Neural Responding Machine for Short-Text Conversation

TL;DR

This paper introduces the Neural Responding Machine, a neural network model for generating contextually appropriate short-text responses, trained on microblogging data, outperforming existing methods in grammaticality and relevance.

Contribution

The paper presents a novel encoder-decoder neural network architecture for response generation in short-text conversations, trained on large-scale microblog data.

Findings

NRM generates grammatically correct responses for over 75% of inputs

Outperforms retrieval-based and SMT-based models in response quality

Effective in producing content-wise appropriate replies

Abstract

We propose Neural Responding Machine (NRM), a neural network-based response generator for Short-Text Conversation. NRM takes the general encoder-decoder framework: it formalizes the generation of response as a decoding process based on the latent representation of the input text, while both encoding and decoding are realized with recurrent neural networks (RNN). The NRM is trained with a large amount of one-round conversation data collected from a microblogging service. Empirical study shows that NRM can generate grammatically correct and content-wise appropriate responses to over 75% of the input text, outperforming state-of-the-arts in the same setting, including retrieval-based and SMT-based models.