TSTR: Too Short to Represent, Summarize with Details! Intro-Guided Extended Summary Generation

TL;DR

This paper introduces TSTR, an extractive summarization method that leverages introductory information to generate detailed extended summaries of scientific papers, outperforming existing models in both automatic and human evaluations.

Contribution

The paper presents TSTR, a novel extractive summarizer that uses intro-guided pointers to improve long scientific summary generation, addressing limitations of previous short-summary-focused methods.

Findings

TSTR achieves significant improvements in Rouge scores over baselines.

Human evaluations favor TSTR for cohesion and completeness.

Effective in generating detailed extended summaries.

Abstract

Many scientific papers such as those in arXiv and PubMed data collections have abstracts with varying lengths of 50-1000 words and average length of approximately 200 words, where longer abstracts typically convey more information about the source paper. Up to recently, scientific summarization research has typically focused on generating short, abstract-like summaries following the existing datasets used for scientific summarization. In domains where the source text is relatively long-form, such as in scientific documents, such summary is not able to go beyond the general and coarse overview and provide salient information from the source document. The recent interest to tackle this problem motivated curation of scientific datasets, arXiv-Long and PubMed-Long, containing human-written summaries of 400-600 words, hence, providing a venue for research in generating long/extended summaries. Extended summaries facilitate a faster read while providing details beyond coarse information. In this paper, we propose TSTR, an extractive summarizer that utilizes the introductory information of documents as pointers to their salient information. The evaluations on two existing large-scale extended summarization datasets indicate statistically significant improvement in terms of Rouge and average Rouge (F1) scores (except in one case) as compared to strong baselines and state-of-the-art. Comprehensive human evaluations favor our generated extended summaries in terms of cohesion and completeness.