arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2611.09040cs.IRcs.DB
收藏

MV-IVF:基于多向量聚类的多向量检索

MV-IVF: Multi-Vector Retrieval via Multi-Vector Clustering

University of Maryland(马里兰大学);Google Research(谷歌研究)

Kishen N Gowda, Laxman Dhulipala, Lars Gottesbüren, Rajesh Jayaram

AI总结:

本文提出MV-IVF多向量检索索引,基于MV-Lloyd算法构建,通过优化搜索与量化加速,在BEIR、LoTTE基准上实现比基线更优的延迟、吞吐量与规模特性。

摘要:

多向量检索将每个查询和文档表示为一组嵌入,并通过细粒度的集合间交互(如MaxSim即Chamfer相似度)对相关性进行评分。这种表达能力相比单向量表示提升了检索质量,但也大幅增加了搜索成本。现有系统主要通过代理表示或过滤-重排序流水线来应对这一挑战,将候选生成简化为单向量搜索。本文研究了互补方向:直接在多向量数据上构建原生IVF风格的索引。我们提出MV-IVF,这是一种基于MV-Lloyd的多向量检索倒排文件索引;MV-Lloyd是一种用于在Chamfer目标下对点云进行聚类的新型Lloyd风格算法。我们的关键观察是,一旦点云被分配到聚类中,中心更新步骤就简化为对每个聚类中展平嵌入的标准单向量聚类。这使得能够高效构建分层IVF树,其中内部节点为中心点点云。在查询时,MV-IVF对树执行最佳优先束搜索以识别有前景的叶聚类,然后对这些叶中的文档进行评分。我们进一步使用低位量化、查询压缩和缓存高效的批量搜索来加速搜索。我们在MVSIC中实现了MV-IVF,这是一个用于多向量检索的优化C++库和统一评估框架,包含多种主要方法。在BEIR和LoTTE文本检索基准上,MV-IVF实现了出色的延迟-召回率权衡、最佳的批量吞吐量前沿,以及良好的构建时间和索引大小特性。在90% Recall@10时,MV-IVF的单查询延迟比基线低达8.03倍,QPS比基线高达173.5倍,证明原生IVF风格索引对于可扩展的多向量检索是实用且有效的。

英文摘要:

Multi-vector retrieval represents each query and document as a set of embeddings and scores relevance via fine-grained set-to-set interactions such as MaxSim (Chamfer similarity). This expressiveness improves retrieval quality over single-vector representations, but makes search substantially more expensive. Existing systems address this challenge primarily through proxy representations or filter-and-rerank pipelines that reduce candidate generation to single-vector search. We study the complementary direction of building a native IVF-style index directly over multi-vector data. We introduce MV-IVF, an inverted-file index for multi-vector retrieval built on MV-Lloyd, a new Lloyd-style algorithm for clustering point clouds under a Chamfer objective. Our key observation is that once point clouds are assigned to clusters, the center-update step reduces to standard single-vector clustering over the flattened embeddings in each cluster. This enables efficient construction of a hierarchical IVF tree with center point clouds at internal nodes. At query time, MV-IVF performs best-first beam search over the tree to identify promising leaf clusters, then scores the documents in those leaves. We further accelerate search using low-bit quantization, query compression, and cache-efficient batched search. We implement MV-IVF in MVSIC, an optimized C++ library and unified evaluation framework for multi-vector retrieval that includes several major approaches. Across BEIR and LoTTE text retrieval benchmarks, MV-IVF achieves strong latency--recall tradeoffs, the best batched throughput frontiers, and favorable build-time and index-size profiles. At $90\%$ Recall-$10$@$10$, MV-IVF achieves up to $8.03\times$ lower single-query latency and up to $173.5\times$ higher QPS than the baselines, demonstrating that native IVF-style indexing is practical and effective for scalable multi-vector retrieval.

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。

↑