Sequence to Backward and Forward Sequences: A Content-Introducing Approach to Generative Short-Text Conversation

Abstract

Using neural networks to generate replies in human-computer dialogue systems is attracting increasing attention over the past few years. However, the performance is not satisfactory: the neural network tends to generate safe, universally relevant replies which carry little meaning. In this paper, we propose a content-introducing approach to neural network-based generative dialogue systems. We first use pointwise mutual information (PMI) to predict a noun as a keyword, reflecting the main gist of the reply. We then propose seq2BF, a "sequence to backward and forward sequences" model, which generates a reply containing the given keyword. Experimental results show that our approach significantly outperforms traditional sequence-to-sequence models in terms of human evaluation and the entropy measure, and that the predicted keyword can appear at an appropriate position in the reply.

Accepted by COLING

Cited by in corpus (14)

• Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory
• Plan-And-Write: Towards Better Automatic Storytelling
• A Multi-Modal Chinese Poetry Generation Model
• A Constrained Sequence-to-Sequence Neural Model for Sentence Simplification
• Correct-and-Memorize: Learning to Translate from Interactive Revisions
• Chat More If You Like: Dynamic Cue Words Planning to Flow Longer Conversations
• A Syntactically Constrained Bidirectional-Asynchronous Approach for Emotional Conversation Generation
• Why Do Neural Response Generation Models Prefer Universal Replies?
• The World is Not Binary: Learning to Rank with Grayscale Data for Dialogue Response Selection
• Affective Neural Response Generation
• Domain Agnostic Real-Valued Specificity Prediction
• Review-Driven Answer Generation for Product-Related Questions in E-Commerce
• Posterior-GAN: Towards Informative and Coherent Response Generation with Posterior Generative Adversarial Network
• Improving Matching Models with Hierarchical Contextualized Representations for Multi-turn Response Selection