Bug Reports-based Test Case Generation (BURGEON)

BURGEON automates test case generation from bug reports, improving bug-triggering input discovery and test oracles for better fault localisation and repair.

The project at a glance

Start date:

04 Apr 2022
Duration in months:

48
Funding:

AFR
Principal Investigator(s):

Tegawendé François d Assise BISSYANDE

About

Because software is pervasive, the user base of software is ever increasing. This results in endless encounters between users and software, which often end up with some unexpected output or incorrect results on the user side. In many cases, users report the bugs to the developers. Oftentimes, however, too many bugs are reported that exceed the human resources and time available for triaging and fixing those bugs. Researchers have therefore proposed techniques to automate several of the bug reports, triaging tasks such as prioritising, classifying, and assigning bug reports. Recently, researchers have started to focus on how to leverage bug reports in actionable software engineering tasks such as fault localisation and automatic program repair. Nonetheless, these approaches generally do not actually dig into bug reports: for example, they use token matching between a bug report and source code to perform localisation at the file level. We argue that the current utilisation of bug reports in software engineering tasks is still suboptimal with great room for improvement and expansion. For example, many bug reports include specific information that would help in generating test cases that could reveal the bug being reported. Such tests are very useful for the purpose of bug localisation and fixing. Our research project BURGEON will focus specifically on test case generation based on bug report information. In particular, we will address two typical problems in automatic testing: finding the right bug triggering inputs, and inferring test oracles.