Generative and Pseudo-Relevant Feedback for Sparse, Dense and Learned Sparse Retrieval

TL;DR

This paper explores the use of generative and pseudo-relevance feedback techniques to improve various types of document retrieval, demonstrating that combining both methods enhances recall significantly.

Contribution

It extends generative relevance feedback to dense and learned sparse retrieval and proposes a combined approach that leverages the strengths of both feedback types.

Findings

GRF improves over PRF by around 10% in precision and recall.

GRF provides external context not captured by PRF.

Combining GRF and PRF significantly increases recall in most experiments.

Abstract

Pseudo-relevance feedback (PRF) is a classical approach to address lexical mismatch by enriching the query using first-pass retrieval. Moreover, recent work on generative-relevance feedback (GRF) shows that query expansion models using text generated from large language models can improve sparse retrieval without depending on first-pass retrieval effectiveness. This work extends GRF to dense and learned sparse retrieval paradigms with experiments over six standard document ranking benchmarks. We find that GRF improves over comparable PRF techniques by around 10% on both precision and recall-oriented measures. Nonetheless, query analysis shows that GRF and PRF have contrasting benefits, with GRF providing external context not present in first-pass retrieval, whereas PRF grounds the query to the information contained within the target corpus. Thus, we propose combining generative and pseudo-relevance feedback ranking signals to achieve the benefits of both feedback classes, which significantly increases recall over PRF methods on 95% of experiments.