The Summary Loop: Learning to Write Abstractive Summaries Without Examples

TL;DR

This paper introduces an unsupervised abstractive summarization method that maximizes coverage and fluency, effectively generating concise, high-quality summaries without requiring labeled training data.

Contribution

It proposes a novel unsupervised training approach that encourages key term inclusion and achieves competitive results with less copied content and sentence compression.

Findings

Outperforms previous unsupervised methods by over 2 R-1 points

Approaches the performance of supervised methods

Produces more abstract summaries with shorter copied passages

Abstract

This work presents a new approach to unsupervised abstractive summarization based on maximizing a combination of coverage and fluency for a given length constraint. It introduces a novel method that encourages the inclusion of key terms from the original document into the summary: key terms are masked out of the original document and must be filled in by a coverage model using the current generated summary. A novel unsupervised training procedure leverages this coverage model along with a fluency model to generate and score summaries. When tested on popular news summarization datasets, the method outperforms previous unsupervised methods by more than 2 R-1 points, and approaches results of competitive supervised methods. Our model attains higher levels of abstraction with copied passages roughly two times shorter than prior work, and learns to compress and merge sentences without supervision.