融合多重特征的钓鱼网站聚类算法研究

发布时间：2018-01-12 23:21

本文关键词：融合多重特征的钓鱼网站聚类算法研究　出处：《中南林业科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：网络钓鱼作为电信诈骗的一种典型欺诈手段,已经引起学术界广发关注。网络钓鱼是一种利用社会工程和技术,针对用户身份数据和金融账号进行欺骗的犯罪机制。网络钓鱼防御是网络钓鱼的一种对抗手段。经过十几年的研究,网络钓鱼防御已经取得了一定的研究成果,但当前的网络钓鱼攻击采用团队化、机构化作战,攻击依旧很猖獗。针对当前网络钓鱼攻击者的团队化、机构化作战的行为方式,提出一种融合多重特征的钓鱼网站聚类算法。通过构造钓鱼网页DOM树,提取出描述钓鱼网站的特征,采用Kulczynski 2算法计算相似度,最后利用K-Means算法进行聚类分析。对eBay、PayPal的聚类结果分析,验证了网络钓鱼攻击者的团队化作案行为特征。本文的主要工作与贡献在于:(1)提出一种融合多重特征的钓鱼网站聚类算法。本论文取样于eBay、PayPal的钓鱼网站。首先,对数据进行预处理,并构造其DOM树。然后使用句法指纹算法进行特征描述,构造特征库进行统一索引,计算相似度。最后,基于这个数据库对钓鱼网页进行相似度聚类。(2)基于算法进行实验分析。对特征库中实验数据对象进行K-means聚类后,得到其聚类结果发现:eBay钓鱼网站聚类结果簇类中发现有83.6%的钓鱼网站相似度值相近或相等,PayPal钓鱼网站聚类结果簇类中发现有73.0%的钓鱼网站相似度值相近或相等。
[Abstract]:As a typical phishing fraud means of telecommunications fraud, has attracted academic attention. GF phishing is a use of social engineering and technology, to carry out criminal deception mechanism for user identity data and financial account. Phishing defense is a means of phishing. After ten years of research, phishing defense has made some achievements, but the current phishing attacks by team, mechanism of combat, attack is still rampant. In view of the current phishing attack team, the way of operational mechanism, proposed a fusion of multiple features of phishing phishing web clustering algorithm. By constructing DOM tree, extraction a description of the characteristics of phishing sites, using Kulczynski algorithm to calculate the similarity of 2, finally using the K-Means algorithm for clustering analysis. The analysis results of eBay, PayPal clustering The verification team into the case, behavior characteristics of phishing attackers. The main work and contributions of this paper are as follows: (1) proposed a fusion of multiple features of phishing sites clustering algorithm. This paper sampling in eBay, PayPal of the phishing site. Firstly, preprocess data, and construct the DOM tree. Then use the fingerprint syntactic feature description algorithm, structure feature library unified index, calculating the similarity. Finally, based on the database of phishing web similarity clustering. (2) the experiments of image analysis based on K-means clustering. After the experimental data in the feature library, the clustering results showed that: eBay clustering results in clusters of phishing sites there are 83.6% phishing sites found similarity values of similar or equal PayPal, the clustering results in clusters of phishing sites phishing sites found similarity value of 73% similar or equal.

【学位授予单位】：中南林业科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.092;TP311.13

【相似文献】