基于神经网络和SNA的C2C电子商务信誉欺诈识别研究

Detection of C2C reputation fraud activities based on neural network and SNA

互联网中信誉欺诈行为严重影响了C2C电子商务的发展。如何有效识别互联网上的信誉欺诈商户是当前的研究热点。阐述了将基于交易历史的社会网络分析用于构建C2C电子商务信誉欺诈识别指标体系的原理及过程,通过使用LVQ神经网络对雅虎奇摩拍卖网采集的大量用户交易数据进行分类,有效识别出控制多个虚假账户以达到信用累计目的的信誉欺诈商户,对维护C2C电子商务交易的稳定性具有重要意义。

The C2C reputation fraud activities have seriously affected the development of C2C business in the World Wide Web.It is currently a hot research topic on how to detect reputation fraud merchants.This paper briefly introduced the principle and process of implementing social network analysis（SNA） based on transactions into building up the detection indices of C2C reputation fraud activities.With the help of LVQ neutral network,this paper successfully detected those accounts who quickly gained too much reputation through controlling many false accounts from the crawled Taiwan Yahoo accounts,which was important for the stability of C2C online business.