Progressive Query Expansion for Retrieval Over Cost-constrained Data Sources

TL;DR

This work tackles retrieval over cost-constrained data sources, where per-document fees can dominate overall cost. It introduces ProQE, a progressive query expansion framework that blends classic pseudo-relevance feedback with LLM-based relevance judgments in an iterative loop, applicable to both sparse and dense retrieval. The approach defines explicit update rules for expansion terms and embeddings, and demonstrates substantial improvements—up to 37% on MRR and R@1—across four datasets while reducing retrieval costs. The results underscore the practical value of cost-aware, iterative LLM-assisted expansion for retrieval-augmented systems and API-based corpora.

Abstract

Query expansion has been employed for a long time to improve the accuracy of query retrievers. Earlier works relied on pseudo-relevance feedback (PRF) techniques, which augment a query with terms extracted from documents retrieved in a first stage. However, the documents may be noisy hindering the effectiveness of the ranking. To avoid this, recent studies have instead used Large Language Models (LLMs) to generate additional content to expand a query. These techniques are prone to hallucination and also focus on the LLM usage cost. However, the cost may be dominated by the retrieval in several important practical scenarios, where the corpus is only available via APIs which charge a fee per retrieved document. We propose combining classic PRF techniques with LLMs and create a progressive query expansion algorithm ProQE that iteratively expands the query as it retrieves more documents. ProQE is compatible with both sparse and dense retrieval systems. Our experimental results on four retrieval datasets show that ProQE outperforms state-of-the-art baselines by 37% and is the most cost-effective.

Figures (2)

• Figure 1: An overview of ProQE for sparse retrieval
• Figure 2: Impact of iterations for NQ and WQ dev sets.