Research On E-commerce Platform Reputation Manipulation Detection Based On Outlier Mining

Up to December 2015,the number of China's online shopping users reached 413 million,in other words,one in every four Chinese was involved in the online shopping,Chinese netizens have entered the era of online shopping.Due to the virtuality of network and the asymmetry of information in business activities,both sides,in particular the seller's credit is particularly important,in order to ensure the smooth progress of e-commerce transactions.The credit feedback system based on the history evaluation plays an important role in reflecting the seller's credit condition.However lately the increasingly rampant e-commerce credit fraud,also known as credit manipulation has seriously affected the function of credit feedback system.Credit risk has become the main barrier of the further development of e-commerce.It is necessary for e-commerce platform to take technical measures to crack down on credit manipulation and maintain the order of the industry.In order to solve the problem,this paper depends on outlier mining which is an advanced technique,aims at building credit manipulation detection model for e-commerce platform,in order to help to crack down on credit manipulation.Firstly,this paper cards the related research about credit manipulation,fraud detection in other industries,frequently-used fraud detection methods.Furthermore,this paper defines the credit manipulation detection issue precisely,builds a credit manipulation detection model,which includes four stages:makes index selection and explains the reasons,determines data collection and processing method,analyzes and compares a variety of detection methods,this paper chooses outlier mining method based on density as the algorithm in multi-sellers contrast analysis,introduces the basic concepts,and realizes the algorithms through Matlab programming,puts forward suspected sellers in-depth analysis method,that includes the establishment of rules base and the analysis of comment records.Thirdly,this paper applys the credit manipulation detection model in taobao,collects the relevant data in taobao,makes detection and analysis,determines the credit manipulation sellers according to the results of the analysis,and puts forward some suggestions for the e-commerce platform and government.In the end,this paper summarizes the research results,and explains the deficiency of this paper and the future research direction.