MoR: Better Handling Diverse Queries with a Mixture of Sparse, Dense, and Human Retrievers

TL;DR

This paper introduces a mixture of retrievers approach that dynamically combines sparse, dense, and human sources to improve retrieval-augmented generation, outperforming individual retrievers and larger models.

Contribution

It proposes a zero-shot, weighted mixture of heterogeneous retrievers that adaptively integrates multiple sources for better information retrieval in RAG systems.

Findings

Mixture of retrievers outperforms individual retrievers and larger models.

The approach achieves +10.8% and +3.9% improvements over single retrievers and 7B models.

Incorporating human sources yields a 58.9% relative performance boost.

Abstract

Retrieval-augmented Generation (RAG) is powerful, but its effectiveness hinges on which retrievers we use and how. Different retrievers offer distinct, often complementary signals: BM25 captures lexical matches; dense retrievers, semantic similarity. Yet in practice, we typically fix a single retriever based on heuristics, which fails to generalize across diverse information needs. Can we dynamically select and integrate multiple retrievers for each individual query, without the need for manual selection? In our work, we validate this intuition with quantitative analysis and introduce mixture of retrievers: a zero-shot, weighted combination of heterogeneous retrievers. Extensive experiments show that such mixtures are effective and efficient: Despite totaling just 0.8B parameters, this mixture outperforms every individual retriever and even larger 7B models by +10.8% and +3.9% on average, respectively. Further analysis also shows that this mixture framework can help incorporate specialized non-oracle human information sources as retrievers to achieve good collaboration, with a 58.9% relative performance improvement over simulated humans alone.