Regularizing Dialogue Generation by Imitating Implicit Scenarios

Abstract

Human dialogues are scenario-based and appropriate responses generally relate to the latent context knowledge entailed by the specific scenario. To enable responses that are more meaningful and context-specific, we propose to improve generative dialogue systems from the scenario perspective, where both dialogue history and future conversation are taken into account to implicitly reconstruct the scenario knowledge. More importantly, the conversation scenarios are further internalized using imitation learning framework, where the conventional dialogue model that has no access to future conversations is effectively regularized by transferring the scenario knowledge contained in hierarchical supervising signals from the scenario-based dialogue model, so that the future conversation is not required in actual inference. Extensive evaluations show that our approach significantly outperforms state-of-the-art baselines on diversity and relevance, and expresses scenario-specific knowledge.

Accepted by EMNLP 2020 (long paper)

References in corpus (10)

• Distilling the Knowledge in a Neural Network
• Sequence to Sequence Learning with Neural Networks
• Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
• A Neural Conversational Model
• Layer Normalization
• Topic Aware Neural Response Generation
• Personalizing Dialogue Agents: I have a dog, do you have pets too?
• Generating Informative and Diverse Conversational Responses via Adversarial Information Maximization
• Hierarchical Recurrent Attention Network for Response Generation
• NEXUS Network: Connecting the Preceding and the Following in Dialogue Generation