logpai/bughub

Provides curated datasets across 12+ major open-source projects (Mozilla, Eclipse, Apache ecosystem) with chronologically-split train/test partitions specifically designed for NLP-based duplicate detection research. Includes complementary bug localization datasets linking reports to source files, plus comprehensive evaluation metrics on duplicate prevalence (6.8%-38.4% across projects) and resolution timelines. Benchmarks against established approaches from two decades of peer-reviewed research in automated bug triage and information retrieval.

123 stars. No commits in the last 6 months.