Unsupervised Learning of Deterministic Dialogue Structure with Edge-Enhanced Graph Auto-Encoder

Authors

  • Yajing Sun Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China
  • Yong Shan Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Chengguang Tang Alibaba Group, Beijing, China
  • Yue Hu Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China
  • Yinpei Dai Alibaba Group, Beijing, China
  • Jing Yu Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China
  • Jian Sun Alibaba Group, Beijing, China
  • Fei Huang Alibaba Group, Beijing, China
  • Luo Si Alibaba Group, Beijing, China

Keywords:

Conversational AI/Dialog Systems

Abstract

It is important for task-oriented dialogue systems to discover the dialogue structure (i.e. the general dialogue flow) from dialogue corpora automatically. Previous work models dialogue structure by extracting latent states for each utterance first and then calculating the transition probabilities among states. These two-stage methods ignore the contextual information when calculating the probabilities, which makes the transitions between the states ambiguous. This paper proposes a conversational graph (CG) to represent deterministic dialogue structure where nodes and edges represent the utterance and context information respectively. An unsupervised Edge-Enhanced Graph Auto-Encoder (EGAE) architecture is designed to model local-contextual and global-structural information for conversational graph learning. Furthermore, a self-supervised objective is introduced with the response selection task to guide the unsupervised learning of the dialogue structure. Experimental results on several public datasets demonstrate that the novel model outperforms several alternatives in aggregating utterances with similar semantics. The effectiveness of the learned dialogue structured is also verified by more than 5\% joint accuracy improvement in the downstream task of low resource dialogue state tracking.

Downloads

Published

2021-05-18

How to Cite

Sun, Y., Shan, Y., Tang, C., Hu, Y., Dai, Y., Yu, J., Sun, J., Huang, F., & Si, L. (2021). Unsupervised Learning of Deterministic Dialogue Structure with Edge-Enhanced Graph Auto-Encoder. Proceedings of the AAAI Conference on Artificial Intelligence, 35(15), 13869-13877. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/17634

Issue

Section

AAAI Technical Track on Speech and Natural Language Processing II