用重排器偏好直接搜索文档，提升检索精度。
Beyond Sequential Reranking: Reranker-Guided Search Improves Reasoning Intensive Retrieval
- 基于重排器偏好构建近邻图，贪心搜索优先文档。
- 在100文档预算下，多个基准提升超3.5点。
- 适合资源受限但需高精度检索的场景。
广泛使用的检索-重排流水线存在两大局限：受限于初始检索的前k份文档质量，以及基于大语言模型的重排器日益增长的计算开销，限制了可有效处理的文档数量。本文提出重排器引导搜索（Reranker-Guided-Search, RGS），通过直接根据重排器偏好检索文档，突破传统串行重排的限制。该方法利用近似最近邻算法生成的邻近图进行贪心搜索，依据文档相似性策略性地优先选择待重排文档。实验结果表明，在多个基准上均取得显著性能提升：BRIGHT提升3.5点，FollowIR提升2.9点，M-BEIR提升5.1点，且所有实验均在仅100文档的重排器预算内完成。分析表明，在固定嵌入与重排器模型条件下，战略性选择待重排文档能显著提升检索精度，尤其在重排器预算有限时效果更明显。
原文摘要 · Abstract (English)
The widely used retrieve-and-rerank pipeline faces two critical limitations: they are constrained by the initial retrieval quality of the top-k documents, and the growing computational demands of LLM-based rerankers restrict the number of documents that can be effectively processed. We introduce Reranker-Guided-Search (RGS), a novel approach that bypasses these limitations by directly retrieving documents according to reranker preferences rather than following the traditional sequential reranking method. Our method uses a greedy search on proximity graphs generated by approximate nearest neighbor algorithms, strategically prioritizing promising documents for reranking based on document similarity. Experimental results demonstrate substantial performance improvements across multiple benchmarks: 3.5 points on BRIGHT, 2.9 on FollowIR, and 5.1 on M-BEIR, all within a constrained reranker budget of 100 documents. Our analysis suggests that, given a fixed pair of embedding and reranker models, strategically selecting documents to rerank can significantly improve retrieval accuracy under limited reranker budget.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。