You are on page 1of 16

STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY

COMMUNICATION AND INFORMATICS


Dita Kusumasari and Onny Rafizan

STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT


COMMUNICATION AND INFORMATICS POLICY

Study on Implementation of Big Data System for Supporting


Communication and Informatics Policy

Dita Kusumasari 1 and Onny Rafizan 2


1,2) Aptika and IKP Research and Development Center, Ministry of Communication and Information

Jl. Medan Merdeka Barat No.9, Jakarta E-mail:


dita001@kominfo.go.id 1, onny002@kominfo.go.id 2

Manuscript received on September 16, 2017, revised on November 27, 2017, approved on December 15, 2017

Abstract

Media Monitoring became one of the tasks and functions of the Ministry of Communications and Information Technology, in accordance with
Presidential Instruction No. 9/2015. The process should be done in short time while maintaining or even improving the accuracy of analysis of
the media. Therefore Big Data technology becomes a promising solution, related abilities of Big Data to process variety of data in a large scale,
also provide accurate reports for stakeholders. This study adopted the Modified Waterfall method that commonly used in manufacture of
software. This method is expected to explores and creates an appropriate recommendation. Big Data Implementation process will require time,
cost, and human resources, so that stakeholders and related user is expected to prepare the implementation process properly in order to be
effective.

Keywords: Media Monitoring, Big Data, Modified Waterfall method, Implementation

Abstract

Media monitoring is one of the duties and functions of the Ministry of Communication and Information Technology, in accordance with
Presidential Instruction No. 9 of 2015. The process of monitoring the media must be completed quickly without reducing the accuracy of
the analysis of the media. Hence technology Big Data is one of the promising solutions, related to its nature that is able to process data
on a very large and varied scale and provides accurate reports for use by policy makers. This research adopts the method

Modified Waterfall which is commonly used in manufacturing software. This method is expected to explore, explore and produce suitable
alternative recommendations. Implementation process Big Data will require a lot of money, human resources, and time, so that the policy
makers and user related parties are expected to be better prepared so that the implementation process can run effectively and optimally.

Keywords : Media Monitoring, Big Data, Method Modified Waterfall, Implementation

as well as areas that can be reached, especially in the


PRELIMINARY distribution of information across countries. People in
one country cannot easily obtain and access
Development in use information relating to other countries, and vice versa.
Information and communication technology has changed the The development of the internet in an era of advanced
way of communicating, especially in dissemination technology allows the circulation of information that is
information. At first method increasingly large, fast, and almost unlimited by space
communication and information distribution is limited to and time. This causes the information circulating from
written media (paper, letters) and electronic media (radio, day to day to be very large, covering various
television and telephone) so that the information circulating information in the field
is still very limited, both from the scale of information in
circulation.

81
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

social, political, economic, technology, science, food, effective way to process it, especially if the information
and so on. generated from the data is needed to help make
Similar conditions also occur in the scale of decisions
government. The last few years have emerged for stakeholders Policy.
appeal to divert It takes a fast and precise way to be able to process
information / documents physical Becomes data the data into information. Hence the principle Big Data
electronic. One of the purposes of this appeal is to
facilitate management and use of data when needed. very suitable applied, where the principle Big Data that
On the one hand, this can be done immediately if the is, to be able to manage very large and varied data,
type and format of all data is uniform. But on the other and process it into the information needed in time
hand, the more various types and formats of data, the
greater the amount and variety of data it will take a that very short. With
long time to process the data into structured and find alternative implementation options Big Data which is
readable information. suitable to be applied to the Ministry of Communication
and Information Technology, especially in the Media
Monitoring section, is expected to help user related in
making
It is undeniable, information is a very decisions through implementation Big Data.
important weapon today. Organization
as government of course Research methods
has a very diverse and large amount of data. To be This study uses a qualitative approach to
able to make the right decisions, in this case the explore existing needs. In finding alternative
government implementation, this study uses methods Modified
as maker Policy, Waterfall. Because the limitation of this research does
requires an effective way to manage the data into not cover the creation and implementation of the
useful information as a consideration in making system, if it refers to
decisions.
Modified Waterfall, stage done
The media is one part of the government's only limited to the stage System Design.
attention, one of which is stated in Presidential Therefore, by referring to this method, and considering
Instruction No. 9 of 2015 concerning Management of the research boundaries set, the research stages are
Related Public Communications adjusted as in Figure 1.
Duty and function Ministry
Communication and Informatics, namely:
1. Review data and information Team Training FGD

that be delivered
Research

ministries and non-ministerial government


agencies;
2. Performing Media Monitoring and analyzing
Formulation Design Strengthening
media content related to Needs choice
Collection
selection

Policy and program Related Directorate


alternative
Data
alternative

government;
3. Develop a single narrative related to government policies
and programs to the public in accordance with the
direction of the President. Formulation

Recommendation

Management of diverse data with a very large


amount requires a Picture 1. Study Method Approach

82
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

Study of literature The government owns it


Good governance can only occur when Lots information to be delivered
decision making is based on adequate information and requires an effective way and role of government
independent judgment (Sullivan in Subiakto, 2014: spokesperson to convey this information to the public.
248). This can be achieved by the existence of factual Theoretically, government public relations or Government
and reliable information, which can only be obtained Public Relations (GPR) have a duty
from a free press, which functions as watchdog
to explain the impact of government programs and
policies on its citizens, including controversial issues
society over government. In a country that guarantees that circulate.
freedom of press and information, the government must be The Working Cabinet Government is
ready against circumstances designing so that the abundant data in conventional
the. This is where the role of government information and new media can be used as a form of public
institutions, namely as a center of communication with opinion that can form a positive image for the
the public in explaining government plans and government. Through Presidential Instruction Number
programs to the community 9 of 2015 concerning Public Communication
Public could understand Management, the Ministry of Communication and
the influence and role of these policies in their lives. Information Technology has been assigned to monitor
the media.

Figure 2. Public Relations Elements

Media Monitoring The data contained in the Directorate of


According to the Directorate General of Information Management and Provision (PPI) is
Information and Public Communication (Ditjen IKP), unstructured with a very fast increase in the amount of
what is meant by Media Monitoring is the activity of data. The data must be processed in a short time to
monitoring content circulating in the media, both print, become information. It takes a statistical method that
broadcast and o. nline and social media. As for is accurate in displaying the information, which can
monitoring the media, with reference to the public help
relations elements as shown in Figure 2.
stakeholders Policy in

83
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

receive the information needed to make decisions. called Pamedi ( Paques Media Intelligence).

For this reason, the PPI Directorate at the


Directorate General of IKP has collaborated with PT. Eight
Eleven Indonesia in 2015 until
2016, and produce a Media Monitoring management
application with Big Data that

Figure 3. Display Dashboard Pamedi

Pamedi ( Paques Media Intelligence) • Expert Analysis & Reporting: Pamedi also allows
Pamedi as a tool used for media monitoring at to make a report
the PPI Directorate has been implemented since 2015 that could changed corresponding
(see Figure 3). Pamedi's abilities are divided into three 1: needs complete with graphics.

Characteristics Big Data


• Media Monitoring: a systematic monitoring of A data is categorized as " Big Data "Not only
topics or issues on social media and news media, because of the large amount of data. There are
several distinguishing characteristics Big Data with
which allows users to access them systematically real-time
information and measurement tools in the app to other systems.
enable users to make better decisions and System Big Data have Volume very large
respond to important situations. data, which usually exceeds ordinary servers in
general and this data will continue to grow every day.
Data can reach more than 100 TB and is usually
stored on external infrastructure (not on
• Media Measurements & Analysis:
make it possible user to present graphs with maintain alone). 2
quality analysis based on the results obtained in other than that Big Data also has varying data
the application / user account. ( Variety), with format
as well as very diverse types of data

2 Taken from Prof. Dr. Mochamad Ashari about Big Data Industry
1 Taken from the presentation of PT. Eight Eleven Indonesia about
and Academic Point of View,
Pamedi on 2 September 2015 in the 2015 Indonesian Big Data Conference.

84
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

so that it requires a special process to be able to RESULTS AND DISCUSSION


process it. Big Data also must be able to process the
data in a very fast time ( Velocity) so that data can be In accordance with Presidential Instruction No. 9 years
useful not only because of the information it produces 2015, the duties of Public Communication
but also because of the speed required to process it Management include; Reviewing data and information
into that information. submitted by ministries and non-ministerial
government agencies; and Doing
Media Monitoring and
Characteristics Big Data the fourth is the truth analyze media content related to government policies
of the data itself and programs. In carrying out these two tasks, it was
( Veracity). Information that is processed from this found that " gap "
data in order to become useful and reliable Among Policy and program
information, we must also look at the source of the government that walk with issue
data used. media coverage, so there needs to be anticipatory
Hence on Big Data, the truth of the data is one thing efforts by the government. Therefore, media
that must be considered. monitoring is an important part of the Ministry of
Communication and Informatics, to be precise for the
PPI Directorate, DG IKP. This need has been
expressed in a Business Process as shown in Figure
4.

Figure 4. Media Monitoring Business Process

Monitoring process the media analysis results, namely Public Issues Monitoring (MIP) and Media
involving 22 media with print, 5 details: 12 media Content Analysis (MCA).
media online e, and 5 selected asMedia TV. Media
a representative sample. This is a medium 1. Public Issues Monitoring (MIP)
matter due to nature MIP is part of the media monitoring process
today's media tend to have a can carried out through tracing headlines news. Criteria headlines
grouped based on
the preaching. Broadly speaking, the media that is being monitored ie headlines news that
monitoring process resulted in two reports contains information or issues

85
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

is being reported by the media. So far, the process of a medium with other media. New business processes
determining issues has been carried out by looking at in the work of MIP with intervention Sentiment Analysis
the subjectivity of news readers. If applying Big Data by and SNA,
method shown in Figure 5.
Sentiment Analysis can speed up the process of news
analysis in the processing of MIP reports, by 2. Media Content Analysis ( MCA)
translating one's point of view into machine language. Media Content Analysis (MCA)
is a process monitoring media
based on the content or content of the news. The
news content that is seen is the content of the news
with headlines the most. As with MIP, method
application
Sentiment Analysis and Social network
Analysis ( SNA) can speed up the analysis process and
enrich the results of news analysis.
The difference with MIP lies in the object of
analysis. If the MIP analysis refers only to headlines news,
then MCA is generated from the analysis of the content of a
news item in the media.

Implementation Stages Big Data


Because Big Data covers a very wide area,
Figure 5. MIP Business Process then its implementation in this study will be
categorized into three stages, namely IT Management
Besides Sentiment Analysis, method use Social & Governance, HR and Systems, as shown in Figure
Network Analysis ( SNA) can simultaneously enrich the 6.
results of the analysis, one of which is by providing
information about the relationship between one issue in

Figure 6. Implementation Stages Big Data

86
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

IT Management and Governance Technology have role urgent


a. IT Management on the continuity of information, starting from an
Gartner describes dimensions Big Data as 3V information being created to being destroyed.
that is Volume, Velocity, Variety ( Gartner, 2009). Enterprise successful ones
Along with its development, Big Data not only treats IT as a significant part of carrying out business
processes. Business processes and IT must
covers 3V but expands to 5V ie collaborate and work together
Volume, Velocity, Value, Veracity, and Variety. so IT can get into
In technical, Big Data is a in governance and management.
large data sets, whether structured, semi-structured, or COBIT have
5 some
unstructured, so that they cannot be processed using the the levers derived from purpose
device relational database ordinary (Nitin Sawant, 2013). organization that has been defined. The lever is an
influencing factor
The data that appear have the opportunity to governance and management from enterprise
be able to provide policy guidance without being IT, namely:
aware of it before (Milton, 2009). Big Data 1) Principles, policies and frameworks;
2) Process;
is a technology trend to take a new approach in 3) organizational structure;
understanding the world and making business 4) Culture, ethics and behavior;
decisions (John FO, 2013). These decisions are made 5) Information;
based on a very large volume of structured, 6) Services, infrastructure and applications; and
unstructured and complex data (eg tweet, videos, 7) Human resources, capabilities and competencies.
commercial transactions).
According to COBIT 5, that information is
effective information capable
According to Bill Schmarzo, process
meet the needs of information consumers ( stakeholder).
integration Big Data in an enterprise, it has a business
In the case of Big Data,
maturity index which consists of the following phases
enterprise ( organization) is stakeholder and one of the
(Bill, 2013):
main pillars is the quality of information. Big Data
harvesting is expected to produce
1. Business Monitoring.
quality information that
2. Business Insights.
support decision making. Good quality information will
3. Business Optimization.
result in good organizational decisions that will
4. Data Monetization.
increase profits enterprise.
5. Business Metamorphosis

Big Data is a process of collecting data to find


b. IT Governance with framework COBIT 5
patterns and correlations that may not be obvious at
COBIT ( Control Objectives for
first, but may be useful in making business decisions.
Information and Related Technology) is one of the
Such data is often useful personal data which can be
frameworks ( framework) created by ISACA which
categorized as
serves to serve as a guide in order to achieve
organizational goals. The organization in question is a enterprise
that is, organizations that carry out IT functions as part
Volunteered data, Observed data or
of their business processes.
Inferred data ( Richard Chew, 2013).

According to COBIT, information is a key


resource for a enterprise.

87
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

Human Resources 3) Data Transformation


Due to the complexity of the system Big Data, One of the properties of Big Data apart from a
it takes a variety of technical capabilities to be able to very large data size, also a very wide variety of data
implement it. However because Big Data Engineering itself types. In order for these data to be analyzed properly,
is still something new and is dealing with technology sometimes the data needs to be converted into
and new job positions, so currently there is no another format so that it is possible to analyze it. The
standard specification regarding the HR competencies required competencies include:
needed for this field. Based on the work process of Big
Data in general, namely
• ETL Tools (eg: Informatica, DataStage, SSIS,
Redpoint, etc.)
Collect, Store, Transform, and Analysis 3, then there are • Scripting (eg: Linux / Unix commands, Python,
four things that must be considered related to the HR Ruby, Perl, etc.)
competencies needed in implementation Big Data, including:
4) Data Analysis
The final stage is analyzing the data that has
1) Data Collection
been previously collected and collected, processing
Data to be processed in the system
the data into information, to become statistical results
Big Data usually taken from the website or API ( Application
if needed. The competencies required include:
program interface), in general
with use technique
crawling. The HR competencies required include:
• MapReduce, Hadoop, Cloudera, IBM Big Insights,
Hortonworks, MapR, etc.
• APIs data
• Data mining or machine learning (eg: Mahout,
• SQL and Data Modeling
Neural Network, etc.)
2) Data Warehouse • Statistical analysis software (eg: R, SPSS, SAS,
Data that has been retrieved from various Weka, MATLAB, etc.)
sources will be stored in server • Programming skills (eg: Java, Scala, Ruby, C
which has been prepared for the system Big Data. ++, etc.)
Correspondingwith his name, Big Data
requires a very large data storage capacity, because System
the entry of very large and varied data into server If Big Data implemented for media monitoring,
then what will be the core of the system is the method
everyday. One of the reasons it is called used to analyze the text ( Text Analysis). There are
Data Warehouse because the process of storing, many methods and algorithms that can be used to
processing, and retrieving data from server analyze text, depending on what kind of results we are
will be very different from Database the usual. The expecting.
competencies required include:

• Relational Databases (eg: MySQL, MS SQL One of the processes in Public Issues
Server, Oracle, DB2, etc.) Monitoring is analyzing issues in a news. One method
• NoSQL (eg: HBase, SAP HANA, HDFS, Cassandra, that can be used in analyzing this issue is to use a
MongoDB, CouchDB, combination of them Social Network Analysis to see
Vertica, Greenplum, Pentaho, Teradata, etc.) the network of these issues and methods Sentiment
Analysis, which is a study of analysis of opinions,
sentiments, evaluations, praise, attitudes and
3 " The Key Skills Needed by Big Data Engineers ”.
emotions of people
http://insights.dice.com/2014/08/21/key-skills-needed-
big-data-engineers /. Retrieved 19 October 2016.

88
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

towards an entity or object which can be a product, seen


could actors that most
service, individual, organization, event, or a topic. Use influence, the relationship between actors, other
parties in the network, and so on. There is a simple
Social Network Analysis ( SNA) looks more at the description of implementation Big Data ( combination sentiment
analysis and SNA) in media monitoring shown in
actors involved in an issue. SNA works by analyzing network
Figure 7 and Figure 8.
(network) formed between actors, so that

Figure 7. Combination Sentiment Analysis and SNA

Figure 8. Simplified description of the combination of SNA with Sentiment Analysis

89
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

1. Social Network Analysis (SNA) In this case because user related is the Directorate of
Use Social Network Analysis PPI at the Directorate General of IKP, so we position
can be applied to view the network ourselves as the government in viewing news. News
(network) which is created between actors involved in that is considered positive by the government may be
an issue, as shown in the illustration in Figure 9. viewed negatively by the public. Vice versa. That is
why the subjectivity in viewing news needs to be
clarified so that it can be translated into a machine.

Figure 9. SNA illustration

Figure 10. News Readers Subjectivity


Social Network Analysis could
present complex matters, such as the various roles of
There are two important things to that should
actors in social networks, contexts,
note, namely:
community, and etc.
1. The Collective Viewpoint As
Its application in Media Monitoring will make it possible
government, angle view that
to see which social networks are involved in a
collected from several individuals forming a
particular issue. Through the SNA, two important points
collective point of view (for example by forming
that will change the scale of an issue can be seen,
a News Assessment Team). The difficulty is a
including:
point of view that must be mutually agreed
• If the same issue is discussed by many people
upon. Therefore, the selection of competent and
in a short time, then the issue has high potential
representative Team members is very
to become a large-scale issue.
important, so that the collective point of view
that is formed can truly represent the
• If there is an issue that is discussed by actors
government.
who have a network of thickness ( density) large
network, it is likely that the issue will be
2. Individual Perspectives
discussed by people who are included in the
Using individual points of view will be much
network (eg followers on twitter).
easier to translate into
language machine. Although
One person's subjectivity will be different from
2. Sentiment Analysis the others, but the person's subjectivity that is
One of the most important things before doing determined can still be accepted and used as
Sentiment Analysis is to determine the subjectivity of an angle
the news reader, as to whom we will see news.

90
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

government perspective in viewing news. As a concept illustration Naïve Bayes


Classification, existing objects can be pre-labeled
Sentiment Analysis make it possible user first (Dell. 2015). To
to analyze an issue by breaking a sentence into word makes it easy, the object illustration is labeled BLUE
for word. Method and RED. In Bayesian Analysis, known as prior
Sentiment Analysis which can be used and in accordance with the probability ( initial probability). Because of the number BLUE
duties and functions of the Communication and Information Media double the amount
Monitoring include: RED, hence it is believed that the new object is more
likely to be labeled BLUE. It could also be written like:
• Naïve Bayes Classifiers
Naïve Bayes Classifiers is a
a method derived from the Bayesian theory. This model is
Number of BLUE
based on conditional independency resulting from
Prior probability objects
from predictions class target. =
for BLUE Total number of
Based on Bayes' theory, posterior probability can be objects
written as follows:
Number of RED
Prior probability objects
=
for RED Total number of
In language, the above equation can be simplified objects
into:
If it is applied in analyzing the media, the first
step to take is to label the inner texts Corpus used (for
example: negative or positive). After labeling, done
Posterior Probability or that opportunity
appears after testing, in Bayes theory results from the
initial probability multiplied by the probability, and Supervised Training by using
divided by the test result. This is a grouping method dataset which is owned. As well as Machine Learning, the
more trained, the more accurate the results will be.
in a manner statistics that Especially on Naïve Bayes Classification. This method
used to predict opportunity has characteristics that when used on a large scale
new membership that appears in a group. Dell also data will produce more accurate results.
put out textbook
about Naïve Bayes Classifier that
described simply. 4
• Standford's Sentiment Treebank
Different from Sentiment Analysis
which generally breaks down sentences and looks word
for word, this method sees the text in a complete
sentence structure. The word structure in a sentence is
not removed, so that the meaning of a word can be
seen in one complete sentence.

Figure 11. Illustration of object classification Set words classifiers could


works well on long sentences with
rely words that
simple but has strong sentiments such as "great" or
4 Statistics - Textbook, Naïve Bayes Classifier. Dell. "extraordinary". However the resulting accuracy for
https://documents.software.dell.com/statistics/textbook/n classification one
aive-bayes-classifier # technical notes

91
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

sentences in determining the negative or positive government which holds one of the strategic
sentiment never exceeded 80%. In the case of sentences sectors, namely information, the choice to build its
with tone neutral for short messages on twitter, the own Big Data system can be considered. The
accuracy obtained even tends to be below 60% (Wang et advantages of this option include:
al., 2012) 5

• Building a system can be an investment in the


management of public information, which will
become more complex as the communication
media develops

• Information security is more guaranteed because the


system that is built is owned by yourself

• Data sovereignty in the hands of the government


that built the system. Costs
• more affordable if
versus subscription service continuously
Figure 12. Example results Standford's Sentiment
Treebank 6
• The dependence of the government on
reduced private sector, because it is managed
Alternative Implementation Options Big Data
independently
Whatever method is to be used for text
While the weaknesses include:
analysis, corpus remains a crucial element. This is
• Construction takes time
related to variation corpus in Indonesian which is still
long until it can be used
very limited. As for corpus which is quite complete
• Requires human resources with certain areas
usually owned by vendor certain and closed to the
of expertise
public. Making Corpus
• Requires a large server to store and process
data. Big Data requires reliable server
the very urgent, because could
capability and capacity so that the data
determine accuracy the result Sentiment
obtained can be processed quickly and
Analysis.
accurately
The last stage is one of the important
considerations related to how the Big Data system will
• Requires assurance from top management
be implemented. There are three alternative options
to could ensure
that can be taken into consideration, including:
it gives support, both in terms of governance, human
resources, and costs
1. Build Your Own System Namely building a system Big
• There must be someone escorting the
Data as a whole from the start, although in the
manufacturing process from start to finish so that
process it can be done hire third party. As an
the final result can be suitable Need to provide
agency
• adequate infrastructure (Network, Server, etc)

2. Subscribe
5 S. Richard, P. Alex, WY Jean, C. Jason. Recursive Deep Models Currently there are several companies / vendors
for Semantic Compositionality Over a Sentiment Treebank that offer this Big Data service.
The vendor notable including:
6 Screenshot taken from the test results on the Treebank website
• i811
Stanford.
• Indonesia Indicator / ebDesk
http://nlp.stanford.edu/sentiment/treebank.html. Retrieved 15 June
2016
• Mediatrac

92
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

• Maverick • In terms of information security, important


• Mediawave system parts can be owned alone (eg: Data
• Incentia Sovereignty, Corpus,
• Awesometrics etc)
• IMMC (Indonesia Media Monitoring Center) • The system can be made according to your needs / wants

• IMM (Intelligence Media Management) • As users will more


understand the system, because it follows the
Despite the drawbacks on
manufacturing process from start to finish Does not
each vendor, the option to subscribe to remain an
• require a lot of technical human resources, because the
option if the services provided can be on custom
work will be done share
with outsiders
as needed. The advantages of this option include:
While the weaknesses include:
• There must be someone overseeing the
• System management and reliability are guaranteed by
the service provider manufacturing process from start to finish so
that the final result can match expectations
• HR is needed only as users, so there is no
need for high qualifications of technical
• Requires assurance from top management
human resources. The system can be used
to could ensure
• immediately server which is great due to data
it gives support, both in terms of governance, human
• processing on the part of the service provider
resources, and costs
• Need to provide adequate infrastructure
(Network, Server, etc)
While the weaknesses include:
• Higher costs (Determined by the provider, Because the process is a
estimates for services that have been used cooperation between the government and outside parties, it
such as IMM around 30 million per month)
is necessary to have a written commitment
that made detailed to
• The survival of the system is determined ensure the system created together can be completed
by provider service, so that and avoid problems that can arise in the future.
cause dependence on
vendor
• Raw data, algorithms, and corpus,
held by the service provider, as it is core that CLOSING

traded. Conclusion

• Requires detailed system adjustments in In system Big Data which will be implemented
order to service a Big Data system for media monitoring,
given corresponding with Text Mining by method Sentiment Analysis
needs. Because there will be different uses will be core in that system. There are two methods Sentiment
and needs between Analysis most suitable to apply, that is Naïve Bayes
corporate with the government. Classifiers and Stanford Treebank.
3. Build Partly
The government builds part of the owning system Method Sentiment Analysis the most suitable
score strategic for for the monitoring media are Stanford Treebank, because
government (such as Corpus), while the rest is this method can see the relationship between words in
built by outsiders. The advantages of this option each sentence, it is different from other methods which
include: generally interpret sentences by breaking up

93
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

words. However, because the method is still in the large allocation of funds for vendors each year.
manufacturing stage Corpus
used in English, then Making Corpus alone can
Naïve Bayes can be an option to be applied for media conducted independently or in collaboration with the
monitoring at this time. university, so that the government and academics can
There are three alternatives related to the share knowledge and resources ( resources).
stages of system implementation Big Data that can be
done, among others: building your own system,
subscribing to an existing one, or building a partial
system. Each of these alternatives has advantages Thank-you note
and The author would like to thank the Aptika and IKP
weakness that need to pay attention to Research and Development Center, Human Resources Research
into consideration. and Development Agency, Kominfo, where this paper is part of the
Team's research that has been carried out at the Aptika and IKP
Suggestion Research Center with the title "Big Data System Implementation
Method Sentiment Analysis best combined Study
with Social Network to Support Policy
Analysis, so that the output the result is not only stop Communication and Informatics " which is financed from
at monitoring the media, but until finding key person or DIPA 2016 Aptika Puslitbang and IKP. The author also
actors related to these issues. It would be a pity if resource thanks the research team which consists of researchers at
amount that has been allocated the Aptika Research and Development Center and IKP and
the team from Telkom University.
only used to
media monitoring. Utilization Big Data can be more
optimal if expanded to do REFERENCES
predictive analysis in estimating
Ashari, Mochamad, 2015. Big Data Industry
the potential of an issue becomes large. Going forward
and Academic Point of View. Indonesian Big Data
through the system Big Data, Kominfo together with other
Conference.
government agencies can anticipate and react more quickly
to issues that are predicted to become large, so that they IBM Big Data & Analytics Hub, 2013. 4v's of
can be controlled. Big Data.
www.ibmbigdatahub.com/infographic/four
Process in the system Big Data which will be - vs-big-data. Retrieved 17 February
implemented simply can be divided into three 2016.
categories, namely Input, Process, and Output. The Presidential Instruction No. 9 of 2015 on
recommended combination of methods is at stage process. Public Communication Management.

Expected recommendation this could Krishnan, K., 2013. Data Warehousing in the
maximizing function Big Data without Age of Big Data. USA: MK Publications.
change the main function of media monitoring. Liu, Z., Ping, Y., Lixiao, Z., 2013. A Sketch of
Implementation Big Data To support media monitoring, Big Data Technologies. Seventh
it is best to build a partial system, so that the important International Conference on Internet Computing
part of the system is like Corpus can be owned by for Engineering and Science. School of
yourself, and the system that is built can be designed Information Science and Technology, Shanghai
and supervised as needed. This option can reduce the Sanda University Shanghai, China.
government's dependence on other parties, as well as
pressure

94
STUDY OF SYSTEM IMPLEMENTATION BIG DATA TO SUPPORT THE POLICY
COMMUNICATION AND INFORMATICS
Dita Kusumasari and Onny Rafizan

Matsudaira, Kate, 2014. " The Key Skills Schmarzo, Bill, 2013. Understanding How
Needed by Big Data Engineers ". Data Powers Big Business. USA: John Wiley &
http://insights.dice.com/2014/08/21/key- Sons, Inc.
skills-needed-big-data-engineers /. Accessed
Schonberger, VM, & Kenneth, NC, 2013. Big
October 19, 2016.
Data: A Revolution that Will Transform How we
Morrisan, CW, Andi, & H., Farid, 2010. Live, Work, And Think. New York, USA:
Mass-Media Communication Theory, Culture and Houngthon Mifflin, Harcourt Publishing.
Society. Jakarta: Ghalia Indonesia.

Munassar, Nabil, MA, & A., Govhardan, Sommerville, Ian, 2011. Software engineering
2010. A comparison Between Five Models Of 9th Edition. Boston: Addison-Wesley
Software engineering. IJCSI Publishing Company.
International Journal of Computer Science Issue, Vol.
Statistics - Textbook, Naïve Bayes Classifier.
7, pp. 95.
Dell.
PT. Eight Eleven Indonesia about https://documents.software.dell.com/statist
Pamedi. 2 September 2015 ics / textbook / naive-bayes-
classifier # technical notes. Retrieved 15 June
Prajapati, V., 2013. Big Data Analytics with R
2016.
and Hadoop. Birmingham, UK: Packt Publishing.
Subiakto, Henry & Rachmad, I., 2014.
Political Communication, Media and Democracy
Romney, Marshall, B. & Paul, JS, 2005.
(Print Second). Jakarta: Golden
Accounting Information Systems (9th Edition). Jakarta:
Prenadamedia Group.
Four Salemba.
T., Firat, & Keane, JA, 2013. Big Data
Sagiroglu, Seref, & Duygu, S., 2013. Big Data:
Framework. IEEE International
A Review. IEEE International Congress on Big Data,
Conference on Systems, Man, and Cybernetics.
Gazi University, Department of Computer
School of Computer Science, The University of
Engineering, Faculty of
Manchester, UK. pp. 1494-1499.
Engineering, pp. 42-47.

Sawant, N., & Himanshu, S., 2013. Big Data


Law Number 14 of 2008
Application Architecture Q&A. New York: Springer
regarding Freedom of Information.
Science Business Media.
Zikopoulos, Paul, C., 2013. The Power of Big
Schell, R., 2013. Security - A Big Question for
Data: The IBM Big Data Platform. USA: Mc Graw
Big Data. IEEE International Conference on Big
Hill.
Data University of Southern California, USA.

95
Journal of the Telematics and Information Society

Volume: 8 No. 2 (October - December 2017) Hal .: 81-96

96

You might also like