Multi-Step Reasoning Over Unstructured Text with Beam Dense Retrieval

TL;DR

This paper introduces BeamDR, a multi-step dense retrieval method that constructs evidence chains for complex question answering without relying on semi-structured data, demonstrating competitive performance.

Contribution

BeamDR is a novel multi-step dense retrieval approach that uses beam search to form reasoning chains in unstructured text, advancing multi-hop question answering.

Findings

Competitive with state-of-the-art systems

Effective in capturing implicit evidence relationships

Operates without semi-structured information

Abstract

Complex question answering often requires finding a reasoning chain that consists of multiple evidence pieces. Current approaches incorporate the strengths of structured knowledge and unstructured text, assuming text corpora is semi-structured. Building on dense retrieval methods, we propose a new multi-step retrieval approach (BeamDR) that iteratively forms an evidence chain through beam search in dense representations. When evaluated on multi-hop question answering, BeamDR is competitive to state-of-the-art systems, without using any semi-structured information. Through query composition in dense space, BeamDR captures the implicit relationships between evidence in the reasoning chain. The code is available at https://github.com/ henryzhao5852/BeamDR.