Reranking Passages with Coarse-to-Fine Neural Retriever Enhanced by List-Context Information

TL;DR

This paper introduces a coarse-to-fine neural reranker that uses list-context attention to incorporate multiple candidate passages, improving relevance ranking by considering contextual information efficiently.

Contribution

It proposes a novel list-context attention mechanism and a cache policy learning algorithm to encode large candidate sets in a single pass, integrating coarse and fine ranking in joint optimization.

Findings

Improved passage reranking accuracy over baseline models

Efficient encoding of large candidate lists with the C2F approach

Joint optimization enhances relevance scoring

Abstract

Passage reranking is a critical task in various applications, particularly when dealing with large volumes of documents. Existing neural architectures have limitations in retrieving the most relevant passage for a given question because the semantics of the segmented passages are often incomplete, and they typically match the question to each passage individually, rarely considering contextual information from other passages that could provide comparative and reference information. This paper presents a list-context attention mechanism to augment the passage representation by incorporating the list-context information from other candidates. The proposed coarse-to-fine (C2F) neural retriever addresses the out-of-memory limitation of the passage attention mechanism by dividing the list-context modeling process into two sub-processes with a cache policy learning algorithm, enabling the efficient encoding of context information from a large number of candidate answers. This method can be generally used to encode context information from any number of candidate answers in one pass. Different from most multi-stage information retrieval architectures, this model integrates the coarse and fine rankers into the joint optimization process, allowing for feedback between the two layers to update the model simultaneously. Experiments demonstrate the effectiveness of the proposed approach.