Ahmadreza Saboor Yaraghi, Golnaz Gharachorlu, Sakina Fatima, Lionel Briand, Ruiyuan Wan, Ruifeng Gao
Software Engineering Empirical Research Radar
Efficient Black-Box Fault Localization for System-Level Test Code Using Large Language Models
Paper detail page in SEER Radar.
TSE / ASE Journal First 2026
Debugging / Fault Localization / Diagnosis / Repair; AI / LLM for SE; Testing / QA
Structured Fault Localization
Abstract / Summary
A static, black-box method estimates a system-level test script trace from one failure log, prunes the code presented to an LLM, and ranks faulty test-code locations without repeated executions or SUT source access. The industrial evaluation reports block-level Hit@3 of 81% and substantial token and inference-time reductions.
External Links
Local PDF is not available on SEER Radar yet. When a public source is recorded, this page will add a local reading link with source attribution.