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
2025-05-15
Summary
This paper talks about SweRank, a new AI system that helps programmers quickly find the exact spot in their code where a software bug or issue is happening, using a smart ranking method and a special dataset called SweLoc.
What's the problem?
The problem is that when software breaks or has a bug, it can be really hard and time-consuming for developers to figure out which part of the code is causing the problem, especially in large and complex projects.
What's the solution?
The researchers built SweRank, which first searches through the code and then ranks the most likely locations of the bug, making it much easier to pinpoint the issue. They tested this system using the SweLoc dataset and found that it works better than both older methods and newer agent-based approaches.
Why it matters?
This matters because it saves programmers a lot of time and effort when fixing software, leading to faster updates, fewer errors, and better software quality for everyone.
Abstract
SweRank, an efficient retrieve-and-rerank framework using the SweLoc dataset, achieves state-of-the-art performance in software issue localization, outperforming both traditional models and agent-based systems.