• No results found

Protecting E-Commerce Systems From Online Fraud

N/A
N/A
Protected

Academic year: 2021

Share "Protecting E-Commerce Systems From Online Fraud"

Copied!
6
0
0

Loading.... (view fulltext now)

Full text

(1)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3549

Protecting E-Commerce Systems From

Online Fraud

First Author

Second Author

P.PhaniAlekhya

Student,

Department of Computer

Science and Engineering,

QIS College of Engineering

& Technology, ongole,

Andhra Pradesh

,

India

.

Sk.Mahaboob Basha

Assistant Professor,

Department of Computer

Science and Engineering,

QIS College of Engineering

&Technology, ongole,

Andhra Pradesh

,

India

.

Abstract- Due to the advent of Internet technologies, E-commerce widely adapted mode of business in modern times. With the growth of E-commence domain credit card usage has become a common phenomenon. This has given chance to adversaries to commit fraud. In the real world, there were plenty of instances of fraud cases. It has its impact on financial outfits that issue credit cards, the E-commerce business entities and also the customers of the E-commerce applications. To overcome this problem and to build confidence in the stakeholders of the E-commerce many techniques came into existence. As simple pattern matching methods are inadequate to solve the problem many modern techniques came into existence. They are based on Genetic programming, Sequence Alignment, Machine learning, Fuzzy logic, Data mining and Artificial intelligence. These techniques are capable of detecting fraudulent transactions. In this paper we explore various techniques being used. We also build a prototype application which demonstrates the efficiency of Genetic Programming to detect credit card fraud. The empirical results revealed that the proposed solution is effective

.

Keywords – Credit card fraud detection, fraud detection techniques, E-commerce

INTRODUCTION

Credit card fraud detection is an active and continuous research area as it involves monetary transactions. The research is important in the light of web technologies that support E-commerce applications in building and managing. Moreover businesses in the world are going online. The online businesses are able to reach global markets and they reach global customers. The E-commerce applications make it possible to shop items anytime and anywhere. There are no time and geographical restrictions. This has plenty of advantages for both consumers and businesses. There are many kinds in E-commerce. B2B (Business to Business), B2C (Business to Consumer) and C2C (Consumer to Consumer) are the three popular forms of commerce applications. The transactions in E-commerce might have fraudulent transactions along with genuine ones. It is essential to have a system that can detect fraudulent transactions. Various techniques such as Hidden Markov Model, Artificial Intelligence, Sequence Alignment, and Genetic Programming came into existence to combat this problem.

In this paper we implement Genetic Algorithm (GA) in order to detect credit card fraud. The GA has provision for evaluating populations continuously until a best fit is found. The operators like cross over and mutation help in

(2)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3550

accurate prediction of fraud cases. We also built a prototype application that demonstrates the efficiency of the proposed solution. The experimental results revealed that the application is very useful in real time systems. The remainder of the paper is organized as follows. Section II reviews relevant literature. Section III presents some of the fraud detection techniques. Section IV provides details about the proposed fraud detection technique and also the prototype details. Section V shows experimental results while section VI concludes the paper.

RELATED WORKS

Credit card fraud detection has become an inevitable part of E-commerce applications. As the applications involve monetary transactions, fraud detection techniques are indispensable. Lot of research has been made in this area [1], [2], and [3]. Many algorithms came into existence to detect credit card fraud [4], [2]. The algorithms include fuzzy logic [5], sequence alignment algorithm [6], [7], data mining techniques [8], [9], machine learning and artificial intelligence approaches [10], [11], [12]. There are other techniques such as Web Services – Based CCFD, CCFD with Artificial Immune System [13], [14], Cardwatch [15], Bayesian Belief Networks [4] , Intrusion Detection [13], Case Based Reasoning for CCFD [4], Advanced Fraud Detection [16], CCFD based on computational intelligence [13], CCFD using self-organizing maps [13]. Many are based on pattern matching, Meta learning and artificial intelligence. In this paper we compare some of the techniques which are good for CCFD.

COMPARISON OF CCFD SYSTEMS

In this section we compare various credit card fraud detection techniques such as Dempster–Shafer theory and Bayesian learning, Hidden Markov Model (HMM) and Genetic Programming (GP).

Dempster–Shafer theory and Bayesian learning

approach

The Dempster–Shafer theory and Bayesian learning approach is the combination of two approaches [17], [18], [19]. The evidences from past and correct are combined in order to detect fraud. Information fusion is the approach followed by this hybrid technique. Figure 1 shows the overview of this hybrid approach.

Fig. 1 – Hybrid approach for CCFD (excerpt from [16])

As seen in figure 1, this approach has four components. They are rule based filter, Bayesian learner, transaction history database and Dempster-Shafer adder. The evidences found from multiple components are fused and the detection is made. This approach is more accurate but consumes more resources besides being slow

.

CCFD using Hidden Markov Model (HMM)

This model is used to analyze credit card transactions. It needs training data and also test data in order to detect fraud. It uses K-means data mining algorithm internally. The K-means algorithm takes all credit card transactions

(3)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3551

and number of clusters as input and generates clusters that are used in HMM. All the transactions are divided into low, medium and high, the three clusters. Once the clusters are formed they are kept in a HMM. The HMM is used for every new transaction. The amount in new transaction should belong to either low, or medium or high. If not the transaction is suspected to be fraudulent and the corresponding people or organizations are alerted. The general overview of HMM is as shown below.

Fig. 2 – Overview of HMM Model (excerpt from [16])

CCFD using GP

Genetic programming is widely used for solving various problems. In this paper we implemented a Genetic Algorithm to detect credit card fraud. This algorithm makes use of existing transactions of credit cards. It uses multiple criteria to detect fraud. The criteria include credit card usage frequency, credit card usage location, overdraft on

the credit card, and credit card book balance. The overview of the GP approach is as shown in figure 3.

Fig. 3 – Overview of proposed architecture for GP

As can be seen in figure 3, the data of credit cards is taken from data warehouse. Then the data is subjected to rules engine. The rules engine contains fraud case rules. The filter and priority components take care of filtering and priority setting. The genetic algorithm is responsible to detect fraud.

PROTOTYPE IMPLEMENTATION

A prototype application has been built in Java platform. The application is developed with Graphical User Interface (GUI) to be user-friendly. The environment used to build the application includes a PC with 4GB RAM, Core 2 dual processor running Windows XP operating system. NetBeans is used as IDE. The important application screens are presented in figure 4 and 5.

(4)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3552

Fig. 4 – Dataset used for experiments

As seen in figure 4, the dataset contains credit card transaction details for number of instances. This data is used by GA proposed in this paper. The GA makes use of the components as described in figure 3 in order to detect fraudulent transactions. The results of detection are shown in figure 5.

Fig. 5 – Results of Fraud Detection

As can be seen in figure 5, based on various criteria mentioned earlier, the fraud detection is done. The criteria are used as part of genetic programming model proposed in

this paper. The results help in understanding the fraud transactions and they can be used to train the system further in order to make new rules and achieve higher accuracy of fraud detection.

EXPERIMENTAL RESUTLS

We compared the results of our approach with that of Artificial intelligence, Hidden Markov Model, Sequence Alignment, and Machine Learning. The comparison is made in terms of true positives and false positives. Highest true positives and least false positives are achieved by GP

.

Fig. 6 – Performance of CCFD techniques with respect to true positives

As can be seen in figure 6, the performance of Genetic Programming is 100%. It has achieved 100% true positives when compared with other techniques such as AI, HMM, SA and machine learning.

0

20

40

60

80

100

120

T

r

u

e

P

o

s

i

t

i

v

e

s

Comparison of CCFD Techniques

CCFD Techniques

(5)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3553

Fig. 7 – Performance of CCFD techniques with respect to false positives

As can be seen in figure 7, the performance of Genetic Programming is more. It has achieved very less percentage of false positives when compared with other techniques such as AI, HMM, SA and machine learning.

CONCLUSIONS AND FUTURE WORK

In this paper we studied the problem of credit card fraud in E-commerce applications. We explored various approaches to solve the problem. The knowledge of various approaches can improve the scope of protecting E-commerce applications. Finally we implemented genetic algorithm for credit card fraud detection. As the adversaries change their means of attack every time, it is important to have constant vigil on the methods they use and update the techniques accordingly. In this paper build a prototype application in Java platform in order to demonstrate the proof of concept. The application uses genetic algorithm to detect credit card fraud. Data mining and other techniques are available to solve this problem. However, we preferred GA as it is

efficient in detecting credit card fraud. The experimental results reveal that the proposed application is useful and can be used in real world systems.

REFERENCES

[1] Tej Paul Bhatla, Vikram Prabhu & Amit Dua “Understanding Credit Card Frauds,” 2003.

[2] Linda Delamaire, Hussein Abdou, John Pointon, “Credit card fraud and detection techniques: a review,” Banks and Bank Systems, pp. 57-68, 2009.

[3] Barry Masuda, “Credit Card Fraud Prevention: A Successful Retail Strategy,” crime prevention, Vol. 6, 1986.

[4] Ezawa.K. & Norton.S,”Constructing Bayesian Networks to Predict Uncollectible Telecommunications Accounts,” IEEE Expert, October;45-51, 1996.

[5] Peter J. Bentley, Jungwon Kim, Gil-Ho Jung and Jong-Uk Choi, “Fuzzy Darwinian Detection of Credit Card Fraud,” In the 14th Annual Fall Symposium of the Korean Information Processing Society, 14th October 2000.

[6] Amlan Kundu, S. Sural, A.K. Majumdar, “Two-Stage Credit Card Fraud Detection Using Sequence Alignment,” Lecture Notes in Computer Science, Springer Verlag, Proceedings of the InternationalConference on Information Systems Security, Vol. 4332/2006, pp.260-275, 2006.

[7] Amlan Kundu, Suvasini Panigrahi, Shamik Sural and Arun K.Majumdar, “BLAST-SSAHA Hybridization for Credit Card Fraud Detection,” IEEE Transactions On Dependable And Secure Computing, vol. 6, Issue no. 4, pp.309-315, October-December 2009.

[8] Philip K. Chan ,Wei Fan, Andreas L. Prodromidis, Salvatore J. Stolfo, “Distributed Data Mining in Credit Card Fraud Detection,” IEEEIntelligent Systems ISSN, Vol. 14 , Issue No. 6, Pages: 67 – 74, November 1999.

[9] C. Phua, V. Lee, K. Smith, R. Gayler, “A Comprehensive Survey of Data Mining-based Fraud Detection Research,” Artificial Intelligence Review, 2005.

[10] Ray-I Chang, Liang-Bin Lai, Wen-De Su, Jen-Chieh Wang, Jen-Shiang Kouh, “Intrusion Detection by Backpropagation Neural Networks with Sample-Query and Attribute-Query,” Research IndiaPublications, pp.6-10, November 26, 2006.

[11] Ghosh, D.L. Reilly, “Credit Card Fraud Detection with a Neural-Network,” Proceedings of the International Conference on SystemScience, pp.621-630, 1994.

0

5

10

15

20

25

M

ac

h

in

e

Le

ar

n

in

g

S

eq

u

en

ce

A

lig

n

m

en

t

H

M

M

A

I

G

P

F

a

l

s

e

P

o

s

i

t

i

v

e

s

Comparison of CCFD Techniques

CCFD

Techniques

(6)

ISSN: 2231-2803

http://www.ijcttjournal.org

Page3554

[12] R. Brause, T. Langsdorf, M. Hepp, “Neural Data Mining for Credit Card Fraud Detection, “International Conference on Tools with Artificial Intelligence, pp.103-106,1999.

[13] Fan, W. Miller, M.Stolfo, S.Lee & P Chan, “Using Artificial Anomalies to Detect Unknown and Known Network Intrusions,” Proc. of ICDM01, pp.504-507, 2001.

[14] Manoel Fernando Alonso Gadi, Xidi Wang, Alair Pereira do Lago, “Credit Card Fraud Detection with Artificial Immune System,” Lecture Notes in Computer Science, Vol. 5132/2008, pp.119-131, 2008.

[15] E. Aleskerov, B. Freisleben, B. Rao, “CARDWATCH: A Neural Network Based Database Mining System for Credit Card Fraud Detection,” Proceedings of IEEE/IAFE Conference on Computational Intelligence for Financial Engineering (CIFEr), pp.220-226, 1997.

[16] S. Benson Edwin Raj and A. Annie Portia, “Analysis on Credit Card Fraud Detection Methods”, ICCCET, 2001, p1-5 [17] M. Mehdi, S. Zair, A. Anou and M. Bensebti,” A Bayesian Networks in Intrusion Detection Systems,” International Journal of Computational Intelligence Research, Issue No. 1, pp.0973-1873 Vol. 3, 2007.

[18] Lam, Bacchus, “Learning bayesian belief networks: An approach basedon the MDL principle,” Computational Intelligence, Vol. 10, Issue No. 3, pp.269–293, August 1994. [19] Sam Maes, Karl Tuyls, Bram Vanschoenwinkel, Bernard Manderick, “Credit card fraud detection using Bayesian and neural networks,” Interactive image-guided neurosurgery, pp.261-270, 1993.

Figure

Fig. 1 – Hybrid approach for CCFD (excerpt from [16])
Fig. 2 – Overview of HMM Model (excerpt from [16])
Fig. 4 – Dataset used for experiments
Fig.  7  –  Performance  of  CCFD  techniques  with  respect  to  false positives

References

Related documents

KEYWORDS :Arduino, BLDC Motor, Chassis Controller, Hall Effect RPM Sensor, Battery, Treadmill..

Antropol. Inherited predisposition to myeloproliferative neoplasms. Ther Adv Hematol. Neoplastic hematopathology: Experimental and Clinical Approches. Some speculations on

In the light of cultural back- ground, it was hypothesized that there would be a negative correlation between PTSD symptoms and re- silience among burn patients, female burn

The incision was closed with 5-0 Chromic P-3 passed on a plastic needle holder and Adson tissue forceps with teeth. The final count was performed when the

These were (1) evidence of contestation over the parameters of nationally defined membership and rights regimes; (2) proposals for alternative solidari- ties rooted in commitments

The results provided evidence that there is a significant negative relationship between increasing volatility in the exchange rate and export flows in both the short and long run

Specifically, we investigate renewable energy research trends for the past 20 years, based on geographic distribution of studies, methods used in the corpus, categorization

Byarugaba et al Virology Journal 2014, 11 173 http //www virologyj com/content/11/1/173 RESEARCH Open Access High pathogenicity and low genetic evolution of avian paramyxovirus type