Abstractive Headline Generation for Spoken Content by Attentive Recurrent Neural Networks with ASR Error Modeling

TL;DR

This paper introduces an attentive RNN model with integrated ASR error modeling to generate headlines from spoken content, effectively leveraging abundant text data despite ASR errors.

Contribution

It presents a novel approach combining ASR error modeling with attentive RNNs for abstractive headline generation from spoken content, reducing the need for large spoken data.

Findings

The proposed model performs well even with different ASR recognizers.

ASR error modeling improves headline generation accuracy.

The approach enables training from text data alone.

Abstract

Headline generation for spoken content is important since spoken content is difficult to be shown on the screen and browsed by the user. It is a special type of abstractive summarization, for which the summaries are generated word by word from scratch without using any part of the original content. Many deep learning approaches for headline generation from text document have been proposed recently, all requiring huge quantities of training data, which is difficult for spoken document summarization. In this paper, we propose an ASR error modeling approach to learn the underlying structure of ASR error patterns and incorporate this model in an Attentive Recurrent Neural Network (ARNN) architecture. In this way, the model for abstractive headline generation for spoken content can be learned from abundant text data and the ASR data for some recognizers. Experiments showed very encouraging results and verified that the proposed ASR error model works well even when the input spoken content is recognized by a recognizer very different from the one the model learned from.