【摘要】 安卓应用程序开发者试图在程序中嵌入广告库来获取收益，然而广告库的广泛引入增加了用户隐私泄露风险。为了准确自动识别嵌入的广告库，从静态分析的角度出发提出了一种基于语义的安卓应用程序广告库检测方法。该方法使用元信息识别主模块和非主模块，然后对非主模块进行语义特征提取，最后采用经典的机器学习方法训练广告库检测分类器。在Andrzoo的公开数据集上进行了实验验证，结果表明该方法取得了99.2%的召回率，并应用于安卓市场广告库检测，获得的准确率达97%。更多还原

【Abstract】 Android application developers try to embed ad libraries in their programs to gain revenue,however,the widespread introduction of ad libraries increases the risk of user privacy leakage. In order to accurately and automatically identify the embedded ad libraries,a semantic-based method for detecting ad libraries in Android applications is proposed from the perspective of static analysis. The method uses meta-information to identify main and non-main modules,then semantic features are extracted for non-main modules,and finally a classical machine learning approach is used to train an ad library detection classifier. Experimental validation is conducted on Andrzoo’s public dataset,and the results show that the method achieves a recall of 99.2% and is applied to Android Market ad library detection,obtaining an accuracy of 97%.更多还原