Long Document Summarization in a Low Resource Setting using Pretrained Language Models

TL;DR

This paper presents a method for long document summarization in low-resource settings by combining GPT-2 based sentence salience detection with pretrained BART, significantly improving ROUGE-L scores on legal briefs.

Contribution

It introduces a novel GPT-2 perplexity-based salience algorithm tailored for low-resource long document summarization, enhancing BART's performance.

Findings

ROUGE-L score improved by 6.0 points using the proposed method

Salience detection aligns well with human expert labeling

Method outperforms several baseline salience detection approaches

Abstract

Abstractive summarization is the task of compressing a long document into a coherent short document while retaining salient information. Modern abstractive summarization methods are based on deep neural networks which often require large training datasets. Since collecting summarization datasets is an expensive and time-consuming task, practical industrial settings are usually low-resource. In this paper, we study a challenging low-resource setting of summarizing long legal briefs with an average source document length of 4268 words and only 120 available (document, summary) pairs. To account for data scarcity, we used a modern pretrained abstractive summarizer BART (Lewis et al., 2020), which only achieves 17.9 ROUGE-L as it struggles with long documents. We thus attempt to compress these long documents by identifying salient sentences in the source which best ground the summary, using a novel algorithm based on GPT-2 (Radford et al., 2019) language model perplexity scores, that operates within the low resource regime. On feeding the compressed documents to BART, we observe a 6.0 ROUGE-L improvement. Our method also beats several competitive salience detection baselines. Furthermore, the identified salient sentences tend to agree with an independent human labeling by domain experts.