influencers and customers. From that way, social media
social media to gain competitive advantage, organization’s acts as an intermediate between consumer and the firm
value chain activities that can be impacted by social media to lead acquisition channel to get more leads that are
are identified (Atzmueller & Martin, 2011). The weight of later managed in a Customer Relationship Management
impact can be measured. This can be analysed for focus, (CRM) system.
cost, as well as differentiation advantage.

List of social media activities that can be performed in

each primary activity can be proposed as follows. Similar
activities can also be extended to secondary activities as
well.

1098 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
Figure 4 Social Media Usage and its impact on
different sources of competitive advantage
Figure 5 Social Media Tools Used to achieve
business goals
D. Key features to be extracted from big data
1) Keyword searching
Organizations can define the topics to be monitored such
that it suits their brand images, products and management
features. Also keywords targeting the competitive
companies also can be defined so that the company
achieve competitive strategies over them.
2) Thematic and sentiment analysis
Data available in social media are not organized in a proper
way and it’s increasingly becoming tedious for businesses
to capture, analyse, and make use of it in a timely manner.
However, the unstructured data is not useful at all.
As a result, there is a need to study how organisations can
(a) extract and analyse social media data related to their
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
business (Sensing), and (b) utilise external intelligence
gained from social media monitoring for specific business
initiatives (Seizing) (Zhang etal., 2014).
3) Analysis of spread patterns
Monitoring is also used to know of how messages spread
in social media. One main approach is to find out causal
relations and follow the impact of the interventions of
social media. Another one is to follow the spread pattern
that makes it able to predict its development. Conversations
in social networking sites with the most potential for
generating and influencing purchasing decisions were
monitored using the analytical tools available on the sites
to analyse the spread and impact of social media messages
(Kumar & Mirchandani, 2012).
4) Share of Voice
This is referred to as measuring the work that is surrounded
by your business firm and your competitors online. By
evaluating the presence of other firms within relevant
market place, companies can well understand the success
of their own business and marketing activities.
If the competitor has a greater influence of the voice, it
may be time to increase your online presence.
5) Sentiment
By evaluating the sentiments in comments of data
available in social media, businesses can understand
the negative points consumers have in mind about their
product or service. Say if a competition company is not
acknowledging for customer complaints, then you can go
into that customer and try to introduce our products.
Also by getting to know needs and wishes of a consumer
within the industry, a business can try to avoid some
mistakes and try to improve them in their own business
strategies (Eman & Younis, 2015).
6) Top Sites and Page Types
It is very much important to know where our competitors
are active online. Firstly, it may let you know some
important sites or forums where a competition brand is
dominating its presence, that your firm not aware of. This
information will be useful to understand whether you also
need to get into that platform to increase your strength or
to be a leader in the industry.
10TH INTERNATIONAL RESEARCH CONFERENCE 1099
PROCEEDINGS
Another advantage of finding out that is to get to know
where the competitor companies are not actively
participating. It would be useful to know if that platform
is not at all needed or if that platform is the best place to
place you footprint so that without competition you can
be a leader there.
7) Influencers
By monitoring social media, you will be able to find out
influencers, who normally criticize about your product or
services or else your competitor’s. By identifying that you
will able to correct your weaknesses so that consumers will
be satisfied. Or else, by knowing competitor’s weakness
you will be able to give a better product or service to your
customers over the competitor.
8) Topics
This is finding out the online conversations about our
business or the competitor.
By finding out this you get to know about the emerging
trends in the business firms which your competitor is
using or you may use that to be the first person to provide
services using emerging trends. By finding out the new
trends, you can become the market lead.
9) Audience Type
By analysing about the people who are discussing about
your product, you will get a clear understanding about the
interested people and therefore get an idea about audience
whom you want to target. Say for an example, you get
identify which age group is mostly interested on your
products, so that you can develop strategies according to
it.
E. Development methods in social media monitoring
In academia, various methods to navigate, collect, analyse
the content, sensing and social media topics have been
proposed. Many techniques categorization unstructured
data were adapted and implemented in the social media
studies and areas of knowledge. These efforts require
interdisciplinary research skills, since they involve the
treatment of raw data regarding the search for information
quality and significance as displayed, most appropriate to
the situation. Reweaves points out that due to the volume
and complexity of data, simple models are inadequate
1100 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
and, therefore, the process of data mining of social media
requires a comprehensive approach, highlighting the
need for a unified approach to explore the data structure
for efficient identification, classification and forecasting
topics, issues and relevant models. Countless techniques
such as extraction, sorting or categorización, feature
analysis, linguistic analysis, content analysis, association
between texts, grouping and summary can be used to
extracted knowledge models or interest sentences and
unexpected textual documents. Researchers say the
clustering technique offers the advantage of revealing
trends, unexpected correlations or patterns in the
structure of the data that had not been budgeted. Scholars
emphasize that the cluster analysis is a technique well
explored in data mining and, according to Abdous, He &
Yen (2012) explored technique that allows display models,
grouping similar words and terms or values which are
also attributed & similarly coded. The same authors say
that even a perspective of data mining, clustering also
can be used for this. Introduction of Artificial intelligence
techniques to monitor social media is also a progress in
the development of social media monitoring. Some of
those type techniques are:
1) Text Mining
Data mining, text mining activities as a source of data
that accepts text work is intended to gather data on other
defined structured text. For example, the classification
of text, segmentation of, subject extraction from text,
production of class particle, emotional analysis, text
summarization, and entity relationship modeling aims
such studies.
Bringing information within the scope of the text mining
efforts in order to achieve the above goals (information
retrieval), syllable analysis (lexical analysis), word
frequency distribution, pattern recognition, information
Extraction, data mining and even visualization (Wei Jin
etal).
Text mining studies aim to achieve results through the
text more statistically. Often for feature extraction using
natural language processing, text mining studies are used.
Introduced in 1995 by Ronen Feldman and Ido Dagan as
“Knowledge Discovery from Text (KDT)”. In the research
terminology text mining is not a clearly defined term.
In analogy to data mining in the Knowledge Discovery
in Databases (KDD) Text Mining is a largely automated
process of knowledge discovery in textual data, which is
 ISBN number: 978-955-0301-41-6
to enable an effective and efficient use of available text
archives. Comprehensively we can say text mining as a
process of compilation and organization of the formal
structuring and algorithmic analysis of large collections of
documents for the appropriate extraction of information
and the detection of concealed content relationships
between texts and text fragments.
The different views of text mining can be sorted by various
typologies. These kinds of information retrieval (IR), the
document clustering, the text data mining and KDD are
repeatedly referred to as sub-forms of text mining.
Most similar to the text mining, data mining with this, it
shares many methods, but not the subject matter: While
data mining is usually applied to highly structured data,
text deal mining with significantly weaker structured text
data. In text mining, the primary data are therefore in a
first step, more structured to allow its development with
methods of data mining. Unlike most tasks of data mining,
multiple classifications are usually welcomed in the text
mining (Miloš & Milana, 2008).
2) Web Mining
Web mining is a process used for meaningful patterns
from Web resources, profiles, physician data mining
techniques (to find a useful correlation is hidden among
other data, to find such a trend Extract the actionable
information for future decisions ) to be applied once.
Its application fields include information filtering and
monitoring of such competitors as patents and technology
development, mining utilization of the web access logs for
analysis, browsing (breadcrumb navigation in your web),
etc. support.
Web mining is part of the data mining specialist for
disclosure of information on the Internet, especially the
World Wide Web. Huge databases are rich in data, as
well as poor information that is hidden in stored data.
Data mining helps to reveal important information and
knowledge applied to the data.
Also, greatly contributes to decision-making, business and
science. Mining the Web is used by a variety of companies
(personalized marketing), as well as governmental
organizations (fight against terrorism). Companies can
establish better relationships with their customers because
they can better meet the needs of consumers, and react
more quickly to their queries. There are also problems
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
in mining web. The technology Web mining does not
cause any problems, while the use of these technologies
in the personal data can do damage. The largest such
issue would be invasion of privacy. Privacy is violated
when information about an individual obtained, used
and disseminated without his knowledge or permission.
Another such problem is that companies that collect data
for a specific purpose, such data can be used for something
completely different. The truth is that these companies
are responsible for the issuance of all of this data, and if
you notice some irregularities followed by a lawsuit, but
there is no law that prevents them from trading in such
data (Sheibani & Amir). As the popularity of the World
Wide Web continues to grow, there is a growing need to
develop new tools and techniques that will improve the
overall usefulness.
3) Named Entity Recognition
The named entity recognition is a subtask of information
extraction activity in the documentary corpus. It is to look
for text objects (that is to say a word or group of words)
be categorized into classes such as personal names, names
of organizations or businesses, place names, quantities,
distances , values, dates, etc.
For example, you could give the following text, labeled by
an entity named recognition system used in a survey.
Jim went to see Kim at 9:00 of May 18.
This text can be recognized as<PERSON>Jim</
PERSON> went to see <PERSON>Kim</PERSON> at
<TIME>9:00</TIME> of <DATE>May 18</DATE>.
The text of this example is tagged with XML tags. Most
labeling systems use formal grammars associated with
statistical models, possibly supplemented by databases
(lists names, names of cities or countries, for example).
In large surveys, manually written grammars database
systems perform best. The downside is that such systems
sometimes require months of drafting work.
Current statistical systems use to share a large amount
of pre-annotated data to learn the possible forms named
entities. There is no need here to write many rules in
hand, but to label a corpus that will serve as a learning
tool. These systems are also very costly in human time. To
solve this problem, recently, initiatives such as DBpedia
Yago or seek to provide semantic corpus that could help
design labeling tools. In the same spirit, some semantic
10TH INTERNATIONAL RESEARCH CONFERENCE 1101
PROCEEDINGS
ontology such as NLGbAse is largely oriented labeling.
Since 1998, the annotation of named entities in texts met
with increasing interest. There are many applications call
for finding information or, more generally understanding
of textual records.
There is not strictly speaking labeling standards. Labels
are heavily skewed depending on the application needs:
we usually find the label type classes roots Person,
Organization, Product, Place, plus the label length and
quantity (time and amount).
F. How these techniques are implemented in social
media monitoring
Social Media monitoring can be broadly defined as using
Social Media data sources, many of which can be taken
automatically using some extracting tools, and choosing
and aiming concepts and word types to analyse market
behaviour and interaction(Sponder & Marshall, 2012).
Thus, data collected from this sort can be summarized,
analysed and classified and passed through various types
of algorithms to get patterns and trends that are useful to
achieve business goals. In order to do this, the social media
need to be monitored on a specific time basis and filter the
useful things and wan to decide which is worth to consider
upon. This is known as social media monitoring.
Monitoring social media using text mining has been
proposed by various researchers as depicted in following
pictures.
Figure 6 Implementing text mining
1102 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Figure 7 Big data analysis of Social Media
III. CONCLUSION AND
RECOMMENDATIONS
Most business firms, government agencies, non-profit
organizations (e.g.: Hospitals, universities, associations,
etc.) and individuals, such as writers and celebrities use
media monitoring as a tool for identifying mentions
of their organization, their brands and leaders in
news media. some organizations also provide media
monitoring tools to track the success of their messages to
find information about competitors and relevant specific
issues on the organization to manage the benchmark
performance against competitors, corporate or brand
image, to gather the information industry to create a
better to understand the strengths and weaknesses of
corporate communications, to identifying new business
opportunities, and for other purposes.
Despite the initially purely private use of social networks
were in the past few years an increasing number of private
companies about to adapt their presence on the Internet
to Web 2.0 developments to benefit from the advantages.
Primarily social media provide platforms for companies,
a relatively inexpensive but accurate marketing tool
because of the wastage of classical forms of marketing
can be significantly reduced by targeting of customers
and potential customers. Furthermore, the operation of
a social media concept supports the development of new
target groups, mainly the move on the internet and can
hardly be achieved through traditional channels. In this
context, especially the resulting economic potential of
viral marketing is to be mentioned the results from the
typical social media concept of networking between users.
Rather than only monitoring social network sites, I
recommend that the use of internal networks can also
be good for performance improvement of a company
which leads to gaining competitive advantage. Such a
system is not only used to improve the working climate
by facilitating and more informal communication among
the employees, but also helps to uncover organizational
weaknesses and eliminate them.
 ISBN number: 978-955-0301-41-6
ACKNOWLEDGMENT
Our special thank goes to the academic staff of the
Department of ICT of the South Eastern University
of Sri Lanka for their continuous support given for us
for working on this research. Also a heartfelt thank is
dedicated to the team “ Business Intelligence” of IFS R&D
International Pvt Limited for supporting to make this
research work a success.
REFERENCES
Wei Jin, Hung Hay Ho and Rohini K. Srihari. 2009.
Opinion Miner: a novel machine learning system for web
opinion mining and extraction. Proc. of the 15th ACM
SIGKDD, 1195–1204.
Eman M.G. Younis . 2015.Sentiment Analysis and Text
Mining for Social Media Microblogs using Open Source
Tools: An Empirical Study, Faculty of Computer and
Information Minia University, Egypt International Journal
of Computer Applications (0975 – 8887) Volume 112.
Riswana Irfan. 2004. A Survey on Text Mining in Social
Networks, The Knowledge Engineering Review, Vol.
00:0, 1–24. Cambridge University Press DOI: 10.1017/
S000000000000000 Printed in the United Kingdom.
Miloš Radovanović and Mirjana Ivanović. 2008. TEXT
MINING:APPROACHES AND APPLICATIONS, Vol.
38, No. 3 , 227-234.
Abdous, M and Yen,H.. 2012.Using Data Mining for
Predicting Relationships between Online Question Theme
and Final Grade. Educational Technology & Society, 15
(3), 77–88.
Ruggiero, A and Vos, M. 2014. Social Media Monitoring
for Crisis Communication: Process, Methods and
Trends in the Scientific Literature. Online Journal of
Communication and Media TechnologiesVolume: 4 –
Issue: 1.
Atzmueller, Martin. Modeling and Mining Ubiquitous
Social Media International Workshops MSM 2011,
Boston, MA, USA, October 9, 2011, and MUSE 2011,
Athens, Greece, September 5, 2011, Revised Selected
Papers. Berlin: Springer, 2012.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Fang, Xing, and Justin Zhan. “Sentiment Analysis Using
Product Review Data.” Journal of Big Data (2015). Han,
Jungkyu, and Min Luo. “Bootstrapping K-means for Big
Data Analysis.” 2014 IEEE International Conference on
Big Data (Big Data).
He, Wu, Shenghua Zha, and Ling Li. “Social Media
Competitive Analysis and Text Mining: A Case Study in
the Pizza Industry.” International Journal of Information
Management: 464-72.
Hu, Xia, and Huan Liu. “Text Analytics in Social Media.”
Mining Text Data: 385-414.
Meyrowitz, Joshua. No Sense of Place: The Impact of
Electronic Media on Social Behavior. New York: Oxford
UP, 1985.
“Patent Issued for Signal Processing Approach to
Sentiment Analysis for Entities in Documents.” Journal of
Engineering 28 May 2014.
Rastunkov, Vladimir, and Mark Rusch. “Using
STATISTICA Text Miner to Monitor and Predict Success
of Marketing Campaigns Based on Social Media Data.”
Practical Text Mining and Statistical Analysis for Non-
structured Text Data Applications: 151-80.
Sheibani, Amir A. “Opinion Mining and Opinion Spam:
A Literature Review Focusing on Product Reviews.” 6th
International Symposium on Telecommunications (IST).
Sponder, Marshall. Social Media Analytics: Effective Tools
for Building, Interpreting, and Using Metrics. New York:
McGraw-Hill, 2012.
10TH INTERNATIONAL RESEARCH CONFERENCE 1103
PROCEEDINGS
DOMESTIC SERVICES MANAGEMENT SYSTEM:
A SRI LANKAN PERSPECTIVE
SHYS Wickramaratne1#, N Wedasinghe 1,
1
Faculty of Computing, General Sir John Kotelawala Defence University,
Ratmalana, Sri Lanka
# yasiru_first@hotmail.com
Abstract – Technology has granted many opportunities
for the people in need of employment around the world.
It has created marketplaces as well as marketspaces
providing a stable economic benefit. However, in this
modern society which have plethora types of media &
mass communication approaches, so far it still failed
to deliver swift and convenient methods to request
domestic services. Home repairing services, beauty, and
health-related services, events organizing related services
and business services are few services still discoverable
through newspapers and yellow pages in Sri Lanka. This
leaves a necessity to create a solid platform for the skilled
micro employees to provide their services continually as
per the customers’ services requests. A mobile App-based
service marketplace will be a good enhancement for the
current context to locate, categorize, and provide rates and
feedback about nearby micro employees who are willing
to provide their services for the domestic customer’s
request. This system will categorize their jobs and its work
providing more niches to the customer’s request. The
main objective of this paper is to emphasize that mobile
technologies can be used to enhance the human life by
helping them to hire services using a convenient platform.
As for the results, this system can influence the 4.7 of an
unemployment rate of Sri Lanka (Labour Force Survey,
2015) by creating a separate easily accessible marketplace
without any entry barriers for the micro employees who
are willing to start their own businesses.
Key Words - Smartphone, Services, Domestic, Micro-
Employees, Android, Unemployment, Cloud-Platform
1104 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
I. INTRODUCTION
The growing capacity of smartphones and in related
technologies have opened up a vast new avenue for the
communication and e-commerce sectors, prompting
many to evolve from using desktop or laptop computers
to handheld mobile devices. 2015 Central Bank reports
show that every 100 people in Sri Lanka own 113 mobile
phones and 13 fixed lines by the end of the year 2015
(Annual Report Central Bank,2015). And latest statistics
issued by the Internet World Stats emphasize 27.4% of the
Sri Lankan population has access to the internet (IWS,
2016).And this statistics correlates that people are using
mobile devices as a simplicity to achieve their daily tasks
more and more. The application of mobile apps to daily
activities are not entirely new to this generation. People
tend to buy goods more through mobile apps compared
to past decade. But unfortunately in Sri Lanka for the
service sector, there are limited ways to request services
for any domestic needs. This Domestic Services Support
System – Android App will be a good addition for the
people to interact and find about more and more about
what kinds of services are available out there or nearby.
These applications will be a native Android app which will
run on any smartphone running on Android OS without a
hassle to use a web application separately. The back end of
this application will be controlled by the Google firebase
cloud server which gives more control over the application
which performs following tasks. And there will be two
separate apps for customers and service providers to make
it more comfortable and convenient to operate their tasks.
 ISBN number: 978-955-0301-41-6
Domestic Support System includes following features:
• Friendly in UI and UX sections
Attractive and user-friendly interfaces will guide the users
and service providers to use the application more often.
Both the apps will start with introductory slides to educate
the parties on apps features.
• User Base Authentication
The service provider identification will be thoroughly
checked according to their information provided as for
the security measures.
• Categorized job market
In the current service sector of Sri Lanka propose the
wide variety of services to the market such as health care,
cleaning, tradesman ships, consultancy, builders etc. By
using this application the customers can access and choose
the exact service for their needs.
• Connects parties more efficiently
The customer app will provide the nearest service provider
according to their service request, making more time
efficient and less stressful for the customer. The customer
location will be taken to determine the service provider
location around him.
• Rate service providers
After the service or services are provided the customer
can give rates for a service provider according to his or
her performance. This option will be more helpful for the
other customers to decide whether they need the service
from that particular service provider.
• Real-time service provider tracking
This application will provide real-time tracking option for
the customer to keep track of the service provider until he
or she arrives at their premises.
• Feedback Monitoring
The customers can also add feedbacks about the service
providers’ performance along with the rating. This open
opinions will help other customers to decide their options
available more straightforwardly. The feedback will also be
useful for the customer to make quick accurate decisions.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
• Creates professional labor pool
In the long run of this application can actually help the
micro employees to become more work efficient at their
work and be professionals.
II. RELATED WORKS
A. Services Vs Goods
Goods and services are the main things people request
in their daily economic life. Sri Lanka shows a very high
growth rate in the service sector and its holds 62.5% out
from other main economic activities. (Annual Report,
2015) IT knowledge advancements and computer literacy
increment have been a direct influence on its growth.
Goods are sold around the country vastly using many
media (Internet, TV, Internet, Mobile apps). We could find
many mobile apps for websites such as eBay, AliExpress,
Amazon etc. (Ebayinc.com, 2017). But unfortunately,
there is no single mobile application in Sri Lanka to
provide categorization for Micro employees. However, this
system fills out the void that created it. Providing a better
marketspace and a marketplace to both of the parties.
B. Server selection
Most of the mobile app developers are tended to use
Amazon Web Services (AWS) as the backbone to their
mobile app nowadays. It provides quite a number of
options for the developers serving as a cloud computing
structure. And this is the backbone of Urbanclap (similar
services providing app). Even though this has so many
features it lacks library updates and deprecated functions.
Here in this Domestic Services Support System as the
backbone it will run along with Google Firebase. The
Firebase was recently acquired by the Google and it
provides seamless support for the app development in
many platforms (Android, iOS, C++, Unity).
C. Database
A Database is an organized collection of data. Databases
can be stored locally on your computer or can be stored
in cloud storages. Every application whether Android,
iOS or web application, it has its own database (W. Cao,
2014). In the Android app, we can create databases using
SQLite, shared preferences, websites or some cloud-based
storage sites. Firebase is also providing database structure
10TH INTERNATIONAL RESEARCH CONFERENCE 1105
PROCEEDINGS
backend for Android, iOS and web applications (Firebase,
Online). Using only the inbuilt SQLite functions of
Android apps is not enough for an app. It is because in the
long run the stability of the app gets affected by the data
accumulated on a daily basis. As for the solution, Google
Firebase database will be used which is NoSQL and
scalable database (C. WODEHOUSE, 2017). And it will
not use any query based SQL languages instead it will use
JSON nodes for index information (Json.org, 2017). This
real-time database will use cloud technologies to share
information across the applications. The moment the data
being updated the app in that instance will be updated
according to the data entered.
D. Convenience
A limited number of websites are available for micro
employees in Sri Lanka such as hodabaas, Mason,
Salondhammi and findbas. But unfortunately, all of
these sites are restricted to their own categories and the
categories are also limited (e.g.: Beauty and Construction)
But Domestic Support System will combine all the
available services together to one place and improve the
convenience to the customers to access them by using
a single mobile app. This system will look into more
market niches. This includes Entertainment services,
Event services, Health related services, Education related
services (Piano lessons, Guitar lessons and etc.)And much
more other services out there in Sri Lanka. (Hodabass.lk,
2017) (findabaas.lk, 2017) (findbass.lk) (Salondhammi.
com, 2017) (masons.lk, 2017)
E. Authenticity
As we search through the web we can find many people
who are willing to provide their services. But when it comes
to services sector security of the people is a major concern
for the developers if they are going to add service provider
information’s to the system. That is why this system has
introduced two separate mobile applications. One for the
consumer and another app for the service provider. By
using the service provider app, the service providers can
list their service to the customer. The mentioned process
will undergo go many authentication steps. Unlike above
websites, this system will use OAuth 2.0 (Oauth.net, 2017)
and OpenID (Openid.net, 2017) which will improve the
security along with this process time. Most of the service
providing websites only request National Identity card
information. This will create questions at the service
providers’ knowledge on their field of work. This can be
1106 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
overcome in this system due to its thorough processes.
Unlike in customer’s app, the service provider app will
include security checks and information validations.
F. Real-time Tracking
After ordering a good/thing it is a normal to wait till it
arrives. Likewise, customers will observe the arrival of
the service provider. In Urbanclap app (Urbanclap.com,
2017) after the customer request a service they wouldn’t
able to find out whether the service provider is going
to provide their service certainly. But in this Domestic
Services Support System, it will be integrated with features
to track the real-time location of the service provider. This
will help customers to proceed to their other duties until
the service provider arrives (M Singhal, 2012). By using
Google Maps API and new Google Maps location share
functions the whereabouts of the service provider can be
easily tracked. Unlike those websites, this app will search
the surrounding area by Google location services not
showing entire labor workforce to that area.
G. Rating
Rating, voting and other feedback mechanisms are heavily
used in today’s mobile apps, app markets (Google Play)
and in social media systems, allowing users to express
opinions about the content they used and still using. Web
sites providing odd jobs and other services do not provide
workers past worked history information or customer
experience by any sort of ratings and feedbacks. These
problems can be neutralized by Android native rating
libraries and integrating feedback features.
III. ANALYSIS & FINDINGS
To examine the demand for the services a face-to-face
interview was conducted around the Galabadha Waththa,
Malapall, and Kottawa area. In this interview, most of the
people have mentioned the difficulty in finding a skilled
worker for their needs. The majority of them rely on
newspaper and neighborhood community to find a worker
(J. Cheng, 2014). This current process consumes time
for the customer. And because of this service providers
scarcity, the customer may have to pay a higher price for
their work they trying to fulfill.
This proves that people who are willing to request services
are going for this sort of extent to find their suitable worker.
Similarly, there is also service bill for guitar lessons,
individual tuition classes, piano lessons, event planning
consultancies.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

IV. METHODOLOGY be written using Java language (Docs.oracle.com, 2017).

Extensible mark-up language (XML) will be used to
A. Logical architecture design the overall user interface of both of the Android
apps. Google firebase will hold its own real-time database
The user-friendly user interface will be the main concern procedures to update the customer requests and service
of this application because by making a good impression provider.
on the people at the first glance, can improve the usage
of this mobile application. And Hassle-free registration
process along with service requesting process is the second
highest prioritized function will be running inside this
application. The following three-tier architecture diagram
summarizes the system’s basic processes (Fig 1).

As mentioned two separate mobile apps coordinate

the two parties. As shown in above architecture. At the
beginning, the customer requests the services through
a mobile application and from application layer it will
redirect into services and database layer. In this layer, all
the JSON nodes requested through application layer will
be submitted and requested output will be generated.
Google Firebase will then redirect this request to the
service provider through service provider application. In
the end, the service provider read the request and confirm
the request to provide services. The admin monitoring
presence or the Google firebase dashboard will help the
developers to monitor and analyse service providers and
customers’ application usage history. And admin will be
in charge of removing any suspicious or unskilled workers
if any customer complaints him or her through the rating
system or from feedbacks.
Figure 2. Interface of the Android app
C. Database Tier
The structure of this applications’ database will be
different from traditional databases. Because the structure
of this database will not use traditional relational database
schema. Instead, the structure of this database will be
built using document-oriented database architecture.
The following (Fig 3) brief about how data stored in the
relational database and non-relational database differ
from each other. (C. WODEHOUSE, 2017)

Figure 1 Basic System Architecture

These two mobile apps are developed using Android

Studio. Which is an integrated development environment Figure 3. Firebase
(IDE) created by Google Inc. (Developer.android.com,
2017) Java, XML and SQL languages will run in these
apps core. Most of this application’s core functions will

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1107
PROCEEDINGS
The following reasons have influenced over the selection
of NoSQL database. (S. Chickerur, 2015)
• Data Structure Issues
Having many different types of data can be influence the
overall structure of the database. In NoSQL database this
issues may occur but with document oriented databases
this issue can be overcome by storing data in one place
without the need of defining the type of data it has to store
in.
• Cloud computing benefits
Google firebase supports development of the mobile app
with many good features. It provides real-time database,
file storage and hosting services in one single place. With
this scaling up the database will not be an issue. Mostly
the cost will be effectively affected, this would help to save
considerable cost.
• Rapid development.
If the development of the app within a limited time frame,
turning out quick iterations or requiring to make regular
updates to the data structure without a lot of interruption
between versions, a relational database will slow you down.
Document-oriented database does not need to be prepared
ahead of time.
V. DISCUSSION
The idea for this application was first laid out by the world’s
corporate giants like eBay, Amazon, etc. They successfully
provided a solid base for the sellers and consumers to
come out and sell their products more freely. But they also
trying to fills the gap between the needs of services to the
customer. In Sri Lankan context also we can see this void
located in heavily E-commerce sector. In order to set itself
apart from traditional classified sites and other directories,
this application will provide matchmaking results for the
JSON nodes requested that guides and help users to sort
through a large array of services in their city, but also make
sure providers get good leads that result in good business
connections. This will make very simple and convenient
to the user to hire a trusted service professional. Thus it is
essential to develop a solid index to match corresponding
service provider to the needs of the requesting customer.
The mentioned technologies and mentioned application
1108 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
of android technology and its combined technologies have
proven to be very resourceful in real life. Many positive
outcomes were achieved by this application of android
technology.
V. CONCLUSION
The demand for the customers’ service request will
continually grow along with the development of the
country. By using Domestic Services Support System,
customers will be able to find individuals with the
experience and knowledge (Professionals) regarding the
service they require more conveniently in a shorter period
of time. And furthermore, it brings the portability to the
users saving their time and money. The unemployment
issues also will be addressed through this system and
problems of micro employees will be considerably
reduced. For the upcoming technologies, these Android-
based apps can be implemented and upgraded as it grows.
REFERENCES
Abraham E. Eviwiekpaefe, “The Trend and Challenges of
Cloud Computing: A Literature Review”. International
Letters of Social and Humanistic Sciences; Vol. 16, pp. 13-
20, 2014
Android guide Firebase <https://firebase.google.com/
docs/>
Amazon Web Services<https://aws.amazon.com/what-is-
aws/> “Annual Report 2015”, Cbsl.gov.lk, 2017. [Online].
Available: <http://www.cbsl.gov.lk/pics_n_docs/10_pub/_
docs/efr/annual_report/AR2015/English/content.htm>.
[Accessed: 15- Apr- 2017].
C. Electrical wiring for House and Industries, P. PPR,
T. service, H. Experts, C. Installation, K. Services, H.
Welding Bass, E. Electrical and D. Engineering, “Home”,
Find A Baas!, 2017. [Online]. Available: <http://findabaas.
lk/>. [Accessed: 25- Mar- 2017].
J. Cheng, C. Danescu-Niculescu-Mizil, and J. Leskovec,
“How community feedback shapes user behavior,” arXiv
preprint arXiv: 1405.1429, 2014.
“Sri Lanka Labour Force Survey Annual Report”, http://
www.statistics.gov.lk, 2017. [Online]. Available: <http://
www.statistics.gov.lk/samplesurvey/LFS_Annual%20
Report_2015.pdf>. [Accessed: 15- Apr- 2017].
 ISBN number: 978-955-0301-41-6
“Masons!”, Masons!, 2017. [Online]. Available: <http://
www.masons.lk/>. [Accessed: 26- Mar- 2017].
Manav Singhal, Anupam Shukla ‘Implementation of
Location-based Services in Android using GPS and
Web Services ‘, IJCSIInternational Journal of Computer
*Science Issues, Vol. 9, Issue 1, No 2, January 2012 ISSN
(Online): 1694-0814
“Introduction- OAuth”, Oauth.net, 2017. [Online].
Available:https://oauth.net/about/introduction/.
[Accessed: 26- Mar- 2017].
T. Xiteb, “Hodabass.lk - The BEST place to FIND a Bass!”
Hodabass.lk, 2017. [Online]. Available: http://hodabass.
lk/. [Accessed: 25- Mar- 2017].
Urbanclap.com, 2017. [Online]. Available: https://www.
urbanclap.com/about. [Accessed: 26- Mar- 2017].
W. Ltd, “Salon Dhammi | Professional Bridal Service in
Sri Lanka”, Salondhammi.com, 2017. [Online]. Available:
http://www.salondhammi.com/. [Accessed: 26- Mar-
2017].
“What is OpenID? | OpenID”, Openid.net, 2017. [Online].
Available:http://openid.net/get-an-openid/what-is-
openid/. [Accessed: 26- Mar- 2017].
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
“Who We Are - eBay Inc.”, Ebayinc.com, 2017. [Online].
Available: https://www.ebayinc.com/our-company/who-
we-are/. [Accessed: 25- Mar- 2017].
M. Studio, “Meet Android Studio | Android Studio”,
Developer.android.com, 2017. [Online]. Available:
https://developer.android.com/studio/intro/index.html.
[Accessed: 16- April- 2017].
“Java Platform SE 8”, Docs.oracle.com, 2017. [Online].
Available: http://docs.oracle.com/javase/8/docs/api/index.
html. [Accessed: 01- May- 2017].
“JSON”, Json.org, 2017. [Online]. Available: http://www.
json.org/. [Accessed: 01- May- 2017].
C. WODEHOUSE, “SQL vs. NoSQL: What’s the
difference?” Hiring | Upwork, 2017. [Online]. Available:
https://www.upwork.com/hiring/data/sql-vs-nosql-
databases-whats-the-difference/. [Accessed: 22- April-
2017].
S. Chickerur, A. Goudar and A. Kinnerkar, “Comparison
of Relational Database with Document-Oriented
Database (MongoDB) for Big Data Applications,” 2015
8th International Conference on Advanced Software
Engineering & Its Applications (ASEA), Jeju, 2015, pp. 41-
47.Doi: 10.1109/ASEA.2015.19
10TH INTERNATIONAL RESEARCH CONFERENCE 1109
PROCEEDINGS ISBN number: 978-955-0301-41-6

AN ONLINE HELP DESK SYSTEM TO

HELP STUDENTS IN LEARNING
MND Salgado1 #, PU Charuka1, KP Bashitha1,
SC Eshwarage1 and MKA Ariyaratne1
Faculty of Computing,
1

General Sir John Kotelawala Defence University, Ratmalana, Sri Lanka

# nipunisalgado93@gmail.com

Abstract -This paper reports about a field study done on I. INTRODUCTION

online helpdesk system for General Sir John Kotelawala
Defence University (KDU) and about the system prototype
which was built according to the analysis of gathered data.
When considering the current learning process prevailing
at KDU, when participating in lectures, students do not
get a chance to present their problems to the lecturers very
often. Therefore, we address the problem of not having
an opportunity for the students, to solve their problems
that arise during the lectures. To overcome this problem,
a Web Based Help Desk System was introduced for KDU.
Through this Online Help Desk System, students will get
the opportunity to present their problems to the relevant
lecturers, whenever they need a further clarification
on any subject material. Students are facilitated by the
option to choose the priority level of a question. So, unlike
other help desk systems, this system will ensure that each
question posted on the system will be answered as soon
as possible according to the priority level preferred by
the student, by notifying the lecturers by an email. The
accuracy of the provided answers will be assured since,
only the authorized lecturers of KDU have the eligibility to
get registered to the system as an answer provider. When
developing this Online Help Desk System, the focus was
given to build a system, which will give more time critical
and task sensitive online help.
Keywords - Online Help Desk System, Question and
Answer System, Web Based Application
A. Prolegomena
In the modern world with the development of technology
in the area of learning, e – learning has gain a prominent
place. Among them, more attraction has been drawn
towards Online Help Desk Systems. This can also be known
as Online Question and Answer Systems. Online help
desk systems can bridge the gap between the user’s need
for simplicity and the software’s complexity, by providing
relevant information within the very specific context of a
user’s lack of familiarity with a particular concept. This is
a common definition about the online help desk systems.
When talking about help desk systems, at once the idea
that comes up to our mind is the help desk system that is
available in every software and systems. Those help desk
systems will provide solutions for the problems regarding
the usage of system and services rendered by the systems.
However, in this research, the main concern was drawn
towards the implementation of an online help desk system
in an educational context. That means this system will help
the students to solve their problems that arise during their
lectures regarding subject matters. Therefore, with the help
of this system, students will get the opportunity to clarify
their doubts regarding the subject matters whenever they
need.

1110 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
B. Background and Motivation
Online Help Desk System is a research area that has recently
gained a lot of interest. A large number of universities and
researchers have done research on this topic. At the early
stages of this research area, more attention has been given
to develop this system only concerning the system point
of view. But Ye Chen, a PhD student in the Department of
Information Systems at University of Maryland has done
a research on developing a Help Desk System concerning
also the organizational factors and human behavioural
factors and not only the designing perspective. In his
research he argues that organizational and human
behavioural factors are also important and should take
into consideration to provide the most appropriate service
according to what actually the users expect from online
help desk system. Also he has done this research on a
context of an educational institute. (Chen, 2004a) This was
the main motivation factor which lead to do this research.
C. Problem in Brief
The Department of IT in KDU Rathmalana offers two
degrees. They are the BSc in Information Systems and BSc
in Information Technology. Currently the Department of
IT has three intakes with more than 100 students. Initially
the team has built this system for the students of all
three intakes who are reading for the BSc in Information
Technology Degree.
In each intake, there are a considerable number of
students. Each day the students will have two lecture
sessions, from 8.00 am to 2.15 pm. One lecture session will
be three hours long. When the lecturers are conducting
the lectures, the students will get many problems regarding
the particular subject. Due to the lack of time and since
there is a substantial number of students for a single
lecture session, all the students might not get a chance to
present their problems to the lecturer. Sometimes some
students hesitate to present their problems to the lecturers
in person because they are shy or less confident. Moreover,
because of the busy schedules of the lecturers, the students
will not be able to meet the lecturers even after the lecture
hours to get solutions for their questions. Also, while the
students engage in their further studying at home they
might get doubts. But they will not have an opportunity
to solve those doubts even by surfing the internet, since
the accuracy of the answers that can be gained by various
other websites cannot be assured. These are the main
problems that the students face when they engage in their
studies.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
To address the above mentioned problems the team built
a Help Desk System which will be fully web based and
automated. Therefore, the students will be able to present
their problems to the lecturers from anywhere anytime
and also the lecturers will be able to provide answers for
those questions from anywhere, anytime. Simply, this
system is a platform which connects the students and the
lecturers, for the benefit of the students.
The objectives that were expected to achieve by doing this
research were, to conduct a critical review on technologies
used in currently available Online Helpdesk Systems
and to design and develop an Online Helpdesk System
for the students of the IT Department of KDU, which
will be beneficial to encourage the students to engage in
their studies without doubts and with more clarifications
and understanding, as well as to increase the interest in
learning.
II. LITERATURE REVIEW
A. Introduction
In this section, we critically review the research conducted
in web based helpdesk systems. This review has organized
in three sections namely early developments, recent
advancements and future direction. Finally, this section
presents the research problem and the technology that
need to be adopt to solve the problem.
B. Early Developments
A help desk system can be described as “The process of
handling unstructured data in a structured manner”. As
stated by Gyll and Gyll (Gyll and Gyll, 2003) a general
definition of a help desk system is, to be able to create
a stable environment where it is easy to make decisions
and find solutions to problems. Help Desk Systems has
developed in order to help the organizations to effectively
communicate, organize, track and report issues. The
importance of web based help desk systems is that the
organizations can access information at anytime and
anywhere(Hafifi and Ariza, 2012).
A technical definition of a help desk system is, it is a
collection of one or more programs designed to provide
user assistance embedded in a larger program or computer
system. Although designers frequently integrated help
programs with the application, help systems might also
be separate and run concurrently with the system (Chen,
2004b). The issue in this definition is its narrow scope due
to its application-oriented nature.
10TH INTERNATIONAL RESEARCH CONFERENCE 1111
PROCEEDINGS
Along with the prevalence of a diverse and heterogeneous
range of software and hardware systems, an online help
system serves as a gateway to all kinds of computing
resources. With the massive development of technology,
online help systems are also developed largely. The
evolution of help systems has come from book oriented
to constructive through exploratory. Book oriented online
help represented paper-based documentation, which
highlighted static and linear structure. Although this
traditional organization of contents was familiar to users,
the limitation of this approach was, not quite capable for
satisfying the complicated information needs of particular
users in specific contexts for distinctive tasks. Due to
this it paved the way for hypertext-based online help,
either exploratory or constructive. The significance of
exploratory online help was that it provided multiple paths
to navigate a document thus increasing the interactivity.
Its flaws were unfamiliarity to users and difficulty to read
through. Constructive online help allowed customization
by users, feedback to developers, and capability to re-
conceptualize tasks. This online help conception evolved
along the dimension of user-document interactivity(Chen,
2004c).
Due to these reasons, it is clear that online help systems
should be considered in an organizational context rather
than from an application-oriented context.
According to Sondheimer and Relles (1982) online help
systems are categorized into four dimensions,
(1) Access method - How users input help requests.
(2) Data structure - How the help information is organized.
(3) Software architecture - How users and the help systems
interact.
(4) Contextual knowledge - To which extent the help
information is relevant to the environment and the
tasks of a specific user.
But this categorization had some drawbacks. Bergman
and Keene-Moore (1985) argued that one shortcoming of
this categorization was that these dimensions only took
into consideration software related aspects, omitting user
interface related factors.(Moré et al., 2012)
C. Recent Advancements
In the recent advancements of online help systems
educational institutions has started to incorporate this help
system with e learning by integrating them as question and
answering systems.
1112 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Universities or higher education institutions have started
introducing the open source software and practical
experiences have been reported by them in various
scientific journals. Introductory examples of e-Learning
and their effectiveness have been reported. Construction
and application of an e-Learning environment are often
based on open source packages, such as Moodle, Stack
Overflow, Yahoo!Answers, wikiAnswers etc... (Brill et al.,
2002)
Since e-Learning offers a learning environment which
exceeds spatial and time-based restrictions due to the
use of web technologies, its usage has been continuously
increasing every year(Lebedeva and Zaitseva, 2014). All
information systems built for the purpose of educational
support are summarized under the term “e-Learning”.
Installation and operation of an e-Learning environment
are already playing an important role in educational
facilities, such as universities and anyone can install an
e-Learning system without much effort. (Moré et al., 2012)
Although many e-Learning/question and answer systems
have been implemented by Universities around the world,
these types of systems are not being implemented in Sri
Lankan Universities. This helpdesk system which is to
be implemented have new options such as, notifying the
relevant lecturers according to the relevant categories and
also an option where an e-mail is to be sent to the lecturer
if a student submits an urgent question which we do not
frequently come up with other e-Learning systems. This is
the novelty of this system.
D. Future Trends
The future trends of e – learning systems include
intelligent question answering systems. QA systems have
developed over the past few decades until they reached
the structure that we have nowadays. QA systems, have a
backbone composed of three main parts namely question
classification, information retrieval, and answer extraction.
Therefore, each of these three components attracted the
attention of QA researchers(Allam and Haggag, 2012).
The main challenge that is faced by the future help
desk systems is that how to make it flexible with the
changing technology and thereby provide customers
with a comfortable and reliable customer support which
is based on customer centric approach. The solution for
this challenge is implementing them in a way such that
they will accompany the changes as well as the growth of
 ISBN number: 978-955-0301-41-6
technology (Barskar et al., 2012) (Fukumoto et al., 2013).
The fast rise of information technology, mobility of users,
and security as well as compliance regulations are affecting
the future of help desk more than ever. In this regard, help
desk software programs are viewed to progress towards
an operationally focused management that gives primary
importance on the main competencies of customer care.
With customers getting bolder and technology-savvier,
improving in this regard is seen as necessary. (Ómarsson,
2010)
Changes in technologies are driving many changes in the
help desk. Better help desk software functionalities are
viewed to possibly modify, if not entirely revolutionize, the
future of customer support by focusing more on strategic
solutions and higher-level problems.
When creating the correct help desk solution, it is
important to pay attention to the communication flow
between a business and IT.
Therefore, the IT help desk shifts from being a preemptive
strategy to an important strategic function of a business
that provides essential business value and have a dynamic
part in assuring a company’s success not only at present
but also beyond.(Mishra and Jain, 2016)
Therefore, the identified problem from the literature
review is that developing a helpdesk system, which is
organization oriented and which is cable of coping with
the changing and growing technological advancements.
III. METHODOLOGY AND
EXPERIMENTAL DESIGN
A. Data Gathering
Combination of qualitative and quantitative methodologies
were used as data gathering techniques when designing
the requirement specification. Questionnaires and
interviews were the main data gathering methods that
were used to gather the requirements and the information.
Apart from that, observations were also used. All the
required data were collected by using a selected group
of personnel within an exact period of time. Interviews
were conducted with both the lecturers and the students.
And the questionnaires were used to gather the data from
the students because, it is an efficient technique to gather
data from a large audience. Face-to-face interviews were
conducted with a selected group of lecturers who conduct
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
lectures for all the three intakes, 5-10 students from each
intake and with the Head of the Department. When
conducting the observations, the team participated in a
lecture and observed what really happen in a lecture and
what the problems that the both parties face are. Although
this method was a bit time consuming this was the most
accurate method for data gathering because it helped to
get the experience of the real situation.
B. Date Representation and Analysis
According to collected data, the team identified that most
of the students were used to search their problems in
Google and also they have stated that, since the answers
available in various websites are different from each
other, students find it difficult to get correct answers on
the internet and it was time consuming. Some students
were used to call lecturers or meet them in person to ask
questions. But, they also stated that, most of the time
lectures were not available to answer the calls or meet
students due to their busy schedules. Some students were
used to ask from friends but there were problems in the
accuracy of the answers which provided by friends. So the
current process was time consuming and less efficient. The
data gathered by using the above mentioned techniques
were analyzed by using tables and charts.
Figure 1. Data Representation
Through examining the data, the team were able to clearly
identify the existing process, problems and limitations
prevailing in the current process and the suggestions of
lecturers and students regarding the existing process. Some
of the identified limitations of the current process were
due to various reasons such as, all the students don’t get
the opportunity to present their problems to the lecturers,
inefficiency and time consumption. And through this
analysis it was proven that the students and the lecturers
were highly satisfied with the proposed system. Finally,
after this analysis the team came to the conclusion, that
10TH INTERNATIONAL RESEARCH CONFERENCE 1113
PROCEEDINGS
a fully automated system should be developed to re-
engineer the existing process to make it function better.
C. Approach
The main users of this system are the students and the
lecturers. In addition, a system administrator can also
access the system for registering users, maintenance,
troubleshooting and upgrading. There are five major
different types of inputs concerning students and lecturers.
Students can logon to the system by entering username
and password. Then they can post the question, the
subject area which the question belongs and the priority
of the question. Lecturers can also logon to the system
by entering username and password. The outputs of the
system are solutions for the problems submitted by the
students, ratings of the provided answers, feedbacks and
e-mail notifications.
D. Technology adopted
The most appropriate technology for developing the
system should be decided by considering the system
domain and the requirements. It is much significant to
select the technological methodologies which will be
capable to satisfy both the functional and non-functional
requirements of the system. Since the proposed Online
IT Helpdesk system is a web based system the team had
to pay more consideration when opting the technological
facet. The technological methodologies had to be
selected in such a manner that it will help the system to
be available at anytime, anywhere and also to make it
efficient and effective. With an in-depth analysis of the
system requirements, PHP has been used as the main
programming language to develop the backend of the
system. The database of the system has been implemented
using MySQL. The front end of the system has been
developed using the programming languages HTML and
CSS and also the bootstrap framework was used to make
flexible and user friendly interfaces. Finally, the validations
of the forms have been done using JavaScript.
E. Design
Design of the system is described under two topics as
overall system architecture and modular architecture.
1114 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
1)Overall System Architecture
This section will describe the three tier architecture of the
system. The system as a whole is mainly divided into three
layers as Client Layer, Application Layer and the Database
Figure 2. System Architecture
2)Client layer
This layer builds the interaction between the users and
the system. Mainly in a system the human computer
interaction is achieved through interfaces. Therefore, this
is the layer which holds the interfaces created according to
the user requirements. There is a variation of the visibility
of interfaces to the users. At the beginning all the users get
the access to the same login interface. But after the user
authentication the interfaces accessible for the users are
different. This has been done according to the role which
the users play in the system. The main function of this
system is question submission and providing solutions.
Therefore, the interfaces of the system has been designed
to accomplish this task easily without any hesitation.
3)Application layer
This layer act as the bridge between the client and the
database Layer. This layer will do the manipulation of
processes according to the given inputs to present the
expected output. Functionality, efficiency, accuracy and
the productivity of the system is mainly based on the logic
which is designed at the application layer.
4)Database layer
This is the layer which does the database management
of the system. And also the data storing tactics are
determined in this layer. In the new system there is only
one database to store all the data regarding the system.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

The database is created in a very consistent manner also 10)Answers Rating Module
by applying the concurrency control. In the database all
This module is only accessible by the students. When
the tables are interconnected with each other properly
lecturers post the solutions students get the opportunity to
to make the data manipulations effective and efficient.
rate those answers by giving stars.
Both the data entered by the users and the data which is
processed by the system are stored in the database. 11)Feedback Module

5)Modular Architecture
This section discusses about how the system is divided
into modules and how each module functions. The
modularized approach used in designing has ease the
development and the maintenance of the system.
6)Login and Authentication Module
Only the authenticated users get the access to the system.
After the authentication the accessibility to the system
varies according to the user type.
7)Question Submission Module
This module is only accessible by the students. Here the
students will submit their question, prioritize the question
as low, medium or high according to the time period
that they need the answer and also they will select the
subject area which the question belongs. Then the system
automatically forwards the question to the lecturers who
have expertise in the relevant subject area.
This module is accessible for both the lecturers and the
students. Both users can give feedbacks about the system
and the services rendered by the system.
12)Previous Question Viewing Module
Both the lecturers and students get the access to this
module. This module provides the facility of viewing
previous questions and the answers.
The developed Online IT Helpdesk system consists of
three main interface categorizations, namely interfaces
of administrators, interfaces of lecturers and interfaces of
students. At the beginning, all three types of users, login to
the system by using the same login interface and then the
interfaces are visible according to the user type.
Following figures illustrates some of the main interfaces of
the proposed solution.

8)Solutions Providing Module

This module is only accessible to the lecturers. When a

question is forwarded to the relevant lecturers then that
lecturers provide the solution for that question. If all
relevant lecturers refuse to answer a question within a Figure 3. Welcome Page
given period of time, then that question is forwarded to
all the lecturers who have been registered in the system.

9)Notification Sending Module

If a student prioritizes a question as high, then that relevant

lecturer is notified with an e-mail notification. And also
if a question is prioritized as low and if that question is
answered within a short period of time then students are
notified with an e-mail notification.
Figure 4. Posting a Question

GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1115
PROCEEDINGS
Figure 5. Answering a Question
IV. RESULTS AND DISCUSSION
An evaluation procedure has been carried out to check
whether the system functions properly and whether it
meets the expected requirement specifications. For this
evaluation purpose a stable system prototype has been
used. And it is of utmost importance that this prototype is
very much similar to the final product. In the evaluation
procedure questions were directly forwarded to the
relevant lecturers as expected in the requirements. As an
example when a user posted a software based question
by selecting the software tag that question was only
displayed to the lecturers who has software tag with an
email notification. And when the answer is provided by
the lecturer then a notification email was send to the
student who posted the specific question. The system was
tested using dummy data and the accuracy of the system
was 100%. All the mails and question forwarding was done
without any errors.
After implementing the system lecturers were able to
give comments regarding the answers provided by other
lecturers and students were able to give feedbacks about
the satisfaction of the given answers. And the lecturers
were satisfied with the feedbacks given by the students.
User interactivity with the system was 100% and the
system helped the students in many ways, like they could
ask the questions about their subject matters directly from
their department lectures and thereby, they were able to
get more reliable answers to clear their doubts. And also
each and every student got the opportunity to present their
problems to the lecturers. And they could look and search
the questions which was asked by other students. When
posting a question students’ name was displayed along
with the question and the lecturer name was displayed
when a question is answered and comments were posted
anonymously in order to avoid conflicts between lecturers.
1116 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Therefore, students were motivated by the system to ask
more questions. As a result of that students engaged in
more studies and increased their academic knowledge.
Finally, this system was accomplished in creating an online
academic discussion among the lecturers and the students.
When considering the nature of this system it is clear
that this system has a more dynamic nature. The system
will have to process a number of user requests at once.
Therefore, this system can be further developed by using
the emerging multi-agent technology. This technology
will be able to handle the dynamic nature of this system
in a very reliable manner. Therefore, this can be done as a
future enhancement for the system.
V. CONCLUSION
This online help desk system is a novel system which will
be implemented in KDU because there is no such system
currently available. This system is developed with the
intention of helping the students to clear their doubts
regarding the subject matters. The main motivation
fact that lead us to develop such a system is that, due to
the busy schedules of the lecturers and due to the large
number of students, the opportunity to solve the problems
regarding the subject matters is minimum. There are
mainly two types of users who are directly beneficial with
the system implementation. They are the lecturers and the
students. Because with the help of this system, students will
frequently get the opportunity to present their problems to
the lecturers from anywhere anytime as this is an online
system. And lecturers will get a clear view about the subject
matters that they should be more attentive. According
to the gathered data, it has become clear that both the
students and the lecturers are completely satisfied with the
system. Therefore, the entire project process concludes an
efficient and accurate development of an online help desk
system for KDU based on the collected data. This system
is primarily developed for the IT department of KDU. As
this system is scalable this can be further developed to
cover all the faculties of KDU.
REFERENCES
Allam, A.M.N., Haggag, M.H., 2012. The question
answering systems: A survey. International Journal of
Research and Reviews in Information Sciences (IJRRIS) 2.
Barskar, R., Ahmed, G.F., Barskar, N., 2012. An Approach
for Extracting Exact Answers to Question Answering
(QA) System for English Sentences. Procedia Engineering
30, 1187–1194. doi:10.1016/j.proeng.2012.01.979
 ISBN number: 978-955-0301-41-6
Brill, E., Dumais, S., Banko, M., 2002. An analysis of the
AskMSR question-answering system, in: Proceedings
of the ACL-02 Conference on Empirical Methods in
Natural Language Processing-Volume 10. Association for
Computational Linguistics, pp. 257–264.
Chen, Y., 2004. To help helpdesk: a field study of online help
systems in campus context. Campus-Wide Information
Systems 21, 81–94. doi:10.1108/10650740410529510
Dwivedi, S.K., Singh, V., 2013. Research and Reviews in
Question Answering System. Procedia Technology 10,
417–424. doi:10.1016/j.protcy.2013.12.378
Fukumoto, J., Aburai, N., Yamanishi, R., 2013. Interactive
Document Expansion for Answer Extraction of Question
Answering System. Procedia Computer Science 22, 991–
1000. doi:10.1016/j.procs.2013.09.184
Gyll, K., Gyll, P., 2003. Panthera.
Hafifi, I., Ariza, A., 2012. Help Desk System.
Hirschman, L., Gaizauskas, R., 2001. Natural language
question answering: the view from here. Natural Language
Engineering 7. doi:10.1017/S1351324901002807
Lebedeva, O., Zaitseva, L., 2014. Question Answering
Systems in Education and their Classifications, in: Joint
International Conference on Engineering Education &
International Conference on Information Technology. pp.
359–366.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Mishra, A., Jain, S.K., 2016. A survey on question
answering systems with classification. Journal of King
Saud University - Computer and Information Sciences 28,
345–361. doi:10.1016/j.jksuci.2014.10.007
Moré, J., Climent, S., Coll-Florit, M., 2012. An answering
system for questions asked by students in an e-learning
context. International Journal of Educational Technology
in Higher Education 9, 229–239.
Motoda, H., n.d. Help Desk System with Intelligent
Interface.
Ómarsson, G.I., 2010. Knowledge Management in an IT-
Help Desk environment.
Radev, D.R., Qi, H., Wu, H., Fan, W., 2002. Evaluating
web-based question answering systems. Ann Arbor 1001,
48109.
ACKNOWLEDGMENT
Authors would like to acknowledge all the lecturers of
the Faculty of Computing who assisted us by providing
information on the current process of communication
between lecturers and students.
10TH INTERNATIONAL RESEARCH CONFERENCE 1117
PROCEEDINGS ISBN number: 978-955-0301-41-6

AN INFRASTRUCTURE SERVICE
SUGGESTION SYSTEM
FOR DATA COMMUNICATION NETWORK :
LOCAL AREA NETWORK
HAHV Halwatura1#, RPS Kathriarachchi1
1
Faculty of Computing General Sir John Kotelawala Defence
University Rathmalana , Sri Lanka
# halwatura.vihanga@gmail.com

Abstract - Network Infrastructure development includes

building of networks such as LANs, WANs, Intranets and
Extranets. This is a responsibility of the Network Engineers
or the network architects. They should have the knowledge
as well as the skills to plan and design a data communication
network, be conscious about the best place to have
communication lines, be updated with the latest network
equipment and be aware about the hardware and wiring
needed for the buildings. These tasks are more complex
for larger organizations which needs more accuracy,
dedication and flexibility in their networks. Therefore the
solution proposed in this paper to reduce these challenges
of a network engineer is a Data Communication Network
Infrastructure Recommendation System. This system
will be able to get any floor plan and analyse it in order
to provide recommendations on the installation of the
main networking components such as selecting the server
room and deployment of the backbone cable. It also allows
the user to draft the plan with endpoints by the drag and
drop function, keep a log fie of the saved data and also to
send a System generated network layout of the plan to the
customer. The software is being developed by making use
of the Image processing technology with the help of the
MATlab Software and C#. A series of algorithms will be
used of achieve the objectives of the software. This software
is non-identical to current software in the market.
Keywords - Network Infrastructure, Image Processing
I. INTRODUCTION
Data communications is the transfer of computer
information from one place to another by using electrical
or optical transmission. Such systems are often called data
communications networks. A variety of performance
requirements and system constraints were considered in
the design of the network. . Designing a network consists
of tasks such as making decisions about the network
type that best fits the needs of a particular organization.
In larger sites this task is performed by a senior network
architect: an experienced network engineer familiar with
both network software and hardware.
According to a business perspective, the organizational
commitment is an important characteristic when
planning, implementing and controlling an effective
network strategy. The organization must be committed
to developing an infrastructure that facilitates
communication of the business objectives to the network
planning team. The organization must also develop
internal standards, methods, and procedures to promote
effective planning.
Often consultants and outside vendors are needed to help
plan and implement the network. However, sometimes
consultants are needed to help develop and specify the
objectives and requirements. Although outside consultants
offer benefits such as expertise and objectivity, they also

1118 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
present their own set of challenges. The design engineers
work according to their own experience.(Tutschku et al.,
1997)
These inefficiencies still remain and continue to exist
when a network is being designed because once the
requirements of the area is given, the network engineer
has to use his knowledge to figure out the position of
the cables and resources to be used. The ideal network
design tool can mean different things to different people.
But whether you’re a network designer, network manager
or engineer, sales or marketing manager, or member of
the Research and Development staff, you probably want
intuitive graphical interfaces that resemble computer-
aided design tools. (Bragg, 2000) A Data Communication
Network Infrastructure Recommendation Software is
to be developed to eliminate the main inefficiencies of
the Network engineer by giving them recommendation
to perform the network design. The process of design
involves the analysis of user requirements and their
translation into a technical solution that is ultimately a
compromise resulting from balancing technology against
budget. (Linge and Parsons, 2006) It could decrease the
complexity of the development of a network in a building
or LAN area by giving recommendations to the network
design. The importance of this project is not only to the
Network engineer as mentioned above but also to the
customer company who wants the network to be put up.
This project can benefit both these parties, the system
helps having a direct communication with the customer
as well in forwarding the plan to them just by one click.
The system can provide recommendations to set up the
network so the engineer has another helping hand to
confirm his/her decisions. Before he examine the place he
can input the plan and check else he can double check the
decisions he made against the system.
Rest of the paper is organized as follows. Section 2 provides
a literature review on existing systems of image processing
and automated architectural software’s and identify the
research problem/gap and the possible technology to
solve the problem. Section 3 is on the image processing
technology used in this project. Section 4 presents the
design and implementation and then Section 6 concludes
the research findings.
II. LITERATURE REVIEW
There are many related research papers published related
to the architectural plans. These research papers mainly
focus on the wall detection of the architectural plans
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
because it is a main component of a floor plan which can
define the whole structure of the building. (Macé et al.,
2010)Two different research papers which were focused
on analysing the floor plans used the Hough Transform
along with vectorization. These two methods were mainly
used to detect the lines and room of the floor plan. How
this algorithm was defines was according to a certain
assumptions which were taken about the walls such
as they are thick, they are double lines, longer hat the
rest of the line and also rectangular. (de las Heras et al.,
2013) According to the paper written by Sebastien Mace,
there are there are some drawbacks in using the Hough
Transformation such as it’s complex nature especially
when the images are large. Rooms can also be detected
by using semantic analysis. Semantic analysis judges
whether the syntax structure constructed in the source
program derives any meaning or not. From this process
the semantics can be separated in order to find the labels
of the detected rooms so we could get the room functions
through them. To detect the doors which are represented
in an arc shape in the floor plan can be identified using a
spotting technique to match the symbol of the door. The
Function of the Optical Character Recognitions (OCR) is
needed to identify the labels or text characters in the floor
plan. OCR can also be programmed in such a way that
it can split more than one label to identify the roles of a
floor plan. (Ahmed et al., 2012) In order to increase the
performance and gain accurate results an approach such
as the removal of the components outside the outer walls
are used so the functions can focus on analyzing the main
components of the floor plan image. (Ahmed et al., 2011)
Floor Plans could be analyzed and interpreted in terms of
Geometry. According to the Architectural discipline, the
building design stage can be categorized intro three main
process which are the conceptual design process, design
development, and construction design. Zimmermann,
2005) During the identification of the parts of a floor
plan, identifying a curve can be very challenging. A
preprocessing model based on dominant point detector
which is associated with the maximum curvature point of
an image could be used to segment the curve into straight
lines or shorter curves. The dominant point detection
process is based on the geometry theories. (Nguyen and
Debled-Rennesson, 2011) During very early studies there
were methods to segment the lines and arcs such as using
the straight line segments as previously mentioned and
then fitting those lines to ellipses. These two approaches
are being performed with the help of a series of algorithms
used to identify steeper curves. (Rosin and West, 1990) A
paper has described an algorithm for detecting circular
10TH INTERNATIONAL RESEARCH CONFERENCE 1119
PROCEEDINGS ISBN number: 978-955-0301-41-6

arcs in arbitrary space curves that result from edge
detection in images. A fast algorithm for determining the
minimum error match of a circular arc with the data is
used. The computational burden is reduced over other
curve fitting methods by the constraint that the curve
must be bounded by two defined end points.
Although Geometry is an easier method, it might not
be the most fast and accurate method. When compared
with systems which uses further advancements, according
to my opinion geometric is in the initial stages. Using an
algorithm to identify the layout of the plan is the best. Two
of the methods to input the image of the floor plan to be
analyzed can be defined as an input and output method.
In the input method, it can be done in real time while the
architect is drawing the plan in a device such as a tablet by
using structures already available in a drawing tool. If we
use an output method, the image could be input as a hand-
drawn plan, but if it is hand drawn, the drawing could be
unclear and confusing. A digitized image produced by the
output method requires high storage capacity, they are also
slow in editing. So using an input method is much easier,
reduces the time to process along with the storage space.
It can also minimize complexities that could arise with the
image type compatibility as the image drawn is already
given as the needed image format but the architect must
have the drawing skills and access to a required computing
device such as a tablet which contains drawing tools.
(Dosch et al., 2000) However, there are ways which could
minimize the errors that could occur in digitized images
and they are classified into three sections namely; First
being the recognition of lines and shapes which falls under
the geometric area, second is identifying the building
elements which are walls, doors, etc and the last is about
the spatial articulation which is concerned on the space
and distribution the objects in the spaces.(Koutamaniss
and Mitossi, 1992). Depending on the above analyzed
facts, it is most suitable to use the output method as it is a
flexible approach to an architect.
scanned and vectorized, during this process we can get a
graph structure with attributes which can later be used
with a graph matching process to recognize the image.
Many vector files containing only ASCII-format data can
be modified with simple text editing tools. Individual
elements may be added, removed, or changed without
affecting other objects in the image. It is also easier to
render and save the vector data. But still vector files
cannot easily be used to store extremely complex images,
such as some photographs, where colour information is
paramount and may vary on a pixel-by-pixel basis.
Parts of the image such as the walls of the floor plan can
be recognized by a filling algorithm. Where the darker,
thick areas can be easily identified. This will be used along
with a Hough Transformation this identifies the areas
without a fixed pattern. (Lladós et al., 1997) These were
relatively easier methods with respect to the traditional or
manual methods but errors and challenges still remain.
Even if these methods were used at some point a human
intervention needed to recheck and cover the unidentified
errors. (Ryall et al., 1993a) The Hough technique is
particularly useful for computing a global description of
a feature(s), given (possibly noisy) local measurements..
A sketch based system was proposed to query a floor plan
repository along with pre-processing methods such
Table 1 Literature review summary

Although the above paragraph focuses on the digitized

images, this doesn’t mean that hand drawn images are
not possible to interpret. There have been a number of
previous studies on the interpretation of hand drawn
floor plans. Hand drawn images tend to have more
noise that the digitized images, these are categorized as
pixel, vector and context noise. To address these noise
in the images we can use Incremental Arc Segmentation
Algorithm (Wenyin et al., 2001) Due to the possibility of as differentiation between thick, medium, and thin lines
unclear drawings being input, it is recommended to use and the removal of components outside the convex hull
various algorithms. During this process the image can be of the outer walls. According to this hypothesis, a line of
1120 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
 ISBN number: 978-955-0301-41-6
polygon is selected as wall edge if it is short and is either
convex or concave. (Ahmed et al., 2014) Another system
is proposed that offers recognition of scanned floor plans
(sketched by hand without use of rulers or other tools),
and automatic conversion of walls, stairs, doors and other
architectural elements. A recognition algorithm was the
main solution taken for this problem. Floor plans are
read through a scanner while architectural elements are
recognized and interpreted automatically, which allows
large-scale reduction of input cost. (Shio and Aoki, 2000)
According to the successfully completed review of the
network infrastructure designing problems, solutions and
the technologies, the system which is proposed by this
thesis will be developed using Matlab, C# and MySql to
address the common problems such as the inefficiencies
of the system. Table 1 shows the summarized results. This
was decided based on the results of previous research
studies.
III. IMAGE PROCESSING
TECHNOLOGY
Image processing involves changing the nature of an
image in order to either improve its pictorial information
or human interpretation or render it more suitable for
autonomous machine perception. We shall be concerned
with digital image processing, which involves using a
computer to change the nature of a digital image. It is
necessary to realize that these two aspects represent two
separate but equally important aspects of image processing
(McAndrew, 2001).
A. MATLAB Software
One of the most advanced and sophisticated tools that is
used to analyse and edit images today is MATLAB, it is
user friendly with graphical interfaces and anyone without
prior knowledge about it can learn about it easily. By using
various functions, an image can be edited and analysed by
using MATLAB. When it comes to floor plan detection,
we can use colour detection, noise removal, converting
to grayscale, edge detection and area filling functions.
(Buksh et al., 2014) A software was developed by using
the main functions of the MATLAB. It was an image
editing software where the user can insert the image and
perform a series of functions accordingly. A previously
proposed image editor contains a variety of image editing
functions used by MATLAB with the intention of bring
all of them under a single platform so any level of user
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
will understand it. In a modelling system, a user friendly
GUI is developed and two alternative methods for image
acquisition are executed. Many experimental examples are
used in evaluating the performance of the adopted edge
detectors.
The proposed system that this research paper will focus
on developing the image processing components by using
MATLAB based on the features identified by the already
existing software.
B.Secondary Software
Apart from the main language MATLAB which is used
for the image processing function, two other technical
software will be used to successfully complete the
development of the proposed system. They are;
• C# - for functions which does not include image
processing functions.
• MySql - For the implementation of the database.
IV. METHODOLOGY
This section will show how the above technologies are
utilized for this ongoing project. Matlab will be used to
do all the image processing functions. The image in this
scenario is the Floor plan of the building and it will be
enhanced, restored and morphological image processing
will be done to make the image useful to the network
engineer. The Mysql database is used to store mainly the
login information and log details of the software so they
can let authorized engineers to the customer plans and
view log of the saved plans to preserve there integrity. C#
is used to view, retrieve and login to the software, basically
it is used to operate all other processes that are not related
to image processing. Figure 1 shows a diagram of the
methodology
Figure 1 Methodology illustration
10TH INTERNATIONAL RESEARCH CONFERENCE 1121
PROCEEDINGS ISBN number: 978-955-0301-41-6

IV. DESIGN AND System should be able to;

IMPLEMNTATION •Auto-draw the backbone cable.

Deta Communication
A. Description of the proposed system Network Infrastructure
Recommendation System

Data Communication Infrastructure Recommendation LOGIN

System is to be developed to minimize the difficulties
of a Network engineer during the design of a network. IMPORT PLAN

Considering the problem, it is important to develop a

system to facilitate the services needed by a Network INPUT BANDWIDTH /
END POINT

engineer in designing a Network in a better manner and

overcoming those identified problems in the current
SELECT SEVER
ROOM

manual system. System would be a Data Communication SYSTEM GENERATED

Infrastructure Recommendation System. The system NETWORK LAYOUT

would be built with MAT lab and the main user of the DRAG / DROP
END POINT
Customer
system would be the Network engineer who is responsible Network
for Network design but it can also be used for training of Engineer CALCULATE CABLE
DISTANCE

the network engineers. This system will be designed to

cover some of the main functions such as the ability for the
VEW / PRINT
REPORT

network engineer to import the plan, selecting the server

room where the backbone cable will be auto drawn. An
input form could be used input the bandwidth, topology,
etc. Furthermore, when the engineer wants to know the
number of routers or switches needed for a particular area,
he/she will be able to input the number of endpoints in one
floor so the system will generate the number of switches
needed accordingly.
B. Features of the proposed system
Aim of the project is to develop a Data Communication
Network Infrastructure Recommendation System to
eliminate the above mentioned problems which can make
the job of a Network engineer more efficient, fast and
easier.
Network Engineer should be able to;
•Import the plan of the building.
•Input the bandwidth, topology, Number of endpoints
etc.
•Manually select server room in the plan.
•Send a system generated network layout of a particular
area to the customer via mail
•Drag and drop the end points
•Calculate cable distances
•View/Print a report
1122 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
Figure 2 Usecase diagram of the proposed system
After the engineer enter the username and the password,
they will be able to successfully access the features of
the system. Next the main input being the floor plan of
the building will be input by the Network engineers
to perform the floor analysis. An input form will ask to
fill the bandwidth and the number of end points. The
backbone cable will also be auto-drawn according to an
input request given by the main user. Via a mouse click
command the user is able to select the server room from
the floor plan. The outputs of the system is very simple.
After the process is being done the engineer can print or
view the finished work. If the user wants to clarify the
diagram of the structure with a customer feedback, he/she
can send a system generated network layout of the finished
work with just one click.
According to the successfully completed review of the
network infrastructure designing problems, solutions and
the technologies, the system which is proposed by this
thesis will be developed using Matlab, C# and MySql to
address the common problems such as the inefficiencies
of the system This was decided based on the results of
previous research studies. Figure 2 shows the list of main
functions to be performed by the proposed system and
the actors or the users of the system along with their
relationship with the actions.
 ISBN number: 978-955-0301-41-6
C. Presentation of the application Interfaces
a) Welcome Screen and Login screen
The welcome page for the software which gives an
attractiveness works along with Login screen which will
be loaded after the progress bar is loaded. Figure 3 shows
the developed welcome screen. Figure 4 shows the login
screen.
Figure 3 Welcome Screen
Figure 4 Login screen
b) Image processing form
This form will contain the import button and other buttons
related to the process of the plan of the building. This acts
as the main interface of the system.
Figure 5 shows the image processing form with a plan
already loaded. The original image of the plan before being
loaded is processed into a grayscale image then removed
noise by using sail and pepper technique and then the
image is turned into binary so that it is easier to identify
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
the walls from the black colour and the room from the
white colour.
Figure 5 Image Processing screen
c) Log form
This form captures all details about saved data which
contains the time, date, who changed it as well as what was
changed.
d) Input details form
This is a form which will take the inputs of no. of end
points, bandwidth, topology, etc for further processing of
the image.
e) Send system generated network layout form
Once the network engineer press screen shott button
this form will appear where he/she will have to enter the
customer name so he email will be taken directly from the
database.
V. CONCLUSION
This paper presents a research to address the main
inefficiencies of a Network Engineer when planning,
designing and implementing a network infrastructure
for an organization because depending on the complexity
of the organization structure, a design can be more
advanced. This paper proposed a Data communication
network infrastructure recommendation system to give
best possible options a network engineer could consider
the design. The system is being developed by using
MATlab, MySql and C#. The design and implementation
is also presented in this paper.
10TH INTERNATIONAL RESEARCH CONFERENCE 1123
PROCEEDINGS
REFERENCES
Ahmed, S., Liwicki, M., Weber, M., Dengel, A., 2012.
Automatic Room Detection and Room Labeling from
Architectural Floor Plans. IEEE, pp. 339–343. doi:10.1109/
DAS.2012.22
Ahmed, S., Liwicki, M., Weber, M., Dengel, A., 2011.
Improved Automatic Analysis of Architectural Floor
Plans. IEEE, pp. 864–869. doi:10.1109/ICDAR.2011.177
Ahmed, S., Weber, M., Liwicki, M., Langenhan, C., Dengel,
A., Petzold, F., 2014. Automatic analysis and sketch-based
retrieval of architectural floor plans. Pattern Recognit.
Lett. 35, 91–100. doi:10.1016/j.patrec.2013.04.005
Bragg, A.W., 2000. Which network design tool is right for
you? IT Prof. 2, 23–32.
Buksh, R., Routh, S., Mitra, P., Banik, S., Mallik, A., Gupta,
S.D., 2014. MATLAB based image editing and color
detection. Int. J. Sci. Res. Publ. 4, 1–6.
de las Heras, L.-P., Fernández, D., Valveny, E., Lladós,
J., Sánchez, G., 2013. Unsupervised wall detector
in architectural floor plans, in: Document Analysis
and Recognition (ICDAR), 2013 12th International
Conference on. IEEE, pp. 1245–1249.
Dosch, P., Tombre, K., Ah-soon, C., Masini, G., 2000. A
complete system for the analysis of architectural drawings.
Int. J. Doc. Anal. Recognit. Manuscr. 3, 102–116.
Koutamaniss, A., Mitossi, V., 1992. Automated recognition
of architectural drawings. Presented at the International
Conference on Pattern Recognition, pp. 660–663.
Kumari, P., Gupta, S.K., 2015. Image Enhancement GUI
using MATLAB. Int. J. Innov. Res. Comput. Commun.
Eng. 3.
Linge, N., Parsons, D., 2006. Problem-Based Learning as
an Effective Tool for Teaching Computer Network Design.
IEEE Trans. Educ. 49, 5–10. doi:10.1109/TE.2005.852600
Lladós, J., López-Krahe, J., Martí, E., 1997. A system
to understand hand-drawn floor plans using subgraph
isomorphism and Hough transform. Mach. Vis. Appl. 10,
150–158.
1124 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
Macé, S., Locteau, H., Valveny, E., Tabbone, S., 2010. A
system to detect rooms in architectural floor plan images,
in: Proceedings of the 9th IAPR International Workshop
on Document Analysis Systems. ACM, pp. 167–174.
M. Kudˇelka Jr., 2012. Image Quality Assessment, in:
Šafránková, J. (Ed.), . Matfyzpress, Praha.
McAndrew, A., 2001. An-Introduction-to-Digital-Image-
Processing-with-Matlab-Notes-for-SCM2511-Image-
Processing.pdf.
Nguyen, T.P., Debled-Rennesson, I., 2011. Decomposition
of a curve into arcs and line segments based on dominant
point detection, in: Scandinavian Conference on Image
Analysis. Springer, pp. 794–805.
Rajesh Parihar, V., 2015. IMAGE MODIFICATION
DEVELOPMENT AND IMPLEMENTATION: A
SOFTWARE MODELING USING MATLAB. Int. J.
Comput. Sci. Mob. Comput. 4, pg.72 – 79.
Rosin, P.L., West, G.A.W., 1990. Segmenting Curves into
Elliptic Arcs and Straight Lines. Presented at the IEEE, pp.
75–78.
Ryall, K., Marks, J., Mazer, M., Shieber, S.M., 1993a. Semi-
Automatic Delineation of Regions in Floor Plans.
Ryall, K., Marks, J., Mazer, M., Shieber, S.M., 1993b.
Annotating floor plans using deformable polygons.
Shio, A., Aoki, Y., 2000. Sketch plan: a prototype system
for interpreting hand-sketched floor plans. Syst. Comput.
Jpn. 31, 10–18.
Shrivakshan, G.T., Chandrasekar, C., others, 2012. A
comparison of various edge detection techniques used
in image processing. IJCSI Int. J. Comput. Sci. Issues 9,
272–276.
Solomon, C., Breckon, T., 2011. Fundamentals of digital
image processing: a practical approach with examples
in Matlab. Wiley-Blackwell, Chichester, West Sussex ;
Hoboken, NJ.
Tutschku, K., Leibnitz, K., Tran-Gia, P., 1997. ICEPT-An
integrated cellular network planning tool, in: Vehicular
Technology Conference, 1997, IEEE 47th. IEEE, pp. 765–
769.
 ISBN number: 978-955-0301-41-6
Wang, Z., Sheikh, H.R., Bovik, A.C., 2004. NO-
REFERENCE PERCEPTUAL QUALITY ASSESSMENT
OF JPEG COMPRESSED IMAGES.
Wenyin, L., Zhai, J., Dori, D., Long, T., 2001. A system for
performance evaluation of arc segmentation algorithms,
in: Proc. CVPR Workshop Empirical Evaluation in
Computer Vision.
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Zimmermann, G., 2005. FROM FLOOR PLAN
DRAFTING TO BUILDING SIMULATION-AN
EFFICIENT SOFTWARE SUPPORTED PROCESS, in:
International IBPSA Conference Building Simulation. pp.
1441–1448.
ACKNOWLEDGMENT
I would like to express my gratitude to my supervisor Mr.
RPS Kathriarachchi for the knowledge, support and words
of encouragement he has given me at all times during this
pro
10TH INTERNATIONAL RESEARCH CONFERENCE 1125
PROCEEDINGS
A COMPARISON OF
DELTA-COMMUNICATION
TECHNOLOGIES AND TECHNIQUES
NR Dissanayake1# and GKA Dias1
1
School of Computing, University of Colombo, Sri Lanka
# nalakadmnr@gmail.com
Abstract - Delta-Communication can be seen as the
power of Rich Internet Applications, and there are
different Techniques and Technologies available for the
development of Delta-Communication, which should
be selected carefully into the Rich Internet Application
development. Enough discussions are not available,
which compare and contrast these Delta-Communication
development Techniques and Technologies towards
supporting decision making of selecting them. This paper
provides an overview of the contemporary Techniques
and Technologies available for the Delta-Communication
development, contextually compares them aligning to
some selected criteria, and finally discusses some facts to
be considered when selecting them for the Rich Internet
Application development. A literature survey on the
Delta-Communication development Technologies and
Techniques was conducted, which was followed by a series
of experiments towards getting the empirical evidence for
the comparison. During the contextual comparison, the
Simple Pull Delta-Communication was identified as the
least complex technique and the Websocket was noted as
the highest complex technology.
Keywords - Delta-Communication, Rich Internet
Applications, Techniques and Technologies
I. INTRODUCTION
Rich Internet Applications (RIAs) have become popular
with their increased user experience delivered through
rich GUIs and faster responses. The key concept behind
1126 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
the fast responses is the Delta-Communication (DC),
which enables the communication of necessary data
between the client and the server components, in either
synchronous or asynchronous modes, supporting both
data-pull and push modes (Dissanayake & Dias, 2017).
There are various Technologies and Techniques (TTs)
available for developing the DC in RIAs. These TTs have
been already discussed in different forums, and also many
tutorials are available to demonstrate the development of
them. However, proper comparisons of these TTs are not
available towards supporting decision making in selecting
these TTs for the RIA development.
This paper gathers contemporarily used DC development
TTs, provides an overview of their features, then
compare and contrast them aligning to the facts:
communication mode, complexity, ease of development,
ease of maintenance and modifiability, and scalability. The
paper also discusses some facts to be considered, when
making decisions for selecting the DC TTs for the RIA
development.
For this study, a literature survey was conducted focusing
on identifying the available DC development TTs,
their features, and development details. Parallel to the
literature survey, a series of experiments was conducted
towards getting empirical evidence on the identified DC
development TTs. During the experiments, prototypes
were developed using HTML, JavaScript, jQuery, PHP,
and MySQL; and development process and the prototypes
 ISBN number: 978-955-0301-41-6
were examined focusing on the complexity, ease of
development/maintenance/modifiability, and scalability
factors.
Section II of this paper provides the background of the
DC, stating some history and the definitions for the DC
and related basics. Section III provides an overview of
the contemporarily utilized DC TTs. Section IV delivers
the contextualized comparison of the DC TTs discussing
the results of the comparison. Based on the knowledge
gained through the comparison, section V discusses some
facts to be considered when selecting the DC TTs for
the RIA development. Finally, section VI concludes the
paper specifying the future applications of the knowledge
delivered in this paper.
II. BACKGROUND
This section provides some history of the DC concept,
and also states definitions for the DC and related basics
towards getting an adequate understanding of the concept
of the DC before the comparison of the development TTs.
A. History of Delta-Communication
Microsoft was working on a technology named XMLHTTP
in their Exchange 2000 project (Hopmann, n.d.), and it
was first introduced to the world as an ActiveX control in
Internet Explorer 5.0 in March 1999 (Dutta, 2006) (Smith,
2006); and later it was called the XMLHttpRequest (XHR)
object, which has an Application Programme Interface
(API) in JavaScript (JS).
In 2005, Jesse James Garrett from Adaptive Path coined
the name AJAX, introducing the first JS based DC
technique for the web applications, utilizing the XHR
object (Garrett, 2005). This technique became popular
and took the traditional web applications to a whole new
era called Web2, which is the era of the RIAs. Later W3C
acquired the control of the XHR object and released the
first specification on 2006 (W3C, 2006). Since then the
term AJAX has become another name for the RIAs, where
even some developers refer the RIAs as AJAX applications.
AJAX can be seen as the beginning of the JS based
RIA development approach, and it became a major
breakthrough in the web development area (Salva &
Laurencot, 2009). After its introduction, developers
were learning how to use AJAX to create desktop-like
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
GUIs in the web applications such as Google Maps; and
later they subsequently used AJAX even to create entire
enterprise RIAs (Lawton, 2008). Using the JS’s ability
to manipulate the Document Object Model (DOM) in
HTML documents, AJAX achieves and enhances the
interoperability capability of the web applications (Salva
& Laurencot, 2009).
It should be noted that the AJAX itself is not a technology,
it is a technique; and the technology behind AJAX is
the XHR object with its JS API. The AJAX is a data-pull
technique, employing the traditional request-response
model. Combining HTML and CSS with JS, AJAX has
become a powerful tool in RIA development, providing
the fundamental implementation of the DC.
The main characteristics of AJAX and other DC TTs
introduced after AJAX, are discussed in section III.
B. Definitions for Delta-Communication and Related
Basic Concepts
Before discussing the DC TTs, it is important to
understand what the DC is, its characteristics, and also
related basic concepts.
The main features of the DC are the capability of processing
in the background and then performing partial page
rendering to display the results on the GUI (Dissanayake
& Dias, 2017). The communication is done faster due to
the smaller set of data communicated compared to the
traditional web communications. Considering these facts,
the DC is defined as: “Delta-Communication is the rich
communication model used by the rich features of the
RIAs, for client-component(s) to communicate with the
server-component(s), to exchange only the needful dataset
– for a particular feature executed at the time – which is
smaller, compared to the size of the request/response of
traditional communication. Since the size of the dataset
communicated is smaller, the communication completes
faster, eliminating the work-wait pattern. The processing
of the response is done by the client-components in the
background, therefore the page refreshes are eliminated
and replaced by partial page rendering to update the
content of the GUI with the results of the response. The
user experience can be determined by the implementation
of the feature, in either blocking (synchronous) or non-
blocking (asynchronous) modes” (Dissanayake & Dias,
2017).
10TH INTERNATIONAL RESEARCH CONFERENCE 1127
PROCEEDINGS
Simple Pull Delta-Communication (SPDC) is the simplest
abstract implementation of the DC, and it is defined as:
“Simple Pull Delta-Communication is the basic abstract
Delta-Communication technique, based on the data-
pull mode. It describes the simplest form of data-pull
Delta-Communication, based on the request-response
model; and this technique is technology independent”
(Dissanayake & Dias, 2017).
Aligning to the definition of the SPDC, AJAX technique
can be seen as JS implementation of the SPDC, which is
limited to the browser based applications. The term AJAX
expresses some flaws, and it is outdated with regards to
the latest API version of the XHR object (Dissanayake &
Dias, 2017). Due to the evolution of the XHR object, the
technical scope of the AJAX has been expanded in terms
of both XML and Asynchronous aspects. Based on these
facts, considering the outdated and limited impression of
the term “AJAX”, it can be replaced by the term “JavaScript-
based Simple Pull Delta-Communication” (JS-SPDC)
(Dissanayake & Dias, 2017). The term JS-SPCS indicates
that it utilizes the SPDC technique, and developed using
JS.
III. TECHNOLOGIES AND
TECHNIQUES FOR THE DELTA-
COMMUNICATION
After the introduction of AJAX/XHR, the concept of the
DC had been used in some other TTs, and each TT is
associated with a set of pros and cons. These DC TTs can
be mainly classified under data-pull and data-push modes,
where the data-pull is based on the request-response
model, which the client requests and pull the data from the
server; and in the data-push, the server sends data to the
client without a request. This section provides an overview
of the available DC TTs, indicating their main features.
In-depth discussions of the specifications of these TTs are
intentionally kept out of the scope of this paper. Instead,
the focus is to classify them for a better understanding of
their usage, to be utilized in the comparison given in the
next section.
1128 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
A. AJAX/JS-SPDC
As discussed before, the AJAX can be seen as the simplest
implementation of the DC. It is based on SPDC, works
in data-pull mode, and the complexity is comparatively
lower than the other DC TTs. The main limitation of
the SPDC technique is that it does not support data-
push mode; therefore, it is not suitable for real-time data
communication as in publisher-subscriber model or any
other data-push models.
Several techniques had been introduced to simulate
data-push using SPDC; some of them – like polling and
long-polling – use the same XHR object, thus also called
reverse-AJAX.
B. Polling (Carbou, 2011)
In polling, data-push is simulated by sending automatic
XHR requests to the server periodically, receiving the
response for the frequent requests, and updating the GUI,
without the users’ explicit requests. There is an overhead
of developing periodical requests sending and responses
handling. If the frequency of the requesting is low, then
the server updates will not be received by the client in
real time. To get the updates in real time the frequency
of the automatic request sending needs to be higher, but
then the network overhead will also be higher. In the cases
of where the requests are returned without updates, the
resources for processing them are wasted. Addressing
these weaknesses of polling, the Comet techniques had
been introduced.
C. Comet (Carbou, 2011)
Comet is an umbrella term, which covers Streaming and
Long Polling. In Comet techniques, unlike polling, the
request is held by the server till there are updates to be
sent back to the client. In the case of timeouts, the request
is terminated, and the client can send a fresh request. This
technique reduces the frequency of the requests, thus, also
reduces the overhead on the network compared to polling.
1)Streaming: Under streaming, there are two
implementation techniques: Hidden iFrame, and Multi-
part XHR. In former technique, JS scripts are pushed to
the client, and in later technique, a multi-part response is
written via the same connection, which the request was
sent.
 ISBN number: 978-955-0301-41-6 PROCEEDINGS

2) Long Polling: Long polling uses the pure SPDC mode) use the SPDC as their core technique, where SSE
technique as in AJAX and polling, which holds the request and WS use their own protocols.
for longer time. Compared to streaming, this technique
can be seen as evolved and effective. The complexity is determined by the comparative
complexity of the DC implementation, considering the
D. Server Sent Events (Hickson, 2015) number of components needed for the development of
the communication components. SPDC is the simplest
On 2015, a true data-push protocol named Server Sent form of DC, hence can be considered as the least complex
Events (SSE) was introduced, which is unidirectional, from technique. As the extended versions of the SPDC, the TTs
the server to the client. However, it did not become much such as Polling and Long-polling can be considered as
popular. Compared to data-push simulation techniques, much complex. Though the SSE is a data-push technology,
the complexity is lower, and the development and since it is unidirectional, the complexity can be considered
modifiability are easier since the development overhead as similar to SPDC.
for data-push simulation as in polling or long polling is
not needed. Table 1. Comparison of some features of
DC TTs
E. WebSocket (Fette, 2011) JS-SP- Polling Streaming Long SSE WS
DC Polling
Core SPDC SPDC SPDC / hid- SPDC SSE WS pro-
An advanced bi-directional DC protocol named den iFrame proto-
col
tocol

WebSocket (WS) was introduced in 2011, which supports Mode Pull Push simu- Push simu- Push sim- Push Pull and
lation lation ulation Push
both data-pull and data-push modes. WS helps to reduce Complexity -- + + + - ++
the number of request-response pairs in the network Easiness of De- ++ - - - + --
velopment
compared to polling, and the header size of the WS is Easiness of ++ - - - + --
smaller than HTTP, which leads to increase the scalability Maintenance /
Modifiability
by addressing the C10K problem (Kegel, 2014) compared Scalability + -- - + + ++
to the other DC TTs. The WS gained the attraction of the
web engineers, however, the complexity of WS applications However, the SSE does not need additional components for
is higher. simulating like in polling or comet, thus can be seen as less
complex compared to them. As a bidirectional technology,
the complexity of the WS can be seen as the highest.
IV. COMPARISON OF SOME Additionally, WS based development requires additional
PROPERTIES OF DC TTS code and components for WS server implementation and
event handling, which makes the development is much
Table 1 contains the analysis of the contextualized complex.
comparison of some selected properties of the DC
development TTs. The meaning of the symbols used to The complexity directly affects the easiness of developing
denote the values are as follows. Note that these values are the DC, and also the maintenance and modifiability
comparative to each other, within the context. properties of the system. As the simplest DC technique,
SPDC provides the easiest development experience,
compared to the other TTs, where WS offers the least
• (++) – Higher easiness as the much complex bi-directional DC
• (+) – High technology. Note that this is the comparative easiness of
• (-) – Less developing a single DC feature, not the actual easiness of
the development of the entire system. The comparative
• (--) – Lesser easiness of the maintenance and modifiability properties
could be a similar to the easiness of the initial development.
The factor “Core” indicates the abstract technique or
the technology used in the core of the technique or the
For the scalability property, the scalability of the system
technology. The mode explains whether its data-pull,
towards real-time communication based on publisher-
push, or simulating either push or pull. The data-push
subscriber or similar model is considered. For such
simulating TTs (which are actually working on data-pull
scenarios, WS offers the capacity to provide the best
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY 10TH INTERNATIONAL RESEARCH CONFERENCE 1129
PROCEEDINGS
scalable solution, addressing the C10k issue (Kegel,
2014). JS-SPDC has no mechanism for higher scalable
development; however, note that if real-time updates are
not needed and if the frequency of the network utilization
is less, JS-SPDC can also provide higher scalability. With
the frequent redundant requests and responses, polling
shows the lowest scalability. Long polling and SSE provides
moderate scalability, and considering the higher size of
the data communicated the scalability of streaming can be
considered lower that long polling and SSE.
V. FACTS TO CONSIDER
IN DECISION MAKING OF
SELECTING TTS FOR RIAS
DEVELOPMENT
Several important facts to deliberate when making
decisions about selecting TTs for the implementation of
DC are discussed below (Domenig, n.d.).
Stakeholders’ constraints: Stakeholders of the system may
have technological constraints such as operating systems,
servers, database management systems, etc., based on
their available infrastructures and resources. The selection
of the DC development TTs may need to align to such
constraints since most DC developments are based on
frameworks/libraries/plug-ins. For example, if there is
a requirement for WS and the stakeholder needs a JAVA
based system, a suitable JAVA based framework for WS
development should be considered.
Compatibility: The compatibility of the TTs selected for
all the client-components, server-components, and DC
components is really important. Selection of a TT for
one component may affect the TTs of other components,
therefore may introduce additional learning curves, and
also may introduce limitations for selection of TTs for
other components. For example, if WS is selected for
DC, for both server and client components, additional
frameworks/libraries might be needed, and they should be
compatible with each other.
Scalability: The size of the target user population and
number of concurrent users (approximate) are the facts
to consider when determining the scalability. Scalability
plays a significant role when selecting TTs for client-
components and DC. If a higher scalability is needed, it
1130 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
ISBN number: 978-955-0301-41-6
is recommended to select WS, regardless its complexity.
Real-time updates: If real-time updates are needed,
it is wise to select the DC TT by also considering the
scalability. However, to get the true power of the real-time
data communication, WS can be seen as the best option.
Furthermore, suitable data formats like JSON should be
considered to get the maximum support for real-time
critical systems (Dissanayake, et al., 2015).
Development, maintenance, and modifiability: The
easiness of not only in the initial development, but also in
maintenance and future modifications may also take into
account. Table 1 contains the easiness in development,
maintenance, and modifiability. Correct identification of
the scalability requirements is essential, before considering
the maintenance and modification aspects.
VI. CONCLUSION AND
FUTURE WORK
The paper has given an overview of the contemporary DC
development TTs, and then has done a contextualized
comparison aligning to the facts: communication
mode, complexity, easiness of development, easiness
of maintenance and modifiability, and scalability. It
was identified that the AJAX/JS-SPDC is the simplest
implementation of the DC, which incorporates the lowest
complexity; however, it lacks in scalability. JS-SPDC is
limited to data-pull, thus it does not support data-push.
The WS covers the limitations of the other DC TTs like
scalability, support for both data pull and push. However,
the complexity of WS is seen as the highest among DC
development TTs.
Aligning to this comparison, the paper has discussed some
facts to be considered in the decision making of selecting
proper DC development TTs into RIA development:
stakeholders’ constraints, compatibility, scalability, need
for real-time updates, and the easiness in development/
maintenance/modifiability.
In future, we expect to exploit the knowledge delivered
in this paper to introduce a taxonomy for the DC
development TTs for RIAs. Then we hope to extend that
taxonomy to introduce a complete set of taxonomies
for RIA development TTs aligning to the architectural
elements: Components and Connectors.
 ISBN number: 978-955-0301-41-6
REFERENCES
Carbou, M., 2011. Reverse Ajax, Part 1: Introduction to
Comet, s.l.: IBM.
Dissanayake, N. R., De Silva, D. & Dias, K., 2015. A
Comparison of the Efficiency of Using HTML over XML
and JSON for the Asynchronous Communication in Rich
Internet Applications. Rathmalana, Sri Lanka, s.n.
Dissanayake, N. R. & Dias, K., 2017. Delta Communication:
The Power of the Rich Internet Applications [Accepted].
International Journal of Future Computer and
Communication (IJFCC).
Domenig, M., n.d. Rich Internet Applications and AJAX -
Selecting the best product. [Online]
Available at: http://www.javalobby.org/articles/ajax-ria-
overview/[Accessed 02 June 2014].
Dutta, S., 2006. Native XMLHTTPRequest object.
[Online]Available at: http://blogs.msdn.com/b/ie/
archive/2006/01/23/516393.aspx[Accessed 02 11 2015].
Fette, I., 2011. The WebSocket Protocol, s.l.: Internet
Engineering Task Force.
Garrett, J. J., 2005. Ajax: A New Approach to Web
Applications. [Online] Available at: http://www.
adaptivepath.com/ideas/ajax-new-approach-web-
applications
GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY
PROCEEDINGS
Hickson, I., 2015. Server-Sent Events. [Online] Available
at: http://www.w3.org/TR/eventsource/[Accessed 15 May
2015].
Hopmann, A., n.d. The story of XMLHTTP. [Online]
Available at: http://www.alexhopmann.com/xmlhttp.htm
[Accessed 02 11 2015].
Kegel, D., 2014. The C10K problem. [Online]
Available at: http://www.kegel.com/c10k.html[Accessed
20 04 2015].
Lawton, G., 2008. New Ways to Build Rich Internet
Applications. Computer, August, 41(8), pp. 10 - 12.
Salva, S. & Laurencot, P., 2009. Automatic Ajax application
testing. Venice, s.n., pp. 229 - 234.
Smith, K., 2006. Simplifying Ajax-Style Web Development.
Computer, May, pp. 98-101.
W3C, 2006. The XMLHttpRequest Object. [Online]
Av a i l a b l e a t : h t t p : / / w w w. w 3 o r g / T R / 2 0 0 6 / W D -
XMLHttpRequest-20060405[Accessed 10 05 2015].
10TH INTERNATIONAL RESEARCH CONFERENCE 1131
PROCEEDINGS ISBN number: 978-955-0301-41-6

LIST OF REVIEWERS

Snr Prof Asoka S Karunanda Maj RMM Pradeep

Dr LP Kalansooriya Mrs N Wedasinghe

Dr Sagara Sumathipala Mr TMKK Jinasena

Dr Chamara Liyanage Mr B Hettige

Dr Manjula Maduwanthi Mr PPNV Kumara

Col SPP Pakshaweera Mr ADAI Gunesekera

Mr Indika Baddegama

1132 10TH INTERNATIONAL RESEARCH CONFERENCE GENERAL SIR JOHN KOTELAWALA DEFENCE UNIVERSITY