SweRank: Software Issue Localization with Code Ranking

Revanth Gangi Reddy Tarun Suresh JaeHyeok Doo Ye Liu Xuan Phi Nguyen, Yingbo Zhou Semih Yavuz Caiming Xiong Heng Ji Shafiq Joty · 2025

Read SweRank: Software Issue Localization with Code Ranking by Revanth Gangi Reddy Tarun Suresh JaeHyeok Doo Ye Liu Xuan Phi Nguyen, Yingbo Zhou Semih Yavuz Caiming Xiong Heng Ji Shafiq Joty (2025) 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-2505-07849.txt.

Description

Revanth Gangi Reddy and nine co-authors present SweRank, in a 2025 preprint, a fast way to find the code behind a bug report. Given a plain-language issue, developers must locate the files and functions to change. That hunt eats time on every project. AI agent systems can do it, but they are slow, pricey, and often closed. Classic ranking models are cheap but trip on the wordy, failure-focused style of issue text. SweRank pairs a retriever with a reranker trained for this exact job.

To train it, the team built SweLoc, a large dataset from public GitHub projects. It pairs real issue reports with the code edits that fixed them. On two standard test sets, SweRank topped both older rankers and costly agent systems. The dataset also lifted other retrievers and rerankers when they trained on it. The authors offer both as resources for cheaper, open issue localization.

Catalog record

CREATORRevanth Gangi Reddy Tarun Suresh JaeHyeok Doo Ye Liu Xuan Phi Nguyen, Yingbo Zhou Semih Yavuz Caiming Xiong Heng Ji Shafiq Joty
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-07849

Browse the full library · About the Tower of Babel