Improving Neural Retrieval with Attribution-Guided Query Rewriting

2026-02-12

AI Summary

This work addresses the challenge that neural retrieval models often fail to recall relevant documents when faced with ambiguous or underspecified queries. To mitigate this issue, the authors propose a query rewriting method that leverages token-level attribution scores from a neural retriever to guide a large language model via structured prompts, enabling it to clarify weak or misleading components of the query while preserving the original intent. Notably, this is the first approach to incorporate the retriever’s attribution-based explanations into a feedback loop, thereby jointly optimizing interpretability and retrieval effectiveness. Experimental results on the BEIR benchmark demonstrate that the proposed method significantly outperforms strong baselines, with particularly pronounced gains in scenarios involving implicit or vague information needs.

Technology Category

• Natural Language Processing: Question Answering
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Neural retrievers are effective but brittle: underspecified or ambiguous queries can misdirect ranking even when relevant documents exist. Existing approaches address this brittleness only partially: LLMs rewrite queries without retriever feedback, and explainability methods identify misleading tokens but are used for post-hoc analysis. We close this loop and propose an attribution-guided query rewriting method that uses token-level explanations to guide query rewriting. For each query, we compute gradient-based token attributions from the retriever and then use these scores as soft guidance in a structured prompt to an LLM that clarifies weak or misleading query components while preserving intent. Evaluated on BEIR collections, the resulting rewrites consistently improve retrieval effectiveness over strong baselines, with larger gains for implicit or ambiguous information needs.

Problem

• neural retrieval
• query ambiguity
• retriever brittleness
• query rewriting
• attribution

Innovation

• attribution-guided
• query rewriting
• neural retrieval
• token-level attribution
• retriever feedback