Towards a Neural Network Approach to Abstractive Multi-Document Summarization

TL;DR

This paper explores adapting neural abstractive summarization models, successful in single-document tasks, to multi-document summarization by fine-tuning on limited multi-document data, showing improved performance on benchmark datasets.

Contribution

It introduces a method to extend neural abstractive models from single to multi-document summarization using minimal multi-document training data.

Findings

Outperforms baseline neural models on DUC datasets

Requires only small amounts of multi-document summaries for fine-tuning

Demonstrates effectiveness of transfer learning in MDS

Abstract

Till now, neural abstractive summarization methods have achieved great success for single document summarization (SDS). However, due to the lack of large scale multi-document summaries, such methods can be hardly applied to multi-document summarization (MDS). In this paper, we investigate neural abstractive methods for MDS by adapting a state-of-the-art neural abstractive summarization model for SDS. We propose an approach to extend the neural abstractive model trained on large scale SDS data to the MDS task. Our approach only makes use of a small number of multi-document summaries for fine tuning. Experimental results on two benchmark DUC datasets demonstrate that our approach can outperform a variety of baseline neural models.