Task-Oriented Dialog Systems that Consider Multiple Appropriate Responses under the Same Context
Yichi Zhang Tsinghua University Beijing, China \AndZhijian Ou Tsinghua University Beijing, China \AndZhou Yu University of California, Davis Davis, California, USA This work was done during Yichi Zhang’s summer internship at University of California, Davis.
Abstract
Conversations have an intrinsic one-to-many property, which means that multiple responses can be appropriate for the same dialog context. In task-oriented dialogs, this property leads to different valid dialog policies towards task completion. However, none of the existing task-oriented dialog generation approaches takes this property into account. We propose a Multi-Action Data Augmentation (MADA) framework to utilize the one-to-many property to generate diverse appropriate dialog responses. Specifically, we first use dialog states to summarize the dialog history, and then discover all possible mappings from every dialog state to its different valid system actions. During dialog system training, we enable the current dialog state to map to all valid system actions discovered in the previous process to create additional state-action pairs. By incorporating these additional pairs, the dialog policy learns a balanced action distribution, which further guides the dialog model to generate diverse responses. Experimental results show that the proposed framework consistently improves dialog policy diversity, and results in improved response diversity and appropriateness. Our model obtain
中文速览
任务导向对话系统（task-oriented dialog system）长期面临一个被忽视的难题：同一段对话背景下，系统其实可以采取多种合理的应对策略，但现有模型只会学习数据集中出现频率最高的那一种，导致策略单一、覆盖面窄。针对这一问题，研究者提出了"多动作数据增强"（Multi-Action Data Augmentation，MADA）框架：先用对话状态压缩历史信息，再从全量训练数据中挖掘出每个对话状态所对应的所有合理系统动作，然后在训练时把这些额外的状态-动作配对也纳入学习，从而让策略模型习得一个更均衡的动作分布。在此基础上，研究者还设计了领域感知多解码器网络（Domain Aware Multi-Decoder，DAMD），将增强后的数据框架与端到端多领域对话生成模型深度结合。实验在大规模多领域数据集MultiWOZ上取得了当时最优结果，人工评估也证实生成的回复更多样且合理；这项工作的意义在于，它首次将"一对多"对话属性系统性地引入任务导向对话生成，为构建能应对真实用户多变行为的对话系统提供了切实可行的思路。
原文 arXiv:1911.10484;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1911.10484v2