Rationale for code changes is scattered across artifact types (Goal in commit messages; Need and Alternatives in issues/PRs), and an LLM pipeline (ARGUS) extracts and summarizes these components with high recall (93.2%) but moderate precision (51.4%).
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Recovering Fine-Grained Code Change Rationale from Multiple Software Artifacts
Rationale for code changes is scattered across artifact types (Goal in commit messages; Need and Alternatives in issues/PRs), and an LLM pipeline (ARGUS) extracts and summarizes these components with high recall (93.2%) but moderate precision (51.4%).