Quam: Adaptive Retrieval through Query Affinity Modelling

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Authors

  • Mandeep Rathee
  • Sean MacAvaney
  • Avishek Anand

Research Organisations

External Research Organisations

  • University of Glasgow
  • Delft University of Technology
View graph of relations

Details

Original languageEnglish
Title of host publicationProceedings of the 18th ACM International Conference on Web Search and Data Mining
Subtitle of host publicationWSDM 2025
Pages954-962
Number of pages9
ISBN (electronic)9798400713293
Publication statusPublished - 10 Mar 2025
Event18th ACM International Conference on Web Search and Data Mining, WSDM 2025 - Hannover, Germany
Duration: 10 Mar 202514 Mar 2025

Publication series

NameProceedings of the Eighteenth ACM International Conference on Web Search and Data Mining

Abstract

A central task in information retrieval and the NLP communities is relevance modeling, which aims to rank documents based on their expressed information needs Many knowledge-intensive retrieval tasks are powered by a first-stage retrieval stage for context selection, followed by a more involved task-specific model. However, using this filtering (cascading) approach inherently limits the recall of subsequent stages. Recently, adaptive re-ranking techniques have been proposed to overcome this issue by continually selecting documents from the whole corpus, rather than only considering an initial pool of documents. However, so far these approaches have been limited to heuristic design choices, particularly in terms of the criteria for document selection. In this work, we propose a unifying view of the nascent area of adaptive retrieval by proposing Quam, a query-affinity model of adaptive re-ranking that includes two complementary components: (1) a more principled algorithm for document selection, and (2) a data-driven approach to model document co-relevance during indexing. Our extensive experimental evidence shows that our proposed approach improves the recall performance by up to 26% over the standard re-ranking baselines. Further, the query affinity modelling and relevance-aware document graph components can be injected into any adaptive retrieval approach. The experimental results show the existing adaptive retrieval approach improves recall by up to 12%.

Keywords

    adaptive retrieval, clustering hypothesis, neural re-ranking

ASJC Scopus subject areas

Cite this

Quam: Adaptive Retrieval through Query Affinity Modelling. / Rathee, Mandeep; MacAvaney, Sean; Anand, Avishek.
Proceedings of the 18th ACM International Conference on Web Search and Data Mining: WSDM 2025. 2025. p. 954-962 (Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining).

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Rathee, M, MacAvaney, S & Anand, A 2025, Quam: Adaptive Retrieval through Query Affinity Modelling. in Proceedings of the 18th ACM International Conference on Web Search and Data Mining: WSDM 2025. Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining, pp. 954-962, 18th ACM International Conference on Web Search and Data Mining, WSDM 2025, Hannover, Lower Saxony, Germany, 10 Mar 2025. https://doi.org/10.1145/3701551.3703584, https://doi.org/10.48550/arXiv.2410.20286
Rathee, M., MacAvaney, S., & Anand, A. (2025). Quam: Adaptive Retrieval through Query Affinity Modelling. In Proceedings of the 18th ACM International Conference on Web Search and Data Mining: WSDM 2025 (pp. 954-962). (Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining). https://doi.org/10.1145/3701551.3703584, https://doi.org/10.48550/arXiv.2410.20286
Rathee M, MacAvaney S, Anand A. Quam: Adaptive Retrieval through Query Affinity Modelling. In Proceedings of the 18th ACM International Conference on Web Search and Data Mining: WSDM 2025. 2025. p. 954-962. (Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining). doi: 10.1145/3701551.3703584, 10.48550/arXiv.2410.20286
Rathee, Mandeep ; MacAvaney, Sean ; Anand, Avishek. / Quam : Adaptive Retrieval through Query Affinity Modelling. Proceedings of the 18th ACM International Conference on Web Search and Data Mining: WSDM 2025. 2025. pp. 954-962 (Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining).
Download
@inproceedings{92cfec2b335447b2bd4cd6977cf1f37f,
title = "Quam: Adaptive Retrieval through Query Affinity Modelling",
abstract = "A central task in information retrieval and the NLP communities is relevance modeling, which aims to rank documents based on their expressed information needs Many knowledge-intensive retrieval tasks are powered by a first-stage retrieval stage for context selection, followed by a more involved task-specific model. However, using this filtering (cascading) approach inherently limits the recall of subsequent stages. Recently, adaptive re-ranking techniques have been proposed to overcome this issue by continually selecting documents from the whole corpus, rather than only considering an initial pool of documents. However, so far these approaches have been limited to heuristic design choices, particularly in terms of the criteria for document selection. In this work, we propose a unifying view of the nascent area of adaptive retrieval by proposing Quam, a query-affinity model of adaptive re-ranking that includes two complementary components: (1) a more principled algorithm for document selection, and (2) a data-driven approach to model document co-relevance during indexing. Our extensive experimental evidence shows that our proposed approach improves the recall performance by up to 26% over the standard re-ranking baselines. Further, the query affinity modelling and relevance-aware document graph components can be injected into any adaptive retrieval approach. The experimental results show the existing adaptive retrieval approach improves recall by up to 12%.",
keywords = "adaptive retrieval, clustering hypothesis, neural re-ranking",
author = "Mandeep Rathee and Sean MacAvaney and Avishek Anand",
note = "Publisher Copyright: {\textcopyright} 2025 Copyright held by the owner/author(s).; 18th ACM International Conference on Web Search and Data Mining, WSDM 2025, WSDM 2025 ; Conference date: 10-03-2025 Through 14-03-2025",
year = "2025",
month = mar,
day = "10",
doi = "10.1145/3701551.3703584",
language = "English",
series = "Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining",
pages = "954--962",
booktitle = "Proceedings of the 18th ACM International Conference on Web Search and Data Mining",

}

Download

TY - GEN

T1 - Quam

T2 - 18th ACM International Conference on Web Search and Data Mining, WSDM 2025

AU - Rathee, Mandeep

AU - MacAvaney, Sean

AU - Anand, Avishek

N1 - Publisher Copyright: © 2025 Copyright held by the owner/author(s).

PY - 2025/3/10

Y1 - 2025/3/10

N2 - A central task in information retrieval and the NLP communities is relevance modeling, which aims to rank documents based on their expressed information needs Many knowledge-intensive retrieval tasks are powered by a first-stage retrieval stage for context selection, followed by a more involved task-specific model. However, using this filtering (cascading) approach inherently limits the recall of subsequent stages. Recently, adaptive re-ranking techniques have been proposed to overcome this issue by continually selecting documents from the whole corpus, rather than only considering an initial pool of documents. However, so far these approaches have been limited to heuristic design choices, particularly in terms of the criteria for document selection. In this work, we propose a unifying view of the nascent area of adaptive retrieval by proposing Quam, a query-affinity model of adaptive re-ranking that includes two complementary components: (1) a more principled algorithm for document selection, and (2) a data-driven approach to model document co-relevance during indexing. Our extensive experimental evidence shows that our proposed approach improves the recall performance by up to 26% over the standard re-ranking baselines. Further, the query affinity modelling and relevance-aware document graph components can be injected into any adaptive retrieval approach. The experimental results show the existing adaptive retrieval approach improves recall by up to 12%.

AB - A central task in information retrieval and the NLP communities is relevance modeling, which aims to rank documents based on their expressed information needs Many knowledge-intensive retrieval tasks are powered by a first-stage retrieval stage for context selection, followed by a more involved task-specific model. However, using this filtering (cascading) approach inherently limits the recall of subsequent stages. Recently, adaptive re-ranking techniques have been proposed to overcome this issue by continually selecting documents from the whole corpus, rather than only considering an initial pool of documents. However, so far these approaches have been limited to heuristic design choices, particularly in terms of the criteria for document selection. In this work, we propose a unifying view of the nascent area of adaptive retrieval by proposing Quam, a query-affinity model of adaptive re-ranking that includes two complementary components: (1) a more principled algorithm for document selection, and (2) a data-driven approach to model document co-relevance during indexing. Our extensive experimental evidence shows that our proposed approach improves the recall performance by up to 26% over the standard re-ranking baselines. Further, the query affinity modelling and relevance-aware document graph components can be injected into any adaptive retrieval approach. The experimental results show the existing adaptive retrieval approach improves recall by up to 12%.

KW - adaptive retrieval

KW - clustering hypothesis

KW - neural re-ranking

UR - http://www.scopus.com/inward/record.url?scp=105001669849&partnerID=8YFLogxK

U2 - 10.1145/3701551.3703584

DO - 10.1145/3701551.3703584

M3 - Conference contribution

AN - SCOPUS:105001669849

T3 - Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining

SP - 954

EP - 962

BT - Proceedings of the 18th ACM International Conference on Web Search and Data Mining

Y2 - 10 March 2025 through 14 March 2025

ER -