arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.29067cs.SE
收藏

超越定位:仓库级RAG-APR中的可恢复头寸与残余前沿

Beyond Localization: Recoverable Headroom and Residual Frontier in Repository-Level RAG-APR

University of Melbourne(墨尔本大学);Yanshan University(燕山大学);Aalto University(阿尔托大学)

Pengtao Zhao, Boyang Yang, Bach Le, Feng Liu, Haoye Tian

AI总结:

本研究探讨仓库级RAG-APR中定位强化后的可恢复增益与残余前沿,通过三种范式分析定位、搜索限制、证据质量及接口设计对修复结果的影响。

摘要:

仓库级自动化程序修复(APR)日益将更强的定位视为改进修复的主要路径。我们提出更具体的问题:一旦定位被加强,定位后仍能产生可恢复增益的杠杆是什么?这些增益在我们的协议中被限制,残余前沿又是什么?我们通过SWE-bench Lite,使用三种代表性的仓库级RAG-APR范式(Agentless、KGCompass和ExpeRepair)研究此问题。我们的协议结合Oracle定位、池内最佳K、固定接口添加上下文探针、每条件相同token填充控制和同仓库硬负样本,以及通用封装Oracle检查。Oracle定位提升了所有三种系统,但Oracle成功率仍低于50%。额外候选多样性在采样10-patch池内仍有助于,但该头寸迅速饱和。在两个固定接口下,大多数信息丰富的添加上下文条件仍优于其自身匹配对照。通用封装检查显示不同系统响应差异:在通用封装下,KGCompass和ExpeRepair的增益仍较大,而Agentless则随构建者选择变化较大。提示级融合仍留下较大的残余前沿:最佳固定探针在原生三种系统Solved@10并集之外仅解决了6个实例。总体而言,更强的定位、受限的搜索、证据质量和接口设计共同塑造仓库级修复结果。

英文摘要:

Repository-level automated program repair (APR) increasingly treats stronger localization as the main path to better repair. We ask a more targeted question: once localization is strengthened, which post-localization levers still provide recoverable gains, which are bounded within our protocol, and what residual frontier remains? We study this question on SWE-bench Lite with three representative repository-level RAG-APR paradigms, Agentless, KGCompass, and ExpeRepair. Our protocol combines Oracle Localization, within-pool Best-of-K, fixed-interface added context probes with per-condition same-token filler controls and same-repository hard negatives, and a common-wrapper oracle check. Oracle Localization improves all three systems, but Oracle success still stays below 50%. Extra candidate diversity still helps inside the sampled 10-patch pools, but that headroom saturates quickly. Under the two fixed interfaces, most informative added context conditions still outperform their own matched controls. The common-wrapper check shows different system responses: under a common wrapper, gains remain large for KGCompass and ExpeRepair, while Agentless changes more with builder choice. Prompt-level fusion still leaves a large residual frontier: the best fixed probe adds only 6 solved instances beyond the native three-system Solved@10 union. Overall, stronger localization, bounded search, evidence quality, and interface design all shape repository-level repair outcomes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑