Combining Lexical and Dense Retrieval for Computationally Efficient Multi-hop Question Answering

TL;DR

This paper proposes a hybrid lexical-dense retrieval method for multi-hop question answering that achieves competitive performance with significantly less computational resources, and provides an evaluation of dense retrieval under resource constraints.

Contribution

It introduces a hybrid retrieval approach combining lexical and dense methods, reducing computational costs while maintaining high accuracy in multi-hop QA.

Findings

Hybrid retrieval matches state-of-the-art performance

Significantly reduces computational resource requirements

Provides comprehensive evaluation under limited resources

Abstract

In simple open-domain question answering (QA), dense retrieval has become one of the standard approaches for retrieving the relevant passages to infer an answer. Recently, dense retrieval also achieved state-of-the-art results in multi-hop QA, where aggregating information from multiple pieces of information and reasoning over them is required. Despite their success, dense retrieval methods are computationally intensive, requiring multiple GPUs to train. In this work, we introduce a hybrid (lexical and dense) retrieval approach that is highly competitive with the state-of-the-art dense retrieval models, while requiring substantially less computational resources. Additionally, we provide an in-depth evaluation of dense retrieval methods on limited computational resource settings, something that is missing from the current literature.