SELAB investigates the integration of Artificial Intelligence and Software Engineering, with a strong emphasis on AI-centric methodologies and systems.
End-to-End Multi-Agent SE Automation
DevSecOps pipeline with multi-agent coordination automating the full software maintenance lifecycle from bug reports to patches, tests, and documentation.
Multi-Agent Program Repair
AST-anchored retrieval and multi-agent coordination for automated bug fixing. Agents collaborate to localize faults, retrieve relevant context, and generate validated patches.
Automated Bug Report Generation
CTQRS framework with reinforcement learning to generate complete, reproducible bug reports. Combines crash trace analysis with structured report synthesis.
Structure-Aware Bug Localization
Zero-shot retrieval with AST-augmented semantic search for pinpointing buggy code regions. Bridges natural language bug descriptions to code structure.
Context-Aware Commit Message Generation
Generates meaningful commit messages by modeling temporal continuity and developer history. Captures project-specific conventions and change semantics.
Security Bug Analysis
Cross-project similarity learning and data augmentation for security bug classification. Deep learning models identify and categorize vulnerability patterns.