发表机构
University of International Business and Economics; University of Glasgow（国际经济贸易大学; 格拉斯哥大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究在PLAID基础上提出PLAID-PRF方法，通过对顶部检索结果执行伪相关反馈来重新制定查询向量，利用质心向量降低计算成本。实验表明该方法能有效提升检索效果，相比PLAID有显著改进，且计算开销小，实现高效反馈感知后期交互检索。
AI 中文摘要
多向量密集检索模型，如ColBERT，通过对查询和文档之间的细粒度令牌级交互进行建模，实现了强大的检索效果。PLAID等方法使用基于质心的每个令牌向量量化来减小索引大小并加快检索速度，同时保持强大的有效性。在这项工作中，我们引入了PLAID-PRF，一种对PLAID执行伪相关反馈（PRF）的方法，以基于顶部检索结果重新制定ColBERT的查询向量。与在多向量检索模型上执行PRF的现有方法相比，PLAID-PRF通过利用内部PLAID质心向量保持计算成本低，将它们与传统PRF方法中的令牌类似对待。该方法选择一小套多样化的高效用扩展向量并将它们附加到原始查询中，重新运行PLAID以优化候选生成和最终评分。在标准的域内MSMARCO和四个域外BEIR基准上进行的广泛实验表明，PLAID-PRF始终比各种基线提高检索效果。特别是，PLAID-PRF比PLAID在nDCG@10上提高了4.3%，在MRR@10上提高了7.3%，同时引入的计算开销比以前的PRF方法少得多。结果表明，我们提出的质心感知PRF方法提供了一种有效且轻量级的机制来提高顶级检索结果的质量。总体而言，这项工作实现了有效且高效的反馈感知后期交互检索，而无需昂贵的查询时文档-令牌聚类。
英文摘要
Multi-vector dense retrieval models, such as ColBERT, achieve strong retrieval effectiveness by modelling fine-grained token-level interactions between queries and documents. Methods such as PLAID use centroid-based quantisation of each token's vector to reduce the index size and speed up retrieval while maintaining strong effectiveness. In this work, we introduce PLAID-PRF, a method that performs Pseudo-Relevance Feedback (PRF) over PLAID to reformulate ColBERT's query vectors based on the top-retrieved results. In contrast with prior methods that perform PRF on multi-vector retrieval models, PLAID-PRF keeps computational costs low by leveraging the internal PLAID centroid vectors, treating them similarly to tokens in traditional PRF methods. The method selects a small and diverse set of high-utility expansion vectors and appends them to the original query, rerunning PLAID to refine both candidate generation and final scoring. Extensive experiments on the standard in-domain MSMARCO and four out-of-domain BEIR benchmarks show that PLAID-PRF consistently improves retrieval effectiveness over various baselines. In particular, PLAID-PRF improves over PLAID by up to 4.3% nDCG@10 and 7.3% MRR@10, while introducing substantially less computation overhead than prior PRF methods. The results demonstrate that our proposed centroid-aware PRF method offers an effective and lightweight mechanism to improve the quality of top-ranked retrieved results. Overall, this work enables effective and efficient feedback-aware late-interaction retrieval without expensive query-time document-token clustering.
CommentsSIGIR 2026
