基于SVM和TF-IDF的恶意URL识别分析与研究  

Analysis and Research of Malicious URL Recognition Based on SVM and TF-IDF

作　　者：甘宏[1] 潘丹[1] 

机构地区：[1]广州城建职业学院,广东广州510925

出　　处：《计算机与现代化》2016年第7期95-97,102,共4页Computer and Modernization

基　　金：国家自然科学基金资助项目(61272067);广东省自然科学基金团队研究资助项目(S2012030006242)

摘　　要：随着互联网尤其是移动互联网的快速发展,全球范围内出现了越来越多带欺诈和破坏性质的站点。本文通过分析URL的文本特征和站点特征,提出一种基于机器学习的URL检测方案,用TF-IDF算法细化了URL的站点特征,并结合以上特征使用基于RBF核的SVM进行URL安全检测,得到了96%的准确率和0.95的F1分数。With the rapid development of the Internet , especially the mobile Internet , there are more and more sites that have been brought out and destroyed in the world .In this paper , we propose a URL detection scheme based on machine learning , through analyzing the features of URL ’s text and sites.The URL’s site feature is refined by TF-IDF algorithm, the URL security detection is carried out with SVM kernel based on RBF kernel , and it obtained 96%auuracy and 0.95 F1 sore.

关 键 词：网络安全 URL检测 TF-IDF SVM 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]