Hi, This is Shiwen Shan (单诗雯) 👋
I am currently a third-year Ph.D. student at 
Sun Yat-sen University (SYSU). Before starting my Ph.D., I received my bachelor’s degree from the same institution 😉.
I am supervised by Prof. Su and Prof. Zheng.
My research focuses on software observability, with the goal of making complex software systems easier to understand, diagnose, and evolve. I am particularly interested in agentic software systems, such as programming and GUI agents. My current work explores trajectory observability, tool-use auditing, and failure diagnosis to improve the performance and reliability of agent execution. Please feel free to contact me to discuss research ideas and potential collaborations!
Updating news.
[2025/12] 🎉 One first-authored paper on effective log compression has been accepted by ICSE 2026! Please check for the preprint!
[2025/08] 🎉 Happy to be a visiting student in Singapore Management University (SMU), as a part of RISE Lab, under the supervision of Prof. Yintong Huo!
[2025/06] 🚀 First-authored paper on improving configuration diagnosibility via proactive logging has been accepted by ICSE 2026 (60/660, 9.1%)! Check here for the preprint!
[2024/09] 🎉 Proud to be a PhD student in Sun Yat-sen University.
[2024/03] 🚀 First-authored paper on configuration error localization via logs has been accepted by ISSTA 2024!
[2023/07] 🎉 Had a wonderful journey as a visiting student in ARISE Lab in CUHK, thanks for the warm welcoming! 😊
Some selected publications.
LogFold
The log compressor is designed to identify cross-template redundancies (i.e., the structured tokens) and propose the hybrid encoding strategy (i.e., the mixed-type encoding) for log compression.
Implementation
Rust C++Technologies
Encoding
The configuration logging framework is designed to enhance configuration diagnosability via automatically generated configuration logs.
Implementation
JavaTechnologies
Program Analysis LLMs
The automation tool is employed to localize configuration errors via logs with empowered LLMs.