Zhiyuan Zhong

Southern University of Science and Technology (SUSTech), shenzhen, China

portrait_crop.jpg

I am Zhiyuan Zhong, a final year undergraduate majoring in Computer Science at SUSTech. I am also doing research in Software Quality Lab, supervised by Prof. Yepang Liu.

My current research is on Software Engineering, specifically LLM4SE. I am exploring the application of Large Language Models in discovering potential bugs within libraries. I am also very interested in research fields inlcuding Natural Language Processing, trustworthiness & safety of Large Language Models.

News

Jul 19, 2024 I will participate in the 2024 Information Hub Summer Camp at HKUST(GZ).
Jul 14, 2024 Our latest work on fuzzing fastjson2 is submitted to ASE 2024 (Industry Showcase). We leverage large language models (LLMs) to fuzz fastjson2, a JSON library from Alibaba. Based on unit tests, we utilize LLMs to generate diverse test cases by incorporating JSON-specific mutation rules. While manual inspection reveals that LLM-generated tests can be erroneous with self-contradictory assertions, we demonstrate that LLMs have the potential for classifying false-positive test failures. As of the end of June 2024, we have identified 34 bugs in fastjson2 , with 30 of them already fixed. The full list of bugs can be found here.
Jun 28, 2024 I am a contributor to the fastjson2 (3.6k+ stars) project.

Selected Publications

  1. fuzz_fastjson2.png
    Fuzzing Data-Serialization Library With Large Language Models - A Practical Case In Fastjson2
    Zhiyuan Zhong, Sinan Wang, Hailong Wang, Shaojin Wen , and 3 more authors
    The 39th IEEE/ACM International Conference on Automated Software Engineering (ASE ’24) Industry Showcase Track, under review, 2024