Joint Passage Ranking for Diverse Multi-Answer Retrieval

TL;DR

This paper introduces JPR, a novel joint passage retrieval model that effectively covers multiple distinct answers for a question by modeling passage diversity, significantly improving answer coverage and enabling more efficient downstream question answering.

Contribution

The paper presents JPR, the first autoregressive joint passage retrieval model for multi-answer retrieval that explicitly promotes answer diversity during passage selection.

Findings

JPR outperforms prior methods in answer coverage across three datasets.

Combining JPR with question answering models improves answer generation efficiency.

JPR achieves state-of-the-art results in multi-answer retrieval tasks.

Abstract

We study multi-answer retrieval, an under-explored problem that requires retrieving passages to cover multiple distinct answers for a given question. This task requires joint modeling of retrieved passages, as models should not repeatedly retrieve passages containing the same answer at the cost of missing a different valid answer. In this paper, we introduce JPR, the first joint passage retrieval model for multi-answer retrieval. JPR makes use of an autoregressive reranker that selects a sequence of passages, each conditioned on previously selected passages. JPR is trained to select passages that cover new answers at each timestep and uses a tree-decoding algorithm to enable flexibility in the degree of diversity. Compared to prior approaches, JPR achieves significantly better answer coverage on three multi-answer datasets. When combined with downstream question answering, the improved retrieval enables larger answer generation models since they need to consider fewer passages, establishing a new state-of-the-art.