Newsroom: A Dataset of 1.3 Million Summaries with Diverse Extractive Strategies

TL;DR

NEWSROOM is a large, diverse dataset of 1.3 million news summaries from 38 major outlets, capturing various extractive and abstractive summarization styles, useful for advancing summarization research.

Contribution

The paper introduces NEWSROOM, a new large-scale dataset with diverse summarization styles, enabling better analysis and development of summarization methods.

Findings

Summaries exhibit high diversity in extractive and abstractive strategies.

The dataset is challenging for existing summarization models.

Analysis shows the dataset's potential to improve summarization techniques.

Abstract

We present NEWSROOM, a summarization dataset of 1.3 million articles and summaries written by authors and editors in newsrooms of 38 major news publications. Extracted from search and social media metadata between 1998 and 2017, these high-quality summaries demonstrate high diversity of summarization styles. In particular, the summaries combine abstractive and extractive strategies, borrowing words and phrases from articles at varying rates. We analyze the extraction strategies used in NEWSROOM summaries against other datasets to quantify the diversity and difficulty of our new data, and train existing methods on the data to evaluate its utility and challenges.