Profiling intelligent systems applications in fraud detection and prevention: survey of research articles
Profiling intelligent.pdf (Size: 486.71 KB / Downloads: 21)
This paper surveys intelligent systems (IS)
applications using a literature review and classification of
articles from 1956 to 2009 with a keyword index and article
abstract in order to explore how IS applications in the field of
fraud detection and prevention have developed during this
period. Based on the scope of 36 articles found from Web of
science (SSCI, SCI and A&HCI) database, this paper surveys
and classifies IS applications in the fraud detection and
prevention using the following three categories of intelligent
systems: neural networks, fuzzy ISs, and computational
intelligence. Following applications areas were detected and
described: telecommunications, insurance, auditing, medical
care, credit card transactions, e-business, bid pricing and
INTRODUCTION (HEADING 1)
According to the Random House Dictionary fraud is a
deceit, trickery, sharp practice, or breach of confidence,
perpetrated for profit or to gain some unfair or dishonest
advantage. For the purpose of this article the definition of
fraud proposed by  will be used: Fraud is the obtaining of
financial advantage or causing of loss by implicit or explicit
deception; it is the mechanism through which the fraudster
gains an unlawful advantage or causes unlawful loss.
Attempts to estimate size and damage from the fraud
indicates that fraud costs a very considerable sum of money.
For example, cost of fraud in UK may fall in rough order of
magnitude of ten+ billion pounds per year. However, there
are many weaknesses of the data collection process on crime
. One reason for incomplete data is that the fraud is not
always reported and in addition not always detected, because
those that are victim of fraud are often not even aware that
the fraud has been committed. Intense financial pressure
during the economic crisis has led to an increase of fraud,
according to a survey of fraud experts conducted by the .
APPLICATIONS OF IS IN FRAUD DETECTION AND
Telecommunications is the application area with the
highest number of articles published which reflects the fact
that many telecoms take fraud very seriously, and those who
did not do so saw their customers leave and their costs rise.
This is also due to the competition and regulation among
telecommunications market especially in European Union
that is getting more crowded each year.
Fraud detection and prevention is taken very seriously
among telecommunications companies, which resulted in
the 10 articles found on the applications of IS. Number of
articles range modestly from one article in average per year
until three articles were published in 2009. Neural networks
were used in most of the articles: Kohonen classifying
model, recurrent neural networks, radial basis function
(RBF) model, feed-forward (FF) neural networks, self-
organizing maps (SOMs) and bidirectional artificial neural
network (bi-ANN). Other IS mentioned are fuzzy rules,
artificial bio-computing systems, and adaptive neuro-fuzzy
inference system (ANFIS). Applications are focused either
to on-line and real-time security systems or to pre-billing
systems that classify unusual behavior in consumption, call
detail, and subscription information.
Fraud prevention and detection is a broad category of
research issues. Some specific methodologies and methods
are presented as examples for exploring the suggestions and
solutions to specific fraud detection and prevention problem
domains. Methodologies of fraud detection and prevetion
are attracting much attention and efforts, both academia and
Intelligent systems provide powerful and flexible means
for obtaining solutions to a variety of problems that often
cannot be dealt with by other methods.
Literature review on the usage of intelligent systems
usage in fraud detection and prevention has been conducted
based on the articles cited in Web of science database.
Survey on research articles indicated that IS applications in
fraud detection and prevention are diversified due to
problem domains, and that their applications are proving to
be critical in the process of fraud detection and prevention.