OASIS: prioritizing static analysis warnings for Android apps based on app user reviews
ESEC/SIGSOFT FSE(2017)
摘要
Lint is a widely-used static analyzer for detecting bugs/issues in Android apps. However, it can generate many false warnings. One existing solution to this problem is to leverage project history data (e.g., bug fixing statistics) for warning prioritization. Unfortunately, such techniques are biased toward a project’s archived warnings and can easily miss newissues. Anotherweakness is that developers cannot readily relate the warnings to the impacts perceivable by users. To overcome these weaknesses, in this paper, we propose a semantics-aware approach, OASIS, to prioritizing Lint warnings by leveraging app user reviews. OASIS combines program analysis and NLP techniques to recover the intrinsic links between the Lint warnings for a given app and the user complaints on the app problems caused by the issues of concern. OASIS leverages the strength of such links to prioritize warnings. We evaluated OASIS on six popular and large-scale open-source Android apps. The results show that OASIS can effectively prioritize Lint warnings and help identify new issues that are previously-unknown to app developers.
更多查看译文
关键词
Static analysis,warning prioritization,Android Lint,app user reviews,natural language processing,concept graph
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络