A Latent Variable Recurrent Neural Network for Discourse Relation Language Models

TL;DR

This paper introduces a latent variable recurrent neural network that jointly models word sequences and discourse relations, improving performance on discourse classification and language modeling tasks.

Contribution

It proposes a novel architecture that integrates latent discourse relations into RNNs, enhancing both discourse relation classification and language modeling.

Findings

Outperforms state-of-the-art in discourse relation classification

Improves language modeling by marginalizing over discourse relations

Effective in dialog act classification

Abstract

This paper presents a novel latent variable recurrent neural network architecture for jointly modeling sequences of words and (possibly latent) discourse relations between adjacent sentences. A recurrent neural network generates individual words, thus reaping the benefits of discriminatively-trained vector representations. The discourse relations are represented with a latent variable, which can be predicted or marginalized, depending on the task. The resulting model can therefore employ a training objective that includes not only discourse relation classification, but also word prediction. As a result, it outperforms state-of-the-art alternatives for two tasks: implicit discourse relation classification in the Penn Discourse Treebank, and dialog act classification in the Switchboard corpus. Furthermore, by marginalizing over latent discourse relations at test time, we obtain a discourse informed language model, which improves over a strong LSTM baseline.