Deep Recurrent Generative Decoder for Abstractive Text Summarization

TL;DR

This paper introduces a deep recurrent generative decoder framework for abstractive text summarization that leverages latent structure learning and neural variational inference to produce higher quality summaries, outperforming existing methods.

Contribution

It presents a novel DRGN model integrating latent structure learning with variational inference for improved abstractive summarization.

Findings

DRGN outperforms state-of-the-art methods on benchmark datasets.

Latent structure learning enhances summary quality.

Neural variational inference effectively handles intractable posteriors.

Abstract

We propose a new framework for abstractive text summarization based on a sequence-to-sequence oriented encoder-decoder model equipped with a deep recurrent generative decoder (DRGN). Latent structure information implied in the target summaries is learned based on a recurrent latent random model for improving the summarization quality. Neural variational inference is employed to address the intractable posterior inference for the recurrent latent variables. Abstractive summaries are generated based on both the generative latent variables and the discriminative deterministic states. Extensive experiments on some benchmark datasets in different languages show that DRGN achieves improvements over the state-of-the-art methods.