大规模移动应用第三方库自动检测和分类方法

发布时间：2018-05-29 12:00

本文选题：Android + 第三方库　；参考：《软件学报》2017年06期

【摘要】：移动应用中,广泛使用第三方库来帮助开发和增强应用功能.很多关于移动应用分析以及访问控制的研究工作,需要在分析之前对第三方库进行检测、过滤或者对其进行功能分类.当前,大部分研究工作都以使用白名单的方式来检测第三方库或者对其功能进行分类.然而,通过白名单检测第三方库不完善且不准确,其原因包括:(1)第三方库的种类和数量很大;(2)常见的代码混淆或者第三方库伪装等技术使得白名单方法不能准确地识别第三方库.提出一种第三方库自动检测和分类方法,包括基于多级聚类技术准确识别第三方库以及基于机器学习对第三方库的功能进行准确分类.实验对超过130 000个Android应用进行分析,验证所提出方法的有效性.实验总共检测到4 916个不同的第三方库.在人工标记的数据集上,通过十折交叉验证,对第三方库分类的准确率达到84.28%.将训练好的分类器应用于全部4 916个检测到的第三方库,人工进行抽样验证的准确率达到75%.
[Abstract]:Third party libraries are widely used in mobile applications to help develop and enhance applications. A lot of research work on mobile application analysis and access control needs to detect filter or classify the third party library before analyzing. Currently, most of the research work uses whitelist to detect or classify the functions of third-party libraries. However, the whitelist detection of third-party libraries is incomplete and inaccurate, The reasons include: (1) the category and number of third party libraries are very large (2) the common techniques such as code confusion or third party library camouflage make the whitelist method unable to identify the third party library accurately. A method for automatic detection and classification of third party libraries is proposed, which includes accurate identification of third party libraries based on multilevel clustering technology and accurate classification of third party libraries based on machine learning. More than 130,000 Android applications are analyzed to verify the effectiveness of the proposed method. A total of 4,916 different third-party libraries were detected. The accuracy of classification of third party library is 84.28 by 10% cross verification on the data set of manual mark. The trained classifier is applied to all 4 916 detected third party libraries, and the accuracy of manual sampling verification is 75%.
【作者单位】：智能通信软件与多媒体北京市重点实验室(北京邮电大学计算机学院);高可信软件技术教育部重点实验室(北京大学);北京大学信息科学技术学院软件研究所;
【基金】：国家自然科学基金(61421061,61421091) 国家高技术研究发展计划(863)(2015AA017202)~~
【分类号】：TP181;TP311.13

【相似文献】