MALA: Cross-Domain Dialogue Generation with Action Learning

Abstract

Response generation for task-oriented dialogues involves two basic components: dialogue planning and surface realization. These two components, however, have a discrepancy in their objectives, i.e., task completion and language quality. To deal with such discrepancy, conditioned response generation has been introduced where the generation process is factorized into action decision and language generation via explicit action representations. To obtain action representations, recent studies learn latent actions in an unsupervised manner based on the utterance lexical similarity. Such an action learning approach is prone to diversities of language surfaces, which may impinge task completion and language quality. To address this issue, we propose multi-stage adaptive latent action learning (MALA) that learns semantic latent actions by distinguishing the effects of utterances on dialogue progress. We model the utterance effect using the transition of dialogue states caused by the utterance and develop a semantic similarity measurement that estimates whether utterances have similar effects. For learning semantic actions on domains without dialogue states, MsALA extends the semantic similarity measurement across domains progressively, i.e., from aligning shared actions to learning domain-specific actions. Experiments using multi-domain datasets, SMD and MultiWOZ, show that our proposed model achieves consistent improvements over the baselines models in terms of both task completion and language quality.

Update: Accepted to Proceedings of AAAI 2020

References in corpus (18)

• Categorical Reparameterization with Gumbel-Softmax
• Neural Discrete Representation Learning
• Style Transfer from Non-Parallel Text by Cross-Alignment
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• A Network-based End-to-End Trainable Task-oriented Dialogue System
• Neural Approaches to Conversational AI
• Global-Locally Self-Attentive Dialogue State Tracker
• Hierarchical Text Generation and Planning for Strategic Dialogue
• Unsupervised Text Style Transfer using Language Models as Discriminators
• Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systems
• Latent Intention Dialogue Models
• Unsupervised Discrete Sentence Representation Learning for Interpretable Neural Dialog Generation
• Semantically Conditioned Dialog Response Generation via Hierarchical Disentangled Self-Attention
• Rethinking Action Spaces for Reinforcement Learning in End-to-end Dialog Agents with Latent Variable Models
• Personalizing a Dialogue System with Transfer Reinforcement Learning
• Domain Adaptive Dialog Generation via Meta Learning
• Zero-Shot Dialog Generation with Cross-Domain Latent Actions
• CARL: Aggregated Search with Context-Aware Module Embedding Learning

Cited by in corpus (5)

• GraphDialog: Integrating Graph Knowledge into End-to-End Task-Oriented Dialogue Systems
• Learning Non-Unique Segmentation with Reward-Penalty Dice Loss
• Semi-Supervised Dialogue Policy Learning via Stochastic Reward Estimation
• Multi-Facet Recommender Networks with Spherical Optimization
• Generalizable and Explainable Dialogue Generation via Explicit Action Learning