he as 1) Lower profile; 2) Middle profile;

e-commerce platform is a substantial marketplace with a group of persons such
as merchants, customers, banks, and other commercial societies. Hence, the
e-commerce system is a very vibrant and voluminous one. One of such system is
Credit card payment in online.









of credit cards for online purchases has significantly increased and it caused
an explosion in the credit card fraud. Credit card fraud includes dishonest use
of card or account information without the knowledge of the owner; hence it is
an act of criminal process. Currently, the major problem for e-commerce business is that
fraudulent transactions appear more and more similar to genuine ones 1. Generally,
in real case, the fraud datasets are extremely skewed and scattered. Hence,
statistical fraud detection or simple pattern matching techniques are not
efficient to detect fraud. Therefore, execution of effective
fraud detection system becomes crucial for all card issuing systems to avoid
their losses.

                Hidden Markov Model will be used as a key to find out
the fraudulent transaction by using the expenditure profiles of users. It works
on the user’s expenditure profiles which can be divided into major three types
such as 1) Lower profile; 2) Middle profile; and 3) Higher profile 7-8. For
every credit card, the expenditure profile is different, so it can figure out
an inconsistency of user profile and try to find fraudulent transaction. It
keeps record of expenditure profile of the card holder with the transactions
that are done in online. Thus analysis of card holder’s profile will be a
useful tool in fraud detection system and it is an assuring way to check
fraudulent transaction, although fraud detection system does not keep records
of number of purchased goods and categories 2,3. The set of information
contains expenditure profile of card holder, money spent in every transaction,
the last purchase time, category of purchase etc. The main challenge here is
how to improve detection accuracy, the computational capacity of the detection
system. This factor has become more and more important with the unpredictable
growth of trading data. In E-commerce, virtual cards are used more frequently
than physical cards, which make the payment much easier and create many small
but frequent transactions. The growing number of users and payment transactions
has brought heavy workloads to these systems. The speed of new transactions
coming into the system can reach millions per second while the size of stored
historical transactions can reach several PBs or even EBs. In this case,
processing of the detection task with the incoming transactions with a low
delay is very hard for most traditional systems.

                According to the recent trends, Big Data technology
seems to be the key of solving the challenge of computational capacity. Thus the
proposed fraud detection system has to be implemented in a Big Data tool for
analysing huge data.