基于特征筛选的轻量级层次化钓鱼网站的检测方法研究

Lightweight Hierarchical Phishing Website DetectionBased on Feature Filtering

针对传统方法难以对大规模钓鱼网站进行批量检测的问题,提出基于特征筛选的轻量级层次化检测方法(lightweight hierarchical detection method based on feature filtering,LHFF).该方法首先使用互信息对原始特征集进行筛选,剔除冗余特征,并将筛选后的特征按照提取特征耗时长短划分为URL特征和网站特征,然后根据划分后的特征,使用轻量级层次化检测框架对钓鱼网站进行检测.实验结果表明,LHFF能够在保障良好检测性能的前提下,减少网站检测所需要的时间,满足对大规模钓鱼网站进行批量检测的需求.

For the problem that traditional methods are difficult to detect large-scale phishing websites in batch,a lightweight hierarchical detection method based on feature filtering(LHFF)is proposed.The method first uses mutual information to filter the original feature set,eliminates redundant features,and divides the filtered features into URL features and website features according to the time spent on extracting features,and then uses the lightweight hierarchical detection framework to detect phishing websites according to the divided features.The experimental results show that LHFF can reduce the time required for website detection and meet the demand for batch detection of large-scale phishing websites while guaranteeing good detection performance.