LLM views improve search results by adjusting scores with overlap data
XViewRank: Overlap-Conditioned Score Fusion for LLM-Augmented Sparse Retrieval
Information Retrieval
Summary
When searching for information, different versions of a question can be asked to find better answers. The authors show that combining search results from the original question and these variations using a method that considers how much the results overlap can improve ranking quality. Their approach, called XViewRank, changes how individual scores are used based on overlaps without needing complex language understanding of the text. Tests on standard search benchmarks found improvement in some cases, especially when the different question views returned quite different results.
What this means in practice
- •For search engine developers: Improve search result ranking by conditionally combining scores from multiple query versions using overlap information for sparse retrieval systems.
- •For enterprise knowledge teams: Enhance document retrieval accuracy within large internal databases by leveraging overlap-conditioned fusion of expanded query views.
Authors
Xinxuan Lv, Minghan Li, Guodong Zhou
Abstract
Large language models can reduce vocabulary mismatch in sparse retrieval by generating alternative views of a query. Once these views are searched independently, however, their ranked lists may be complementary, redundant, or affected by query drift, making fixed fusion suboptimal. We present XViewRank, a lightweight retrieval-side method for conditionally fusing BM25 results from an original query, a hypothesis view, and a step-back view. For each pooled candidate, XViewRank retains three normalized BM25 scores and three query-level overlaps between the retrieved result sets. A small nonlinear scorer uses the overlaps as context that changes how the view-specific scores are interpreted, without jointly encoding query and passage text. Under a matched-input protocol, XViewRank improves nDCG@10 and MAP numerically over reciprocal rank fusion on both TREC DL 2019 and 2020, with a statistically significant MAP gain on DL19. Additional analyses show that the clearest benefit occurs when the view rankings have low agreement, while the evidence on DL20 is weaker. The results support overlap-conditioned score fusion as a compact retrieval-side alternative to fixed fusion, while also delimiting where its advantage is empirically established.