Research On Anti-fraud Of Credit Card Online Payment

Credit card frauds disrupted the normal financial order, result in great loss for banks, merchants and cardholders.For the merchants, losses caused by three major areas:refund for the frauds, credit losses and additional bank charges. In recent years, the rate of online trading losses due to fraud stable at 1.4%, but because of the increasing amount of online transactions, losses due to fraud is still rising. Therefore, the paper's work has important theoretical and practical value.Anti-fraud of credit card online payment is defined as using some method to assess the risk of transaction when it is happen to help merchants to make decisions whether to accept this transaction, in order to achieve the purpose of reducing fraudulent transactions. Overseas, credit card developed well, so there have a lot of researches on anti-fraud. In the system services and individual services, many companies and organizations have provided anti-fraud services for credit card online payment, such as Cybersource anti-fraud system, AVS address detection services and so on.In China, research in this area was relatively small. Theoretical research, Yan Hua, Hu Mengliang, who use Bayesian classification algorithm historical data on credit card anti-fraud, Tong Fengru study combined classifier based on credit card fraud recognition and so on; enterprise applications, the IPS company has released credit card anti-fraud system called ANT which uses a neural network model for anti-fraud fraud analysis.This paper presents an outlier mining algorithm based on dissimilarity sum, and using the association rule mining, mines the characteristics of the historical fraudulent transactions, then designes and implement a prototype anti-fraud system to provide anti-fraud Protection for online credit card transactions. And here follows the main work of the paper:(1) Research the status of credit card online payment fraud;(2) Through outlier mining algorithm based on dissimilarity sum and rule extraction techniques, Paper analysis and mine the historical transactions,get the characteristic information from the fraudulent transactions;(3) Paper based on the characteristic information from the fraudulent transactions, associated with existing anti-fraud services, design a anti-fraud system for credit card online trading and achieve a prototype anti-fraud system;(4) The paper designs an experiment to test the effectiveness and the real-timing of the system, and compare with the current anti-fraud systems.Compare the results of the experiment, prove the system which this paper design is adapts to credit card online payment system, and it is more effective than other anti-fraud methods. And the theoretical significance of this paper are:(1) Paper combine with anti-fraud data mining theory and the anti-fraud fact, presents an outlier mining algorithm based on dissimilarity sum.(2) Combine the theory of data mining, e-commerce, and the theory of criminal psychology, apply to credit card online payment anti-fraud. Practical value of the paper is reflected in:(1) To help reduce the probability of fraudulent transactions and reduce the loss of merchants.(2) Increase cardholders'confidence in electronic commerce by reducing fraudulent transactions.Strengthen the real-timing of the system further, and access more mature individual anti-fraud services to the system are the next work of this paper.