SemLoc: Structured Grounding of Free-Form LLM Reasoning for Fault Localization

Zhaorui Yang, HaiChao Zhu, Qian Zhang, Rajiv Gupta, Ashish Kundu UC Riverside Cisco Research Independent Researcher · 2026

Read SemLoc: Structured Grounding of Free-Form LLM Reasoning for Fault Localization by Zhaorui Yang, HaiChao Zhu, Qian Zhang, Rajiv Gupta, Ashish Kundu UC Riverside Cisco Research Independent Researcher (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-29109.txt.

Description

Yang and co-authors present SemLoc, a tool that finds the faulty lines behind software failures. Standard tools rank code by execution patterns. Those signals fail on semantic bugs, where runs follow the same path. Language model tools reason about meaning but give unverifiable answers. SemLoc turns that free-form reasoning into a typed, checkable form. It builds a matrix of broken constraints across tests. On 250 Python programs, SemLoc reached 42.8 percent top-one accuracy.

Catalog record

CREATORZhaorui Yang, HaiChao Zhu, Qian Zhang, Rajiv Gupta, Ashish Kundu UC Riverside Cisco Research Independent Researcher
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56627
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.29109

Browse the full library · About the Tower of Babel