Lightweight Composite Re-Ranking for Efficient Keyword Search with BERT

Yang, Yingrui; Qiao, Yifan; Shao, Jinjin; Yan, Xifeng; Yang, Tao

doi:10.1145/3488560.3498495

Cited by 8 publications

(2 citation statements)

References 41 publications

(64 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our scheme uses a hybrid combination of BM25 and learned term weights, motivated by the previous work in composing lexical and neural ranking [16,22,25,26,42]. GTI adopts that for final ranking.…”

Section: Background and Related Workmentioning

confidence: 99%

Optimizing Guided Traversal for Fast Learned Sparse Retrieval

Qiao

Yang

Lin

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

Recent studies show that BM25-driven dynamic index skipping can greatly accelerate MaxScore-based document retrieval based on the learned sparse representation derived by DeepImpact. This paper investigates the effectiveness of such a traversal guidance strategy during top 𝑘 retrieval when using other models such as SPLADE and uniCOIL, and finds that unconstrained BM25-driven skipping could have a visible relevance degradation when the BM25 model is not well aligned with a learned weight model or when retrieval depth 𝑘 is small. This paper generalizes the previous work and optimizes the BM25 guided index traversal with a two-level pruning control scheme and model alignment for fast retrieval using a sparse representation. Although there can be a cost of increased latency, the proposed scheme is much faster than the original MaxScore method without BM25 guidance while retaining the relevance effectiveness. This paper analyzes the competitiveness of this two-level pruning scheme, and evaluates its tradeoff in ranking relevance and time efficiency when searching several test datasets.

show abstract

Section: Background and Related Workmentioning

confidence: 99%

Optimizing Guided Traversal for Fast Learned Sparse Retrieval

Qiao

Yang

Lin

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

show abstract

“…Our skipping and final ranking adopts a hybrid formula to bound and combine rank scores based on BM25 weights and learned term weights. That is motivated by the recent studies in composing lexical and neural models in re-ranking [43] and in combining scores from sparse retrieval and dense retrieval [15,22,23]. We choose VBMW [29] to demonstrate our scheme because VBMW is generally acknowledged to represent the state of the art [27] for many cases.…”

Section: Background and Related Workmentioning

confidence: 99%

Dual Skipping Guidance for Document Retrieval with Learned Sparse Representations

Qiao¹,

Yang²,

Lin³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

This paper proposes a dual skipping guidance scheme with hybrid scoring to accelerate document retrieval that uses learned sparse representations while still delivering a good relevance. This scheme uses both lexical BM25 and learned neural term weights to bound and compose the rank score of a candidate document separately for skipping and final ranking, and maintains two top-𝑘 thresholds during inverted index traversal. This paper evaluates time efficiency and ranking relevance of the proposed scheme in searching MS MARCO TREC datasets.

show abstract

Privacy-aware document retrieval with two-level inverted indexing

Qiao,

Ji,

Wang

et al. 2023

Inf Retrieval J

View full text Add to dashboard Cite

Previous work on privacy-aware ranking has addressed the minimization of information leakage when scoring top k documents, and has not studied on how to retrieve these top documents and their features for ranking. This paper proposes a privacy-aware document retrieval scheme with a two-level inverted index structure. In this scheme, posting records are grouped with bucket tags and runtime query processing produces query-specific tags in order to gather encoded features of matched documents with a privacy protection during index traversal. To thwart leakage-abuse attacks, our design minimizes the chance that a server processes unauthorized queries or identifies document sharing across posting lists through index inspection or across-query association. This paper presents the evaluation and analytic results of the proposed scheme to demonstrate the tradeoffs in its design considerations for privacy, efficiency, and relevance.

show abstract

Lightweight Composite Re-Ranking for Efficient Keyword Search with BERT

Cited by 8 publications

References 41 publications

Optimizing Guided Traversal for Fast Learned Sparse Retrieval

Optimizing Guided Traversal for Fast Learned Sparse Retrieval

Dual Skipping Guidance for Document Retrieval with Learned Sparse Representations

Privacy-aware document retrieval with two-level inverted indexing

Contact Info

Product

Resources

About