Scholarship

Publications

Papers and preprints on LLM4SE, AI4SE, and intelligent software engineering, with a focus on repository-level code intelligence and AI-powered software maintenance.

2026

arXiv preprint arXiv:2604.26523

RepoDoc: A Knowledge Graph-Based Framework to Automatic Documentation Generation and Incremental Updates

Dong Xu, Mingwei Liu, Xiwen Wang, Jianfeng Zhong, Zibin Zheng

RepoDoc builds repository knowledge graphs to generate modular documentation, propagate semantic impact, and update affected documents incrementally after code changes.

arXiv PDF Code

RepoDoc architecture: repository knowledge graphs, semantic impact propagation, and incremental documentation updates.

arXiv preprint arXiv:2602.20799

Unseen-Codebases-Domain Data Synthesis and Training Based on Code Graphs

Guangsheng Ou, Qiming Zhang, Sirong Chen, Anji Li, Dong Xu, Tiancheng Luo, Dekun Dai, Cuiyun Gao, Long Wang, Jun Zhou, Mingwei Liu, Zibin Zheng

This work uses code graphs to synthesize repository-domain training data, helping models adapt to previously unseen codebases and reason across project-level dependencies.

arXiv PDF

Code graph data synthesis overview — Overview of code-graph-based data synthesis and training for unseen codebases.