一种基于URL语法规则的欺诈网站识别方法  

Fraudulent Sites Identification Based on URL Tree Patterns

作　　者：李华康[1] 赖龙彬[2] 陈光宣[3] 杨一涛[1] 孙国梓[1] 

机构地区：[1]南京邮电大学计算机学院,南京210003 [2]上海交通大学计算机科学与技术系,上海200240 [3]浙江警察学院,杭州310053

出　　处：《计算机科学》2015年第B10期28-33,共6页Computer Science

基　　金：本文受南京邮电大学引进人才科研启动基金（NY321085,NY214069）,江苏省自然科学基金（BK20140895）,国家自然科学基金（61502247）,公安部重点实验室开放课题（2015DSJSYS001）资助.

摘　　要：互联网金融P2P作为一种新型金融业务模式，实现了资金融通、支付、投资和信息中介等服务。互联网欺诈同样采用漫天撒网模式在极短时间内铺开发布范围，扩大受害面积后又快速抹去所有网络行踪。面对海量互联网站点中网络欺诈的后发性及网络服务器数据的易失性，一般在线电子数据取证方法面临海量检索发现慢、全网证据存储压力大等问题。借鉴自然语言处理中使用语法树结构核函数进行语法分析和处理的方法，提出一种基于URL模式树的站点功能分类模型，构造站点URL语法规则和URL路径语法树，并通过语法树核函数的改进对站点行为模式进行分类。实验结果表明，基于uRL语法规则的站点模式识别，能够快速有效地甄别站点类型，为海量互联网站点金融欺诈勘查数字取证提供快速筛选及发现服务。The peer-to-peer （P2P） platform,which is a new financial business model, services the financing, payment, investment and information intermediary. However, the Internet deceivers diffuse the fraudulent sites in wide range to expand victims quickly and destruct all the online information after the illegal income. According to the hysteresis of Internet fraud and volatilization of online data, it is difficult for traditional online digital forensic technologies to iden- tify the fraudulent sites or storage the massive relevant evidences. This paper proposes fraudulent sites identification approach based on URL Tree Patterns referring to language tree model and kernel function technology in Naturel Language Processing. The URL syntax rule and URL Syntax Tree are created to improve the classification of sites patterns. Experimental results show that the pattern recognition based on URL syntax rules can identify the site type with high accuracy. The approach proposed in this paper improves the fraudulent sites identification and discovery for digital forensics in massive Internet sites.

关 键 词：互联网金融 数字取证 URL语法规则 核函数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]