This repository supports collaborative research on AI-generated code, human-AI software development, repository mining, code provenance, AI-code detection, and software maintenance analysis.
CODE/: Source code, scripts, datasets, and experiments.DOC/: Related papers, proposals, meeting notes, presentations, and grant materials.
Lead: Dr. Harvey Siy
Study repository histories, code provenance, code lineage, GitHub mining, historical version recovery, and commit-level development behavior. This aim focuses on constructing human-written code baselines and reconstructing how code artifacts originate, evolve, propagate, change, or disappear over time.
Lead: Dr. Myoungkyu Song
Develop and evaluate methods for detecting AI-generated code using supervised LLM-based models, perturbation-based scoring, predictability measures, AST embeddings, and hybrid classification approaches.
Lead: Dr. Jaydeb Sarker
Analyze how AI-tool adoption affects collaborative software maintenance, including productivity, review effort, rework, debugging activity, integration difficulty, code-quality warnings, code smells, and code complexity.
The project aims to understand how AI-powered coding tools reshape collaborative software development and maintenance, with a focus on code authorship, AI-generated code detection, productivity, quality, and long-term software sustainability.