JavaScript中基于规则的广告脚本检测

发布时间：2018-05-17 09:38

本文选题：JavaScript + 广告脚本检测　；参考：《南京邮电大学》2017年硕士论文

【摘要】：随着互联网的迅猛发展,网络广告成为互联网最重要的商业模式之一。网络广告在促进互联网发展的同时,也带来了用户信息泄露、影响用户网页浏览体验等负面问题。为了屏蔽网页中出现的广告,Adblock、Ghostery等广告屏蔽软件层出不穷,他们大多使用模式匹配的方法,将网页中的URL和选择器属性与已知的广告相关黑名单列表进行匹配再进行过滤,但是对于新出现的广告相关字段以及经过混淆的广告则无法检测。本文经过对动态广告生成原理的分析发现,第三方脚本库中自动执行的JavaScript脚本,往往与动态广告的加载有关。文章使用JavaScript的动态插桩工具Jalangi对符合条件的脚本函数获取其完整的函数调用链信息。首先在JavaScript函数中动态绑定一个用于存储调用路径的属性,对相关函数调用路径进行追踪和保存;然后,在插桩过程中对于setTimeout()、set Interval()、call()、apply()等特殊函数,根据函数的特点不同进行不同的处理以保证函数调用链路的完整传递;最终,根据广告相关JavaScript调用路径获得广告相关和广告无关的脚本文件。在构造分类器阶段,本文首先对经过动态插桩得到的广告相关JavaScript文件进行分析,并总结出用于区分广告相关和广告无关的JavaScript静态特征;再根据静态特征根据已经检测到的广告相关脚本文件进行分类器的训练;最后使用新的脚本文件测试集对分类器的效果进行验证。实验证明该分类器对广告脚本文件检测的准确率达到了93.86%,整个分类器可以较有效的对广告相关JavaScript脚本文件进行识别,便于在网页加载前对脚本文件进行屏蔽并阻止广告的显示,弥补了Adblock等广告屏蔽插件的不足。
[Abstract]:With the rapid development of the Internet, Internet advertising has become one of the most important business models. While promoting the development of Internet, network advertisement also brings negative problems such as user information leakage and web browsing experience. In order to block ads, such as Adblockor Ghostery, they often use pattern matching methods to match and filter the URL and selector attributes in the web pages with known lists of advertising related blacklists. But new advertising related fields and confused ads are undetectable. Based on the analysis of the principle of dynamic advertisement generation, it is found that the JavaScript script executed automatically in the third party script library is often related to the loading of dynamic advertisement. In this paper, Jalangi, a dynamic piling tool of JavaScript, is used to obtain the complete function call chain information for the qualified script function. First, dynamically bind a property used to store the call path in the JavaScript function to trace and save the call path of the related function; then, for special functions such as setTimeOutout during the piling process, According to the characteristics of the function, different processing is carried out to ensure the complete transfer of the function call link; finally, according to the Advertising related JavaScript call path, the Advertising and Advertising independent script files are obtained. In the stage of classifier construction, this paper firstly analyzes the advertising related JavaScript files obtained by dynamic piling, and sums up the static features of JavaScript used to distinguish advertising relevance from advertising independence. Then the classifier is trained according to the static features according to the detected advertisement related script files. Finally, the new script file test set is used to verify the effect of the classifier. Experimental results show that the classifier can detect the advertisement script file with the accuracy rate of 93.86. The whole classifier can recognize the advertisement related JavaScript script file effectively, and it is convenient to block the script file and prevent the display of the advertisement before loading the web page. Make up for Adblock and other advertising blocking plug-in deficiencies.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.092;TP312.2

【相似文献】