Skip to main content

Software testing and debugging with/without AI/LLMs.

Primary supervisor

Yongqiang Tian

In this project, students and me will work together to develop a new technique for software testing and debugging. 

 

The subject under test may be AI/LLM. The technique may involve AI/LLM as well. 

Aim/outline

- develop/extend a new technique in software testing and debugging

- conduct in-depth experiments and analysis

- write a research paper/thesis in a professional way.  

URLs/references

Here are a few representative papers that may be of interest:

  • An Empirical Study of Bugs in Data Visualization Libraries.

    A work to understand the challenges and opportunities in testing specific software.

  • Compilation Consistency Modulo Debug Information

    A work to automatically test compilers.

  • Leveraging Large Language Models to Detect Missed Peephole Optimizations

    A work to test compilers with LLMs.

  • LPR: Large Language Models-Aided Program Reduction

    A work to reduce programs with LLMs, which benefits debugging software.

You DO NOT NEED to read all details of these papers, but at least the abstracts and introductions.

Required knowledge

- self-motivated, willing to spend time and efforts in research

- good programming and analysis skills

- good communication skills (oral and writing)

- good background in software testing and debugging

- (can be obtained during project) a reasonable knowledge of LLMs