How "Multi" is Multi-Document Summarization?

TL;DR

This paper introduces an automated metric to measure how much a multi-document summary relies on multiple sources, revealing that many datasets and systems often generate summaries from single documents rather than combining dispersed information.

Contribution

The paper proposes a novel metric to quantify the dispersion of source information in multi-document summaries and applies it to analyze existing datasets and systems.

Findings

Many datasets contain summaries that are mostly covered by a single document.

State-of-the-art systems often do not effectively combine information from multiple sources.

The proposed metric can guide the development of more multi-source-aware summarization models.

Abstract

The task of multi-document summarization (MDS) aims at models that, given multiple documents as input, are able to generate a summary that combines disperse information, originally spread across these documents. Accordingly, it is expected that both reference summaries in MDS datasets, as well as system summaries, would indeed be based on such dispersed information. In this paper, we argue for quantifying and assessing this expectation. To that end, we propose an automated measure for evaluating the degree to which a summary is ``disperse'', in the sense of the number of source documents needed to cover its content. We apply our measure to empirically analyze several popular MDS datasets, with respect to their reference summaries, as well as the output of state-of-the-art systems. Our results show that certain MDS datasets barely require combining information from multiple documents, where a single document often covers the full summary content. Overall, we advocate using our metric for assessing and improving the degree to which summarization datasets require combining multi-document information, and similarly how summarization models actually meet this challenge. Our code is available in https://github.com/ariecattan/multi_mds.