You are on page 1of 19

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1359-0790.htm

JFC
25,3 Parameters of automated fraud
detection techniques during
online transactions
702 Vipin Khattri
Shri Ramswaroop Memorial University Institute of Technology,
Barabanki, India, and
Deepak Kumar Singh
Jaipuria Institute of Management, Lucknow, India

Abstract
Purpose – This paper aims to provide information of parameters and techniques used in the automated
fraud detection system during online transaction. With the increase in the use of online transactions, the
concerns regarding data security have also increased. To tackle the frauds, lot of research has been done and
plethora of papers are available on the related topics. The purpose of this paper is to provide the clear
pathway for researchers to move in the direction of development of automated fraud detection system to
prevent the fraud during online transaction.
Design/methodology/approach – This literature review analyses and compares the different types of
techniques for detecting fraud during online transaction. An in-depth study of the most prominent journals
has been done and the core methodology of the papers has been presented. This article also shed some light on
different types of parameters used in fraud detection techniques during online transaction.
Findings – There are vast varieties of various fraud detection techniques, and every technique has
completed task in its own way. After studying approximately 41 research papers, 14 books and four reports,
in total 30 parameters have been identified and a detailed study of the parameters has been presented. The
parameters are also listed with their details that how these parameters are used in the security system for
detecting online transaction fraud.
Research limitations/implications – This paper provides empirical insight about the parameters and
their prominence in the development of automated fraud detection security system of online transaction. This
paper encourages the researchers to development of improved fraud detection system.
Practical implications – This paper will pave the way for researchers to do a focused research on the
fraud detection methodologies. The analysis will help in zeroing down the most prevalent topic of research in
this field. The researchers will be able to understand the internal details of parameters and techniques used in
the fraud detection systems. This literature also helps the research to think in a variety of ways that how these
parameters will be used in the development of fraud detection system.
Originality/value – This paper is one of the most comprehensive reviews in its field. It tries and attempts
to fill a void created because of lack of compilation of the laid fraud detection parameters.

Keywords Online transaction, Credit card, Fraud detection techniques,


Parameters of credit card fraud detection
Paper type Literature review

Journal of Financial Crime


1. Introduction
Vol. 25 No. 3, 2018
pp. 702-720
One of the easiest ways to transfer the money from one account to another account is online
© Emerald Publishing Limited
1359-0790
transaction. This event can be performed from anywhere with the help of computer, internet
DOI 10.1108/JFC-03-2017-0024 or transaction device. Online transaction is a suitable and simplest way to perform online
payment. You can use credit card, debit card or just use small credentials of your account for Automated
online transaction. Besides all security measures, different types of frauds have been fraud detection
reported for online payment. A risk in terms of fraud is associated with every online
transaction (Papadopoulos and Brooks, 2011; Barker et al., 2008). It means fraud is an illegal
techniques
activity (Bhatla et al., 2003; Bai and Chen, 2013) that produces loss in terms of trust, money,
time and resources, etc. In online transaction fraud, candidate has to bear financial loss.
Euromonitor International (Delamaire et al., 2009) quoted that 120 million credit cards, debit
cards and charge cards were brought into use in Germany in 2004. The total transactions 703
were performed by cards approximate e375bn in 2004 which was 4 per cent increase from
2003. This increase of credit card is because of high use of credit card and easily achievable
of credit card fraud. Annual report provided information (Financial Fraud Action-UK, 2017)
that only in the UK-card fraud loss was £467m in 2006 and it was increased to £567.5m in
2015. Credit card fraud may be achieved in various ways such as bankruptcy (a person who
is not able to pay for transaction using credit card), theft fraud (credit card is used by
someone else), counterfeit fraud (using credit cards details) and application fraud (credit
card is applied with false information). In current scenario, there is huge increase in number
credit cards, debit card and point of sale machine for performing online transaction as a
result online transaction fraud is also increasing. There are various online fraud detection
techniques available and performs well but society requires an improved fraud detection
system because fraud is becoming more sophisticated. The focus of this paper is to analyzed
different online fraud detection techniques using various methodology and parameters
(Oracle Financial Services Software, 2014) use to present the fraud during online transaction.
Information flow of this paper is organized in five sections as follows. Section 2 analyzed the
online transaction fraud detection techniques. This section also gives the comprehensive
summary of different fraud detection techniques. Section 3 analyzed the parameters used in
different online fraud detection techniques. This section also gives the comprehensive
summary of parameter used in different fraud detection techniques. Section 4 contains the
conclusion. Finally, Section 5 gives the future scope of this study for the researchers to move
forward on clear path way.

2. Fraud detection techniques


Online transaction fraud detection technique is an automated screening tool that is used to
identify whether the online transaction is fraudulent or not. It is very helpful when fraud is
detected immediately during online transaction. Therefore, financial institutions are always
looking better solution for identification of fraud. As fraudster uses sophisticated methods
for committing fraud, new improved technologies need to be identified accordingly.
As emerging of new technologies, new innovations are inevitable for detecting the
fraudulent transaction because fraudsters are committing fraudulent transaction in a new
way every time. The various techniques are used to find the fraudulent transaction such as
hidden Markov model (Agrawal et al., 2015; Prakash and Chandrasekar, 2015; Dhok and
Bamnote, 2012; Prakash and Chandrasekar, 2012; Khan et al., 2014b), artificial neural
network (Behera and Panigrahi, 2015; Bekirev et al., 2015; Carneiro et al., 2015; Van
Vlasselaer et al., 2015; Khan et al., 2014a; Brause et al., 1999; Aleskerov et al., 1997), big data
(Chen et al., 2015), genetic (Assis et al., 2014; Ramakalyani and Umadevi, 2012; Duman and
Ozcelik, 2011), artificial immune system (Halvaiee and Akbari, 2014; Soltani et al., 2012),
Bayesian (Singh and Singh, 2015; Renuga et al., 2014; Panigrahi et al., 2009), migrating bird
optimization (Duman and Elikucuk, 2013; Duman and Elikucuk, 2013), data mining (Philip
and Sherly, 2012) and others (Sánchez et al., 2009; Quah and Sriganesh, 2008; Bentley et al.,
2000; Ganji and Mannem,2012). Every methodology has its own characteristics with their
JFC advantages and disadvantages. In this section, the researcher analyzes different techniques
25,3 and discuss below.
In this study, four steps were involved (Agrawal et al., 2015) to implement fraud
detection. In the first step, the author analyzed the shopping behavior of the customer, i.e.
which type of product customer want to purchase. In the second step, analyzed the spending
behavior of the customer, i.e. how much money is required to purchase the product. In the
704 third step, hidden Markov model (Das, 2013) is used to categories the behavioral pattern of
the customer in a clustered way. In final step, the genetic algorithm (Gen and Cheng, 1997;
Goldberg, 2006) is applied to find the fraudulent transaction on the basis of scoring the
parameters (behavioral pattern) used in the study. The genetic algorithm worked on the
previous transaction data managed by hidden Markov model.
The work of authors (Khan et al., 2014b) proposed the hidden Markov model to identify
the fraudulent transaction. This model works on the basis of current spending pattern and
compares it with previously recorded spending pattern to find suspicious transaction. K-
means clustering algorithm (Everitt et al., 2011; Agarwal and Reddy, 2013) is used to cluster
the spending pattern. The training of hidden Markov model is done by using Baum–Watch
algorithm (Das, 2013) to identify fraudulent transaction.
The authors (Behera and Panigrahi, 2015) implemented a novel approach in three phases.
In the first phase, it checks the authentication details and verification details of the card. In
the second phase, it analyze the transaction pattern and process it to classify using fuzzy
C-means clustering algorithm (Sadaaki et al., 2008) which is used to differentiate the
transaction pattern as normal pattern or suspicious pattern based on past transaction
history of the customer. In third phase, a neural network approach (Bianchini et al., 2009;
Braspenning et al., 1995; Hassoun, 1995; Graupe, 2013) of feed forward with back
propagation and supervised learning algorithm is applied on suspicious transaction pattern
to identify that whether the suspicious transaction is actually fraudulent or not. The author
stated that by applying this hybrid approach, the author can identify fraudulent transaction
as well as reducing the cases of false fraudulent transaction. The author quoted the result
that 93.3 per cent correct and 6.10 per cent incorrect classified transaction found.
The authors (Khan et al., 2014a) implemented simulated annealing algorithm
(Van Laarhoven and Aarts, 1987) to train the neural network by adjusting the weight of the
network for detecting the credit card fraud. Metropolis–Hasting algorithm for simulated
annealing is used to evaluate the best configuration of weights for artificial neural network.
The main purpose for using simulated annealing algorithm is to handle the problem of local
minima and use less time to perform. The author reported 92 per cent success rate for
detecting the fraud cases and 85 per cent success rate for detecting the non-fraud cases.
In this study, the authors (Özçelik et al., 2010) undertook the credit card fraud detection
problem of a bank and tried to improve the performance of the existing solutions. The
digressed from the classical method and did not undertake the typical objective of
maximizing the number of correctly classified transactions; rather they defined a new
objective function. In this novel objective, the misclassification costs are variable, and thus,
correct classification of some transactions is more important than correctly classifying the
others. The genetic algorithms were used which is a novel one in the related literature both
in terms of the application domain and the cross-over operator used. The algorithm was
applied to real life data and the authors claimed that the savings obtained are almost three
times the previous practice.
This study (Duman and Ozcelik, 2011) developed a technique that improved the credit card
fraud detection. The author aimed to reduce the incorrect classification of suspicious
transaction for which each transaction is scored and based on these scores the transactions are
classified as fraudulent or legitimate. The author aimed at minimizing the misclassification Automated
cost. To achieve their objectives, the author implemented two different techniques in the fraud detection
combination named genetic algorithm and scatter search.
In this study, (Singh and Singh, 2015), the author implemented Bayesian learning
techniques
approach to classify the current transaction into suspicious or non-suspicious transaction
based on two parameters i.e. geographical location and customer spending behavior. These
parameters are clustered by using k-means clustering algorithm. The author analyzed the
behavior of customer on the basis of spending category as high spending profile, medium
705
spending profile and low spending profile. It means that if salaried employee is categorized
as medium spending profile then customer does not spend their whole income to purchase
the product. If any deviation occurs in spending profile then transaction is considered as
suspicious transaction category.
In the study of Renuga et al. (2014), the authors implemented Naïve Bayesian Classifiers
and Random Forest approach that are used to track the transaction behavior of the customer
for detecting fraudulent pattern. The data mining techniques (Li et al., 2013; Zaki et al., 2014)
is used in the proposed approach for detecting fraud. The author implemented two models,
first personalized to track current transaction of every individual and compare with
previous transaction history with respect to the individual and second aggregate model to
track the transaction of every individual and compare with previous transaction history
with respect to the all individuals. The author quoted that the Naïve Bayesian approach
gives the best result in terms of performance for personalized model as compared with
random forest classifier and random forest classifier gives best result in terms of
performance for aggregate model as compared with Naïve Bayesian in detecting fraud.
The study of credit card fraud (Bentley et al., 2000), author analyzed the implementation
of evolutionary fuzzy system (Cao, 2007; Sadaaki et al., 2008) to classify the transaction into
suspicious and non-suspicious category. To perform the clustering (Everitt et al., 2011;
Agarwal and Reddy, 2013; Cao, 2007), C-Link, S-Link and K-means are used and genetic
programming(Chen, 2012) is used to generate fuzzy system rule. This study revealed
that100 per cent detection is achieved in case of suspicious transaction and 5.79 per cent
showing false in case of non-suspicious transaction. The author also stated that two aspect,
i.e. accurate and intelligible classifications of data are performed by the use of evolutionary
fuzzy system efficiently.
The table (Table I) shows chronologic flow of the path followed by researchers in the area
of credit card fraud detection. The table clearly depicts that the focus of this crucial study in
1997 and was classically done using data mining techniques. In the late 1990s, neural
networks started playing a huge role in data mining hence the fraud detection research also
followed the path and started using neural networks. After 2000, the focus moved towards
self-organizing maps and Bayesian algorithm. During this time, some authors also tried to
use fuzzy logic and association rule mining approach to solve the problem. Till 2010, the
problem has taken the shape of optimization problem because most of the authors are trying
to find the attributes responsible for detecting fraud and on that basis, framing an objective
function. This inclination brought genetic algorithm in picture as it was one of the most
suitable optimization algorithms. The artificial neural networks once again gain its
popularity when people started different algorithms of artificial neural networks for
optimization. In the fraud detection field, the researchers started using artificial neural
networks with variations like simulated annealing and Naïve Bayesian approach. In today’s
scenario, the number of credit cards is increasing drastically and hence the data of their
usage, hence the field is now merging into the big data research areas. Currently, the
JFC Methodology/technique Authors (Year)
25,3
Hidden Markov model and genetic algorithm Agarwal and Reddy (2015)
Semi-hidden Markov model Prakash and Chandrasekar (2015)
Fuzzy clustering and neural network Behera and Panigrahi (2015)
Neural network committee and clustering Bekirev et al. (2015)
Artificial neural network Carneiro et al. (2015)
706 Cluster analysis and artificial neural networks Véronique et al. (2015)
Big data Chen et al. (2015)
Naïve Bayesian approach Singh and Singh (2015)
Artificial neural network tuned by simulated annealing algorithm Khan et al. (2014a)
Genetic programming Carlos et al. (2014)
Artificial immune systems Soltani et al. (2014)
Naïve Bayesian approach and random forest Renuga Devi et al. (2013)
Migrating birds optimization Duman et al. (2013)
Hidden Markov model Dhok and Bamnote. (2012)
Hidden Markov model Prakash et al. (2012)
Artificial immune system Soltani et al. (2012)
Anti-k nearest neighbor algorithm Gnnji et al. (2012)
Genetic algorithm and scatter search Duman et al. (2010)
Table I. Genetic algorithm Ozcelik et al. (2009)
Bayesian learning algorithm Panigrahi et al. (2009)
Summary of
Association rule mining Sanchez et al. (2008)
methodology/ Self-organizing maps and predictive algorithm Quah and Sriganesh (2008)
technique used for Fuzzy Darwinian Bentley et al. (2000)
detecting online Adaptive neural data mining Brause et al. (1999)
transaction fraud Neural network-based database mining Aleskerov et al. (1997)

researchers are focusing on amalgamation of two or more technologies to detect and


minimize the frauds.

3. Parameters used for fraud detection technique


One of biggest challenges for financial institute is that how to prevent the online transaction
fraud. The fraud activity has negative impact on card holders, merchants and financial
institution. The impact on card holder is least by legislation prevailing while merchants
have negative impact in terms of cost of goods sold, shipping cost, card association fees,
merchant bank fees, administrative fee and loss of reputation and financial institution has
some impact such as administrative manpower cost. The financial institution has given
great importance to implement secure electronic transaction (Bhatla et al., 2003; Oracle
Financial Services Software, 2014) and monitor to find fraudulent transaction. For detecting
the fraud, different types of techniques on the basis of parameters are used by the financial
institution which could analyze transaction during online payment. These parameters play
an important role for finding online payment fraud detection. These parameters are
discussed as follows.

3.1 Address verification system


This is used in case when card not present and this is helpful to identify the international
fraudulent transaction (Bhatla et al., 2003). This is one of the important parameter used in
fraud detection and implemented in online fraud prevention tool (Panigrahi et al., 2009). This
parameter is used to verify address mention for billing of purchase with address information
mention with the card issuers (VISA, 2011). If any mismatch found then transaction will be
categories as suspicious transaction. The authors (Singh and Singh, 2015) used this Automated
parameter to find the mismatch between the billing address and shipping address to identify fraud detection
the current transaction as fraudulent transaction. The authors (Bhusari and Patil, 2011) used
the same to monitor the variation in shipping address, card holder address and billing
techniques
address. If any variation found then transaction is considered as suspicious transaction.

3.2 Card verification value 707


This is a four-digit code which is printed on one side of the card. This code cannot be copied
from skimmed from magnetic strip. This is used to ensure that authorized person is
initiating the transaction and card is in their custody (Bhatla et al., 2003). This parameter is
used to verify the code provided during online transaction with the card issuers
(VISA, 2011). In other words, this code is used to verify that customer has a legal card in
hand during the time of order. As the authors (Panigrahi et al., 2009) reported that this
parameter is used in various real time fraud prevention tools. The authors (Correia et al.,
2015) used this parameter in considering fraudulent transaction in case of many wrong
attempts made by the customer.

3.3 Negative list


This parameter is used to verify that whether pervious transactions was involved in un-
trusted activity such as card number used by another person, card has been reported lost in
the past, card that has chargeback in the past etc. If any such activity was reported then the
account is marked and mention in the negative list. This negative list is used for fraud
detection. This list is helpful to compare the current transaction with past fraudulent history
of account so that further fraud activity could be prevented from this account (Bhatla et al.,
2003; Oracle Financial Services Software, 2014).

3.4 Card frequency


This parameter is used to restrict the number of times card used for online transaction. The
authors used this parameter (Agrawal et al., 2015) to find fraudulent transaction. A value is
calculated on the basis of this parameter and observed that if this value is below than the
predefined critical value then current transaction is considered as non-suspicious
transaction otherwise suspicious transaction. The authors (Panigrahi et al., 2009; Bolton and
Hand, 2001) used this parameter to compare the previously recorded transaction frequency
pattern with high volume transaction frequently. The authors (Correia et al., 2015)
considered this parameter that current transaction is suspicious transaction in case number
of transactions occurred in a short time or same card is used from different location. The
authors (Patel and Singh, 2013) considered the current transaction as suspicious if increase
in card used more number of times. This is used by the authors (Carminati et al., 2014) in
training and feature extraction for identification a transaction as fraudulent transaction. The
author used this parameter (Renuga et al., 2014) to train the model for identification of
fraudulent transaction.

3.5 Location
This parameter is used to record country name and city name where purchase is to be made
and compare it with the list of country and city where suspicious transaction have been
occurred in past Oracle Financial Services Software, 2014). This parameter is used (Sánchez
et al., 2009) to verify the location where the card used was in suspicious category or not. If yes
then transaction can be considered in suspicious category. This parameter used (Bahnsen
JFC et al., 2015) to identify transaction that is to be made from which country so that the
25,3 transaction can be categories as suspicious transaction on the basis of comparing the current
transaction country from list of fraudulent transaction country. The authors (Singh and
Singh, 2015) used to find customer location by using internet protocol (IP) address, this
parameter is used to calculate the distance between two locations with respect to two
consecutive transactions. If calculated distance is required more time to cover it than the
708 observed time then current transaction is considered as fraudulent or suspicious transaction.

3.6 Overdraft
When a person wants to withdraw money from a bank account and available balance is not
sufficient to withdraw, this situation is known as overdraft, and generally, this situation
does not occur frequently. This parameter is used to verify that whether deficit in an account
caused by drawing more money which is beyond the limit. The authors (Ramakalyani and
Umadevi, 2012) used this parameter to find fraudulent transaction. A critical value is
defined for this parameter and on the basis of past and current transaction history, actual
value is calculated. If actual value is greater than critical value then current transaction is
considered as suspicious transaction.

3.7 Daily spending amount


This parameter is used to predict the average amount of daily spending. It has been
observed through daily spending pattern that salaried employee spends the money in the
starting of the month. This parameter is also used to monitor that on which days of the
month, customer spends more money than the normal transaction. The authors (Panigrahi
et al., 2009) used this parameter on the basis of daily/monthly spending amount and
implemented in Rule based filter to find out deviation in spending pattern of the customer.
This parameter is also used (Bolton and Hand, 2001) for break point analysis to compare the
current and previous spending behavior and identify the increasing level of spending
amount of customer.

3.8 Authentication type


This parameter is used to implement the security level (high, medium, low) for screening
during online transaction. If the security level is high then it screens all the parameters to
monitor and if the security level is low then it screens few parameters to monitor. The author
(Panigrahi et al., 2009) used this parameter as VIP or normal category to identify the high
volume of transaction and compare it with previous spending pattern. The authors
(Ramakalyani and Umadevi, 2012) used this parameter to find fraudulent transaction. A
value is calculated on the basis of this parameter and observed that if this value is below
than the predefined critical value then current transaction is considered as non-suspicious
transaction otherwise suspicious transaction.

3.9 Age of card


This parameter is used to give the details of relationship between card acquiring date and
the current date (Oracle Financial Services Software, 2014). Authors (Ramakalyani and
Umadevi, 2012) used this parameter to find fraudulent transaction. On the basis of age of
card, current transaction is categorized such as one year – high risk, two-three years –
medium risk and above three years – low risk. It has been observed that most of the fraud
occurs in case of age of card is less than one year. Therefore, during online transaction,
screening of all account is high with age of card is less than one year.
3.10 Time of card used Automated
This parameter is used to record the date and time of current transaction. This parameter fraud detection
identifies the pattern of transaction time of customer. This parameter is used to provide the
details of current online transaction time and compare the time with respect to location
techniques
where it is used. This study (Bahnsen et al., 2015) used this parameter to monitor the current
transaction time with the previous pattern of transaction time and identify the similarity. If
any deviation in the pattern occurs then current transaction is considered as suspicious
transaction. Author used this parameter in Renuga et al. (2014) to train the model for
709
identification of fraudulent transaction. This parameter (Bhusari and Patil, 2011) used to
monitor the variation in transaction time. If any variation found then transaction is
considered as suspicious transaction.

3.11 Customer income


This parameter is used to give the details of customer income. The customer never spends
money greater than his/her income. If transaction amount of transaction done by customer is
not suited as per his/her income then current transaction can be considered as suspicious
transaction. The authors (Ramakalyani and Umadevi, 2012) used this parameter to find
fraudulent transaction. As customer spends money as per his/her monthly income. The
customer never spends money more than his/her income because if any variation occurs then
current transaction is considered as suspicious transaction.

3.12 Customer position


This parameter is used to give the details of customer position. On the basis of high or low
position, transaction can be considered as low risk or high risk (Oracle Financial Services
Software, 2014). The authors (Ramakalyani and Umadevi, 2012) used this parameter to find
fraudulent transaction. Customer buys the product or spends the money on the basis of
position or status. Customer never spends money above the status. If any deviation occurs
then current transaction is considered as suspicious transaction.

3.13 Customer profession


This parameter is used to give the details of customer profession. The authors (Ramakalyani
and Umadevi, 2012) used this parameter to find fraudulent transaction. Customer spends his/
her money according to his/her profession. If customer belongs to high status profession then
spending pattern will be high. Hence, as per customer’s profession and spending pattern,
current transaction is considered as non-suspicious transaction or suspicious transaction.

3. 14 Customer age
This parameter is used to compare type of product purchasing with respect to the
customer’s age. This parameter is used (Sánchez et al., 2009) and reported that most cases of
fraud were found from 18 to 30 age group of people. This parameter is also used (Khan et al.,
2014a) to train the neural network to identify the transaction as suspicious transaction
according to product purchased. For example, one customer was buying a product which
was not suited the customer’s age. After enquiry, it was found that card was stolen of the
customer and used by another person.

3.15 Watch list scan


When customer wants to buy using credit card/debit card and during transaction it checks a
list of those customer who were involved in suspicious activity during previous transaction,
JFC as a result the current transaction can be considered as a suspicious or fare transaction. This
25,3 parameter is used to verify that whether the account has doubt full entities in checking the
record during the previous transaction or not. In the report (Oracle Financial Services
Software, 2014), this parameter include account id of those customer who are in the watch
list to monitor during transaction.

710 3.16 Amount limit


This parameter is used to restrict the maximum online transaction amount. Fraudster does
not know that how much restriction is implemented on the account transaction so that if the
customer attempts to purchase the product for higher amount then the current transaction
can be considered as fraudulent transaction. Author implemented this parameter as a filter
in advance fraud detection suite (AFDS – which is specifically built to provide for merchants
with tools to better combat online-fraud) (Authorize.Net, 2012) to set the lower and upper
limit of the amount of transaction. If any deviation is found then the current transaction can
be considered as a fraudulent transaction.

3.17 Transaction limit hourly


This parameter is used to restrict the maximum number of online transactions within an
hour. If any fraudster uses the stolen card then fraudster tries to spend all money as a
consequences within hour fraudster use multiple transaction and does not know the limit
hour restriction. As a result, current transaction is considered as a fraudulent transaction.
The author implemented this parameter as a filter in AFDS system (Authorize.Net, 2012) to
restrict the number of transaction per hour to prevent the fraudulent transaction.

3.18 Transaction IP velocity


This parameter is used to verify that whether excessive transaction occurs from same IP
address or does not occur. The fraudsters do not know the screening of the transaction. If
fraudster uses multiple cards for purchasing from single computer than it can be consider as
fraudulent transaction. The author stated (VISA, 2011) that this parameter is used to
monitor the multiple transactions from single IP address and consider as a suspicious
transaction. The author implemented this parameter as a filter in AFDS system (Authorize.
Net, 2012) to monitor the multiple transactions from single IP address.

3.19 Block IP fraud


This parameter is used to verify that whether the IP address which is used for online
transaction is marked as fraud IP or not. The author quoted in Authorize.Net, (2012) that
current transaction can be considered as a fraudulent transaction if it comes from specific IP
address and this specific IP address is considered as fraud IP address.

3.20 Positive status


This parameter is used to check that whether the account is in positive list or not. Financial
institution always monitors all the customers which are related to them. If customer
performs fine all the transactions from start date to till date. Then account is marked as
positive which helps in to identify the current transaction is suspicious or not. This list is
helpful to compare the current transaction with past fraudulent history of account so that
further fraud activity could be avoided for this account (Bhatla et al., 2003).
3.21 Foreign order Automated
This parameter is used to verify that online transaction is from outside country or not. This fraud detection
parameter is used (Carminati et al., 2014) to identify that transactions are to be executed
from outside the country that can be classify as suspicious transaction in training the model.
techniques

3.22 Citizenship
This parameter is used to record the citizenship of the customer who is going to purchase 711
the product and compare it with the list of countries where suspicious transactions have
been occurred in past. The authors (Bahnsen et al., 2015) used this parameter to compare
identify of the country with the list of suspicious country. If found then current transaction
is considered as a suspicious transaction. In Oracle Financial Services Software (2014), this
parameter contains a risk value for each country and can be considered as high or low risk
during transaction.

3.23 Distance IP billing


This parameter is used to check the distance between IP address which is being used to
online transaction and billing address of the customer (Singh and Singh, 2015). If customer
want to purchase some product from certain location but services of financial institution
find the distance between customer current location of IP address and billing address and
this distance is large than the specified range than the current transaction is considered as
suspicious transaction. To insure the security following points are also considered:
 Whether nation of IP address matches billing address country.
 If the country code is related to the IP address, IP address’s estimated state/region,
estimated city of the IP address, estimated latitude of the IP address, Estimated
Longitude of the IP address.
 IP address is behind an anonymous proxy (anonymous proxy is regarded as high
risk).
 If the IP address is an open proxy (transparent).

3.24 Transaction time gap


This parameter is used to record the difference of time between two consecutive transactions.
The authors (Panigrahi et al., 2009) included this parameter to identify the pattern of
transaction time gap and analyze the current transaction as per previous pattern stored. The
authors (Sánchez et al., 2009) involved this parameter to identify that product purchase from
different store in a short period of time which is not feasible, and then, the current transaction
is considered as fraudulent transaction. The authors (Carminati et al., 2014) used this
parameter to find the average time gap between subsequent transactions and trained the
model accordingly.

3.25 Store where purchase was made


This parameter is used to record store identity from where the purchase is done and
compare it with the list of stores where suspicious transactions have occurred in the past.
The authors (Sánchez et al., 2009) used this parameter to compare identify of the store where
purchase is to be made with the list of suspicious store. If found then current transaction is
considered as a suspicious transaction. The author reported in the paper, Gas station is the
store where high frauds have been done as a fraudulent transaction. The author used (VISA,
JFC 2011) this parameter as alert indicator because fraudster always looking for new store to
25,3 commit the fraud.

3.26 Transaction amount


This parameter is used to record the amount of each transaction. The authors (Panigrahi et al.,
2009) implemented this parameter to compare the previously recorded transaction amount
712 pattern in a cluster with the current transaction pattern and categorize as fraudulent or non-
fraudulent transaction. The authors (Bolton and Hand, 2001) used this parameter to consider a
transaction as fraudulent transaction in case of sudden increase in current transaction in
comparison with previously recorded transaction amount pattern. This parameter is also used
(Khan et al., 2014a) to train the neural network to classify the current transaction as fraudulent
or suspicious transaction. This parameter is analyzed (Correia et al., 2015) to consider the
current transaction as a suspicious transaction in case of sudden increase of transaction in
consecutive transaction. The authors (Carminati et al., 2014)included this parameter to train the
model for identification of transaction as suspicious category. The author used this parameter
(Renuga et al., 2014) to train the model for identification of fraudulent transaction.

3.27 Gender
This parameter is used to record the category of human (male or female) and find the pattern
of fraudulent transaction committed by male or female. The authors (Sánchez et al., 2009)
reported, high fraudulent transactions were made by female client. This parameter is also
used (Khan et al., 2014a) to train the neural network to identify the transaction as suspicious
transaction.

3.28 Type of product


This parameter is used to record the type of product which is being purchased by customer
and classify a pattern accordingly. The authors (Panigrahi et al., 2009) used this parameter
to identify the transaction is suspicious or non-suspicious on the basis purchasing the same
product or not. The authors (Sánchez et al., 2009) implemented this parameter to identify the
mismatch between the type of product purchasing and customer’s profile and consider the
transaction as fraudulent transaction.

3.29 Transaction type


This parameter is used by the author to identify that transaction is to be made either from
internet, ATM or POS (Bahnsen et al., 2015). On the basis of different level of risk of
transaction type, current transaction can be categories as suspicious transaction.

3.30 Account status


This parameter is used to find the status of account that whether the account was involved
in fraudulent/suspicious transaction or not from previously recorded status. The authors
(Khan et al., 2014a) included this parameter to train the neural network for identification of
account status for current transaction.
The above table (Table II) shows a clear pattern of focused parameters at different point
of time. In 2009 and before that, the parameters mostly considered are spending amount,
authentication type and transaction amount. After the advent of new security techniques,
the focused changed to those techniques and also on geo-location and IP techniques. The
abovementioned parameter not just checked the static information but also used to monitor
the transaction behavior of the customer. The transaction behavior is very useful in
Year of publications
Sr No. Parameters 2015 2014 2013 2012 Others

1. Address Singh and Singh (2015), Authorize.Net (2012) VISA (2011),


Verification System CyberSource (2015) Bhusari and Patil (2011) ,
(AVS) Panigrahi et al. (2009),
Bhatla et al. (2003)
2. Card Verification Correia et al. (2015), Authorize.Net (2012) VISA (2011),
Value (CVV2) CyberSource (2015) Panigrahi et al. (2009),
Bhusari and Patil (2011),
Bhatla et al. (2003)
3. Negative List CyberSource (2015) Oracle Financial Services Bhatla et al. (2003)
Software (2014)
4. Card Frequency Correia et al. (2015) Carminati et al. (2014) Patel and Singh (2013) , Ramakalyani and Umadevi Panigrahi et al. (2009),
Duman and Ozcelik (2011) (2012) Bolton and Hand (2001),
VISA (2011)
5. Location Bahnsen et al. (2015), Oracle Financial Services Patel and Singh (2013) Ramakalyani and Umadevi Sánchez et al. (2009), VISA
Singh and Singh (2015), Software (2014) (2012), Authorize.Net (2012) (2011)
CyberSource (2015)
6. Overdraft Patel and Singh (2013) Ramakalyani and Umadevi
(2012)
7. Daily Spending Patel and Singh (2013) Ramakalyani and Umadevi Panigrahi et al. (2009) ,
Amount (2012) Bolton and Hand (2001)
8. Authentication Ramakalyani and Umadevi Panigrahi et al. (2009), VISA
Type (2012) (2011)
9. Age of Card Oracle Financial Services Ramakalyani and Umadevi
Software (2014) (2012)
10. Time of card used Bahnsen et al. (2015) Patel and Singh (2013) , Ramakalyani and Umadevi Bhusari and Patil (2011),
Renuga et al. (2014) (2012) VISA (2011)
11. Customer Income Ramakalyani and Umadevi
(2012)
12. Customer Position Ramakalyani and Umadevi
(2012)
13. Customer Khan et al. (2014a), Ramakalyani and Umadevi
profession Oracle Financial Services (2012)
Software (2014)
14. Customer Age 4 Khan et al. (2014a) Sánchez et al. (2009)
15. Watch List Scan Oracle Financial Services
Software (2014)
16. Amount Limit Authorize.Net (2012)
17. Transaction Limit Authorize.Net (2012)
Hourly
(continued)

detecting online
transaction fraud
Parameters used in
Summary of
fraud detection

713
techniques
Automated

Table II.
JFC
25,3

714

Table II.
Year of publications
Sr No. Parameters 2015 2014 2013 2012 Others

18. Transaction IP Authorize.Net (2012) VISA (2011)


Velocity
19. Block IP Fraud Authorize.Net (2012)
20. Positive status CyberSource (2015)
21. Foreign Order Carminati et al. (2014)
22. Citizenship Bahnsen et al. (2015) Oracle Financial Services
Software (2014)
23. Distance IP Billing Singh and Singh (2015)
24. Transaction Time Carminati et al. (2014) Panigrahi et al. (2009),
Gap Duman and Elikucuk
(2013), Duman and
Elikucuk (2013), Philip and
Sherly (2012; Sánchez et al.,
2009)
25. Store where Sánchez et al. (2009),
purchase was made VISA (2011)
26. Transaction Bahnsen et al. (2015), Khan et al. (2014a), Renuga et al. (2014) Ramakalyani and Umadevi Panigrahi et al. (2009),
Amount Singh and Singh (2015), Carminati et al. (2014) (2012) Duman and Elikucuk
Correia et al. (2015), (2013), Duman and
CyberSource (2015) Elikucuk (2013), Philip and
Sherly (2012), Sánchez et al.,
(2009),
Bolton and Hand (2001)
27. Gender Khan et al. (2014a) Sánchez et al. (2009)
28. Type of Product Panigrahi et al. (2009),
Duman and Elikucuk
(2013), Duman and
Elikucuk (2013), Philip and
Sherly (2012), Sánchez et al.
(2009),
Bhusari and Patil (2011)
29. Type of Bahnsen et al., (2015)
Transaction
30. Account Status Khan et al. (2014a)
preventing the fraud detection. In present scenario, there is new concept, i.e. clean fraud. It Automated
means that it is very difficult to differentiate between valid and invalid transaction. In other fraud detection
words, more sophisticated way to implement fraudulent transaction. This type of fraud can
only be detected by analyzing the transaction behavior of customer.
techniques
The above graph (Figure 1) is showing the frequency per centage of parameters used in
different fraud detection techniques. Plethora of literature is studied by author and various
parameters used in different fraud detection technique are found. On the basis of frequency
of parameters, the author draws the statistics in the form of graph. From the graph, it is 715
observed that parameters like address verification system (AVS), card verification value
(CVV2), card frequency, location, time of card used and transaction amount are used most
prominently.
The annual fraud benchmark report was published in 2016 (CyberSource, 2015) by cyber
source (Payment Management of Digital Economy). In this report, statistics (Figure 2) are
shown that various automated screening fraud detection tools are used and these tools have
various parameters use in screening during the transaction. This graph shows that
parameters like address verification service, card verification number, customer order
history and negative list are used most prominently among various parameters. An
automated screening tool can work efficiently if it uses most suitable parameters.
If we compare both the graphs (Figures 1 and 2) it is observed that few parameters (like
address verification service, card verification number, location etc.) having high frequency
per centage are common in both. It is clear that the report provided by CyberSource (2015)
support the direction of author for this research. In other words, the author is motivated to

Figure 1.
Frequency of
parameters used in
online transaction
fraud detection
techniques
JFC
25,3

716

Figure 2.
Frequency of
parameters used in
online transaction
fraud detection
techniques
(Cybersource, 2015)

achieve the target for developing the improved fraud detection system for online
transaction.

4. Conclusion
As time grows, online business in all areas is also growing, and there is no restriction or
reducing gap between customer and financial business for online transaction. In the same
manner, online fraud transaction is also increasing which should be prevented or reduced.
Therefore, an improved more sophisticated system for detecting and preventing the online
fraud transaction require which is based on various parameters used in detection
techniques. Fraud detection not only depends on the static fact but it also considers the
behavioral pattern of the customer of past transaction which is helpful to find suspicious
transaction.

5. Future scope
Credit card fraud has exponentially increased in the recent years. Apart from the excessive
use of credit card, one main reason is increase in the online use of credit cards. To reduce the
risk of merchants’ and credit card companies, managing the process in an automatic and
efficient way and building an accurate but easy to manage credit card risk monitoring
system is the key. In this paper, discussion on different parameters and different types of
fraud detection techniques has been done. In future, by using this information, candidate
will develop an improved online transaction fraud detection system. This system will Automated
consider a good trade-off between finding the fraudulent cases and time consuming for fraud detection
identification of fraudulent cases. By using these parameters, researchers may involve a
broad view of implementation and do various more sophisticated implementation of fraud
techniques
detection system towards more intelligence and automatic decision oriented.

References 717
Agarwal, C.C. and Reddy, C.K. (Eds) (2013), Data Clustering: Algorithms and Applications, Chapman
and Hall/CRC, New York, NY.
Agrawal, A., Kumar, S. and Mishra, A.K. (2015), “Implementation of novel approach for credit card
fraud detection”, 2nd International Conference on Computing for Sustainable Global
Development (INDIACom), IEEE, pp. 1-4.
Aleskerov, E., Freisleben, B. and Rao, B. (1997), “Cardwatch: a neural network-based database mining
system for credit card fraud detection”, Proceedings of the IEEE/IAFE Computational
Intelligence for Financial Engineering (CIFEr), pp. 220-226.
Assis, C.A., Pereira, A.C., Pereira, M.A. and Carrano, E.G. (2014), “A genetic programming approach for
fraud detection in electronic transactions”, IEEE Symposium on IEEE Computational Intelligence
in Cyber Security (CICS), pp. 1-8.
Authorize.Net (2012), “Fighting online fraud”, available at: www.authorize.net/content/dam/authorize/
documents/FDS_0504.pdf (accessed 15 November 2016).
Bahnsen, A.C., Aouada, D., Stojanovic, A. and Ottersten, B. (2015), “Detecting credit card fraud using
periodic features”, IEEE 14th International Conference on IEEE Machine Learning and
Applications (ICMLA), pp. 208-213.
Bai, F. and Chen, X. (2013), “Analysis on the new types and countermeasures of credit card fraud in
mainland China”, Journal of Financial Crime, Vol. 20 No. 3, pp. 267-271.
Barker, K.J., D’Amato, J. and Sheridon, P. (2008), “Credit card fraud: awareness and prevention”, Journal
of Financial Crime, Vol. 15 No. 4, pp. 398-410.
Behera, T.K. and Panigrahi, S. (2015), “Credit card fraud detection: a hybrid approach using fuzzy
clustering and neural network”, Second International Conference on IEEE Advances in
Computing and Communication Engineering (ICACCE), pp. 494-499.
Bekirev, A.S., Klimov, V.V., Kuzin, M.V. and Shchukin, B.A. (2015), “Payment card fraud detection
using neural network committee and clustering”, Optical Memory and Neural Networks, Vol. 24
No. 3, pp. 193-200.
Bentley, P.J., Kim, J., Jung, G.H. and Choi, J.U. (2000), “Fuzzy Darwinian detection of credit card fraud”,
The 14th Annual Fall Symposium of the Korean Information Processing Society, 14 October.
Bhatla, T.P., Prabhu, V. and Dua, A. (2003), “Understanding credit card frauds”, Cards Business Review,
Vol. 1 No. 6.
Bhusari, V. and Patil, S. (2011), “Study of hidden Markov model in credit card fraudulent detection”,
International Journal of Computer Applications, Vol. 20 No. 5, pp. 33-36.
Bianchini, M., Maggini, M. and Scarselli, F. (Eds) (2009), Innovations in Neural Information Paradigms
and Applications, Vol. 247, Springer.
Bolton, R.J. and Hand, D.J. (2001), “Unsupervised profiling methods for fraud detection”, Credit Scoring
and Credit Control VII, pp. 235-255.
Braspenning, P.J., Thuijsman, F. and Weijters, A.J.M.M. (1995), Artificial Neural Networks: An
Introduction to ANN Theory and Practice, Springer Science and Business Media, Berlin, Vol. 931.
Brause, R.W., Langsdorf, T.S. and Hepp, H.M. (1999), “Credit card fraud detection by adaptive neural
data mining”, Proceedings of the 11th IEEE International Conference on Tools with Artificial
Intelligence, pp. 103-106.
JFC Cao, B.Y. (Ed.) (2007), Fuzzy Information and Engineering, Springer Science and Business Media,
Berlin.
25,3
Carminati, M., Caron, R., Maggi, F., Epifani, I. and Zanero, S. (2014), “BankSealer: an online banking
fraud analysis and decision support system”, IFIP International Information Security
Conference, Springer Berlin Heidelberg, pp. 380-394.
Carneiro, E.M., Dias, L.A.V., da Cunha, A.M. and Mialaret, L.F.S. (2015), “Cluster analysis and artificial
718 neural networks: a case study in credit card fraud detection”, 12th International Conference on
IEEE Information Technology-New Generations (ITNG), pp. 122-126.
Chen, S.H. (Ed.) (2012), Genetic Algorithms and Genetic Programming in Computational Finance,
Springer Science and Business Media, New York, NY.
Chen, J., Tao, Y., Wang, H. and Chen, T. (2015), “Big data-based fraud risk management at Alibaba”,
The Journal of Finance and Data Science, Vol. 1 No. 1, pp. 1-10.
Correia, I., Fournier, F. and Skarbovsky, I. (2015), “The uncertain case of credit card fraud detection”,
Proceedings of the 9th ACM International Conference on Distributed Event-Based Systems,
pp. 181-192.
CyberSource (2015), “Annual fraud benchmark report: a balancing act”, available at:www.cybersource.
com/content/dam/cybersource/NA_2016_Fraud_Benchmark_Report.pdf (accessed 15 January
2017).
Das, S. (2013), Computational Business Analytics, CRC Press, New York, NY.
Delamaire, L., Abdou, H.A.H. and Pointon, J. (2009), “Credit card fraud and detection techniques: a
review”, Banks and Bank Systems, Vol. 4 No. 2, pp. 57-68.
Dhok, S.S. and Bamnote, G.R. (2012), “Credit card fraud detection using hidden Markov model”,
International Journal of Advanced Research in Computer Science, Vol. 3 No. 3.
Duman, E. and Elikucuk, I. (2013), “Applying migrating birds optimization to credit card fraud
detection”, Pacific-Asia Conference on Knowledge Discovery and Data Mining, Springer Berlin
Heidelberg, pp. 416-427.
Duman, E. and Elikucuk, I. (2013), “Solving credit card fraud detection problem by the new
metaheuristics migrating birds optimization”, International Work-Conference on Artificial
Neural Networks, Springer Berlin Heidelberg, pp. 62-71.
Duman, E. and Ozcelik, M.H. (2011), “Detecting credit card fraud by genetic algorithm and scatter
search”, Expert Systems with Applications, Vol. 38 No. 10, pp. 13057-13063.
Everitt, B.S., Landau, S., Leese, M. and Stahl, D. (2011), Cluster Analysis, John Wiley and Sons. Ltd,
New York, NY.
Financial Fraud Action-UK (2017), “fraud the facts 2016”, available at: www.financialfraudaction.org.
uk/wp-content/uploads/2016/07/Fraud-the-Facts-A5-final.pdf (accessed 5 February 2017).
Ganji, V.R. and Mannem, S.N.P. (2012), “Credit card fraud detection using anti-k nearest neighbor
algorithm”, International Journal on Computer Science and Engineering, Vol. 4 No. 6, p. 1035.
Gen, M. and Cheng, R. (1997), Genetic Algorithms and Engineering Design, John Wily and Sons,
New York, NY.
Goldberg, D.E. (2006), Genetic Algorithms, Pearson Education.
Graupe, D. (2013), Principles of Artificial Neural Networks, World Scientific, Vol. 7.
Halvaiee, N.S. and Akbari, M.K. (2014), “A novel model for credit card fraud detection using artificial
immune systems”, Applied Soft Computing, Vol. 24, pp. 40-49.
Hassoun, M.H. (1995), Fundamentals of Artificial Neural Networks, MIT press, MA.
Khan, A.U.S., Akhtar, N. and Qureshi, M.N. (2014a), “Real-time credit-card fraud detection using artificial
neural network tuned by simulated annealing algorithm”, Proceedings of International Conference
on Recent Trends in Information, Telecommunication and Computing, ITC, pp. 113-121.
Khan, M.Z., Pathan, J.D. and Ahmed, A.H.E. (2014b), “Credit card fraud detection system using hidden Automated
Markov model and K-clustering”, International Journal of Advanced Research in Computer and
Communication Engineering, Vol. 3 No. 2, pp. 5458-5461.
fraud detection
Li, J., Cao, L., Lim, E.P., Zhou, Z.H., Ho, T.B., Cheung, D. and Motoda, H. (2013), Trends and Applications
techniques
in Knowledge Discovery and Data Mining, Springer Berlin Heidelberg.
Oracle Financial Services Software (2014), “Know your customer risk assessment guide”, available at:
https://docs.oracle.com/cd/E41197_01/books/ECM%206.2.2/kyc_rag_2.0_4ed.pdf (accessed 1
January 2017). 719
Özçelik, M.H., Duman, E., Is ik, M. and Çevik, T. (2010), “Improving a credit card fraud detection system
using genetic algorithm”, International Conference on IEEE Networking and Information
Technology (ICNIT), pp. 436-440.
Panigrahi, S., Kundu, A., Sural, S. and Majumdar, A.K. (2009), “Credit card fraud detection: a fusion
approach using Dempster–Shafer theory and Bayesian learning”, Information Fusion, Vol. 10
No. 4, pp. 354-363.
Papadopoulos, A. and Brooks, G. (2011), “The investigation of credit card fraud in Cyprus: reviewing
police effectiveness”, Journal of Financial Crime, Vol. 18 No. 3, pp. 222-234.
Patel, R.D. and Singh, D.K. (2013), “Credit card fraud detection and prevention of fraud
using genetic algorithm”, International Journal of Soft Computing and Engineering, Vol. 2
No. 6.
Philip, N. and Sherly, K.K. (2012), “Credit card fraud detection based on behavior mining”, TIST
International Journal for Science, Technology and Research, Vol. 1, pp. 7-12.
Prakash, A. and Chandrasekar, C. (2012), “A novel hidden Markov model for credit card fraud
detection”, International Journal of Computer Applications, Vol. 59 No. 3.
Prakash, A. and Chandrasekar, C. (2015), “An optimized multiple semi-hidden Markov model for
credit card fraud detection”, Indian Journal of Science and Technology, Vol. 8 No. 2,
pp. 165-171.
Quah, J.T. and Sriganesh, M. (2008), “Real-time credit card fraud detection using computational
intelligence”, Expert Systems with Applications, Vol. 35 No. 4, pp. 1721-1732.
Ramakalyani, K. and Umadevi, D. (2012), “Fraud detection of credit card payment system by
genetic algorithm”, International Journal of Scientific and Engineering Research, Vol. 3
No. 7, pp. 1-6.
Renuga, T.D., Rabiyathul, A.B. and Kamaladevi, M. (2014), “Fraud detection in card not present
transactions based on behavioral pattern”, Journal of Theoretical and Applied Information
Technology, Vol. 61 No. 3.
Sadaaki, M., Hidetomo, I. and Katsuhiro, H. (2008), Algorithms for Fuzzy Clustering: Methods in
c-Means Clustering with Applications, German, Springer.
Sánchez, D., Vila, M.A., Cerda, L. and Serrano, J.M. (2009), “Association rules applied to credit card
fraud detection”, Expert Systems with Applications, Vol. 36 No. 2, pp. 3630-3640.
Singh, P. and Singh, M. (2015), “Fraud detection by monitoring customer behavior and activities”,
International Journal of Computer Applications, Vol. 111 No. 11.
Soltani, N., Akbari, M.K. and Javan, M.S. (2012), “A new user-based model for credit card fraud
detection based on artificial immune system”, 16th CSI International Symposium on IEEE
Artificial Intelligence and Signal Processing (AISP), pp. 029-033.
Van Laarhoven, P.J. and Aarts, E.H. (1987), “Simulated annealing”, Simulated Annealing: Theory and
Applications, Springer Netherlands, pp. 7-15.
Van Vlasselaer, V., Bravo, C., Caelen, O., Eliassi-Rad, T., Akoglu, L., Snoeck, M. and Baesens, B. (2015),
“APATE: a novel approach for automated credit card transaction fraud detection using
network-based extensions”, Decision Support Systems, Vol. 75, pp. 38-48.
JFC VISA (2011), “Global visa card-not present merchant guide to greater fraud control”, available at: www.
fnbtanzania.co.tz/downloads/merchant-services/Global_Visa_Card-Not-Present.pdf (accessed 10
25,3 November 2016).
Zaki, M.J., MeiraJr, W. and Meira, W. (2014), Data Mining and Analysis: Fundamental Concepts and
Algorithms, Cambridge University Press, MA.

Further reading
720
Arthur Williams, D. (2007), “Credit card fraud in Trinidad And Tobago”, Journal of Financial Crime,
Vol. 14 No. 3, pp. 340-359.
Balan, L. and Popescu, M. (2011), “‘Credit card fraud’, the annals of the ‘Stefan cel Mare’ University of
Suceava”, Fascicle of The Faculty of Economics and Public Administration, Vol. 11 No. 1, pp. 81-85.
Kohonen, T. (1998), “The self-organizing map”, Neurocomputing, Vol. 21 Nos 1/3, pp. 1-6.
Mena, J. (2003), Investigative Data Mining for Security and Criminal Detection, Butterworth-Heinemann.
Minelli, M., Chambers, M. and Dhiraj, A. (2012), Big Data, Big Analytics: Emerging Business Intelligence
and Analytic Trends for Today’s Businesses, John Wiley and Sons, NJ.
Mo, H. (Ed.) (2009), Handbook of Research on Artificial Immune Systems and Natural Computing:
Applying Complex Adaptive Technologies: Applying Complex Adaptive Technologies, IGI Global.
Zhang, C. and Zhang, S. (2002), Association Rule Mining: Models and Algorithms, Springer-Verlag,
Berlin.

Corresponding author
Vipin Khattri can be contacted at: vipinkhattri@gmail.com

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com

You might also like