Research And Implementation Of Capturing Network Data Based On The Reputation Study Of C2C Websites

Lack of on-line-trust has become the key which prevent the on-line auction from quick and health growth.Defects of subscriber reputation evaluation with major C2C websites cause trust risk of the sites.This research is a sub-project of Project of online trade anti-fraud.The main substance is to design and implementation one high-performance,operational and stable worm.The worm will be able to pump mass trade data from pages of 3 major C2C sites inland,and restore them to the database with customized format as fundamental material for further machine study and analyze.The research is based on.NET framework and C# language.The ability of HTTP visit and multi-thread which embedded in C# is the key of capturing network data program.The research will find the classification of defrauder from mass data based neural network and SVM algorithm use nonsupervision learning pattern.The implementation worm has run over thousands of hours with high stability,and got more than 2 million complete trade records,met the aim of original design.