针对安卓应用的软件自动修复算法研究与系统构建

Recently, Android apps has become indispensable, its security and reliability has gained more and more attention. To reduce the cost of fixing defects and software vulnerabilities, automated program repair techniques has become a hot topics in scientific research. Compared to traditional software systems, mobile applications has its unique properties. However, research on automated repair system of mobile applications is still at its early stage. This proposal aims to systematically study defects for Android apps and the unique properties of its patches, investigate the current status of automated repair of Android apps, and to propose a more efficient and stable approach for automated repair. Firstly, this proposal will investigate semantic repair of Android apps, and to integrate it with online repair techniques, in order to improve the quality of generated patches and the efficiency of patch generation algorithm. Secondly, this proposal will collect and analyze bug-fix patterns for Android apps, and using multi-objective optimization algorithm to search for patches that satisfy functional and non-functional properties. Lastly, we will study the relationship between the quality of test suite and automated repair for Android apps in order to propose a more efficient algorithm for test generation and to improve the quality of generation patches.

近年来，移动应用已经渗透到了各个方面，其安全性和可靠性引发了越来越多的关注。为了减少修复程序缺陷和安全漏洞过程中的开销，自动修复技术成为了科学研究的热点。与传统软件系统相比，移动应用有其特殊性。 然而，移动应用自动修复的研究也还处在初级阶段。本项目旨在系统化调查安卓程序缺陷和补丁的特殊性，探究安卓程序自动修复现状，以及提出更加高效、稳定的自动修复方案。首先，本项目将探索针对安卓程序的基于约束求解的补丁生成算法，并将其与现有在线修复技术结合，以提升补丁的质量和补丁生成的效率。其次，本项目将采集和分析安卓应用的补丁模式，并采用多目标优化算法搜索满足功能性和非功能性需求的补丁。最后，我们将研究程序测试用例质量与自动修复的关系，以期提出高效的测试用例生成算法，并进一步提高生成的补丁的质量。

近年来，移动应用已经渗透到了各个方面，其安全性和可靠性引发了越来越多的关注。为了减少修复程序缺陷和安全漏洞过程中的开销，自动修复技术成为了科学研究的热点。与传统软件系统相比，移动应用有其特殊性。 然而，移动应用自动修复的研究也还处在初级阶段。本项目旨在系统化调查安卓程序缺陷和补丁的特殊性，探究安卓程序自动修复现状，以及提出更加高效、稳定的自动修复方案。首先，本项目将探索针对安卓程序的基于约束求解的补丁生成算法，并将其与现有在线修复技术结合，以提升补丁的质量和补丁生成的效率。其次，本项目将采集和分析安卓应用的补丁模式，并采用多目标优化算法搜索满足功能性和非功能性需求的补丁。最后，我们将研究程序测试用例质量与自动修复的关系，以期提出高效的测试用例生成算法，并进一步提高生成的补丁的质量。