Embrace Divergence for Richer Insights: A Multi-document Summarization Benchmark and a Case Study on Summarizing Diverse Information from News Articles

TL;DR

This paper introduces a new multi-document summarization task focused on capturing diverse information across news articles about the same event, supported by a curated dataset and analysis of LLM evaluation biases.

Contribution

It presents a novel dataset for diverse multi-document summarization, analyzes LLM evaluation biases, and studies LLM capabilities in summarizing diverse news content.

Findings

LLMs cover less than 40% of diverse information on average.

Identified biases in LLM-based evaluation metrics.

Provided best practices for automatic evaluation of diverse summarization.

Abstract

Previous research in multi-document news summarization has typically concentrated on collating information that all sources agree upon. However, the summarization of diverse information dispersed across multiple articles about an event remains underexplored. In this paper, we propose a new task of summarizing diverse information encountered in multiple news articles encompassing the same event. To facilitate this task, we outlined a data collection schema for identifying diverse information and curated a dataset named DiverseSumm. The dataset includes 245 news stories, with each story comprising 10 news articles and paired with a human-validated reference. Next, to enable consistent automatic evaluation, we conducted a comprehensive analysis to pinpoint the position and verbosity biases when utilizing Large Language Model (LLM)-based metrics for evaluating the coverage and faithfulness of summaries. Through correlation analyses, we outline the best practices for effectively using automatic LLM-based metrics on the DiverseSumm dataset. Finally, we study how LLMs summarize multiple news articles by analyzing which type of diverse information LLMs are capable of identifying. Our analyses suggest that despite the extraordinary capabilities of LLMs in single-document summarization, the proposed task remains a complex challenge for them mainly due to their limited coverage, with GPT-4 only able to cover under 40% of the diverse information on average.