Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note2 min read

PLAID-PRF: Pseudo-Relevance Feedback with Centroid-like Tokens in PLAID

Source

arXiv

arxiv.org/abs/2607.18626v1

Read the source

Source

Plain-Language Brief

Researchers propose a method called PLAID-PRF to improve the effectiveness of multi-vector dense retrieval models like ColBERT. This method uses centroid-based quantization in PLAID, a model that reduces index size and speeds up retrieval while maintaining strong effectiveness. PLAID-PRF performs pseudo-relevance feedback (PRF) on PLAID to reformulate ColBERT's query vectors based on the top-retrieved results. The method selects a small set of high-utility expansion vectors, appends them to the original query, and reruns PLAID to refine both candidate generation and final scoring. Experiments show that PLAID-PRF improves retrieval effectiveness over various baselines on standard and out-of-domain benchmarks, with less computation overhead than prior PRF methods.

What the Document Says

  • Researchers introduce PLAID-PRF, a method for improving multi-vector dense retrieval models.
  • PLAID-PRF uses centroid-based quantization of tokens' vectors in PLAID.
  • It performs pseudo-relevance feedback (PRF) on ColBERT's query vectors based on top-retrieved results.
  • The method selects high-utility expansion vectors and appends them to the original query for refinement.
  • Extensive experiments show PLAID-PRF improves retrieval effectiveness over various baselines.
  • It demonstrates less computation overhead than prior PRF methods.

Key Concepts Mentioned

  • Multi-vector dense retrieval models
  • Pseudo-Relevance Feedback (PRF)
  • Centroid-based quantization
  • PLAID

Claims Made by the Document

  • PLAID-PRF improves retrieval effectiveness over various baselines.
  • PLAID-PRF introduces less computation overhead than prior PRF methods.

Evidence Used

  • Improvements shown in experiments on MSMARCO and BEIR benchmarks.

Methods / Approach

  • Pseudo-Relevance Feedback (PRF) over PLAID

Limitations Stated by the Authors

Not stated in source

Limitations Noted by the Agent

Agent interpretation — not from the source.

The effectiveness of the method might be dependent on the specific dataset and retrieval task.

Open Questions

  • How robust is the method across different types of datasets and tasks?

Notes for Future Q&A

May include interpretation. Labeled where it does.

The text does not provide specific details about the datasets used in experiments or how the high-utility expansion vectors are selected.

Raw Abstract / Excerpt

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.