Tangzhi Xu

Softwiser Group

Nanjing University

xutz@smail.nju.edu.cn

I am a fourth-year Ph.D. candidate in the Softwiser Group at the Institute of Computer Software, Nanjing University, supervised by Assoc. Prof. Yuan Yao and Prof. Feng Xu.

My current research focuses on Coding Agent.

news

Jul 30, 2026	Our paper “CogPath: A Constraint-Guided Context-Reduction Framework for LLM-Based Test Generation” was accepted by SCAM 2026.
May 11, 2026	Our paper “Streamlining Repository Tasks with Effective Snippet Retrieval” was accepted by ACM TOSEM.
Jan 23, 2026	Our paper “Commit Message Generation Based on Large Language Models” was accepted by Journal of Software (软件学报).
Aug 01, 2025	Our paper “Comprehend, Imitate, and then Update: Unleashing the Power of LLMs in Test Suite Evolution” was accepted by ASE 2025.
Jul 15, 2025	Our paper “LoRA Decompose: Serving Fine-Tuned Models into LoRA-Like” was accepted by ECAI 2025.

selected publications

SCAM

CogPath: A Constraint-Guided Context-Reduction Framework for LLM-Based Test Generation

Jianhan Liu, Tangzhi Xu, Yuan Yao, and 2 more authors

In IEEE International Working Conference on Source Code Analysis and Manipulation, 2026 (CCF-C)

Bib PDF Code

@inproceedings{liu2026cogpath,
  title = {CogPath: A Constraint-Guided Context-Reduction Framework for LLM-Based Test Generation},
  author = {Liu, Jianhan and Xu, Tangzhi and Yao, Yuan and Xu, Feng and Ma, Xiaoxing},
  booktitle = {IEEE International Working Conference on Source Code Analysis and Manipulation},
  year = {2026},
}

TOSEM

Streamlining Repository Tasks with Effective Snippet Retrieval

Tangzhi Xu, Cong Li, Zhaogui Xu, and 7 more authors

ACM Transactions on Software Engineering and Methodology, Jun 2026 (CCF-A)

DOI Bib PDF Code

@article{xu2026repoet,
  title = {Streamlining Repository Tasks with Effective Snippet Retrieval},
  author = {Xu, Tangzhi and Li, Cong and Xu, Zhaogui and Jiang, Yanyan and Yao, Yuan and Zhu, Xiaorui and Xu, Feng and Di, Peng and Xu, Chang and Su, Zhendong},
  journal = {ACM Transactions on Software Engineering and Methodology},
  year = {2026},
  month = jun,
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  issn = {1049-331X},
  url = {https://doi.org/10.1145/3821433},
  doi = {10.1145/3821433},
  keywords = {RAG, language models, program repair, repository-level tasks},
}

JOS

Commit Message Generation Based on Large Language Models

Shengbin Xu, Linjie Jia, Tangzhi Xu, and 4 more authors

Journal of Software, May 2026

In Chinese; published online

DOI Bib HTML PDF

@article{xu2026commitmessage,
  title = {Commit Message Generation Based on Large Language Models},
  author = {Xu, Shengbin and Jia, Linjie and Xu, Tangzhi and Zhu, Xiaorui and Yu, Ping and Xu, Feng and Ma, Xiaoxing},
  journal = {Journal of Software},
  year = {2026},
  month = may,
  note = {In Chinese; published online},
  doi = {10.13328/j.cnki.jos.007640},
}

ASE

Comprehend, Imitate, and then Update: Unleashing the Power of LLMs in Test Suite Evolution

Tangzhi Xu, Jianhan Liu, Yuan Yao, and 3 more authors

In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering, May 2025 (CCF-A)

DOI Bib PDF Code

@inproceedings{xu2025comprehend,
  title = {Comprehend, Imitate, and then Update: Unleashing the Power of LLMs in Test Suite Evolution},
  author = {Xu, Tangzhi and Liu, Jianhan and Yao, Yuan and Li, Cong and Xu, Feng and Ma, Xiaoxing},
  booktitle = {Proceedings of the IEEE/ACM International Conference on Automated Software Engineering},
  year = {2025},
  doi = {10.1109/ASE63991.2025.00046},
}

ECAI

LoRA Decompose: Serving Fine-Tuned Models into LoRA-Like

Yibo Han, Tangzhi Xu, Zenan Li, and 4 more authors

In Proceedings of the European Conference on Artificial Intelligence, May 2025 (CCF-B)

DOI Bib PDF

@inproceedings{han2025lora,
  title = {LoRA Decompose: Serving Fine-Tuned Models into LoRA-Like},
  author = {Han, Yibo and Xu, Tangzhi and Li, Zenan and Zhang, Yijia and Miao, Youshan and Yao, Yuan and Xu, Ningyi},
  booktitle = {Proceedings of the European Conference on Artificial Intelligence},
  year = {2025},
  doi = {10.3233/FAIA251354},
}

ISSTA

Clause2Inv: A Generate-Combine-Check Framework for Loop Invariant Inference

Weining Cao, Guangyuan Wu, Tangzhi Xu, and 4 more authors

In Proceedings of the ACM SIGSOFT International Symposium on Software Testing and Analysis, May 2025 (CCF-A)

DOI Bib PDF Code

@inproceedings{cao2025clause2inv,
  title = {Clause2Inv: A Generate-Combine-Check Framework for Loop Invariant Inference},
  author = {Cao, Weining and Wu, Guangyuan and Xu, Tangzhi and Yao, Yuan and Wei, Hengfeng and Chen, Taolue and Ma, Xiaoxing},
  booktitle = {Proceedings of the ACM SIGSOFT International Symposium on Software Testing and Analysis},
  year = {2025},
  doi = {10.1145/3728920},
}