Topic Enhanced Controllable CVAE for Dialogue Generation (Student Abstract)

Yiru Wang, Pengda Si, Zeyang Lei, Yujiu Yang

AAAI 2020 pp. 13955-13956

doi:10.1609/AAAI.V34I10.7250 /aaai/2020/wang2020aaai-topic-a/

Abstract

Neural generation models have shown great potential in conversation generation recently. However, these methods tend to generate uninformative or irrelevant responses. In this paper, we present a novel topic-enhanced controllable CVAE (TEC-CVAE) model to address this issue. On the one hand, the model learns the context-interactive topic knowledge through a novel multi-hop hybrid attention in the encoder. On the other hand, we design a topic-aware controllable decoder to constrain the expression of the stochastic latent variable in the CVAE to reduce irrelevant responses. Experimental results on two public datasets show that the two mechanisms synchronize to improve both relevance and diversity, and the proposed model outperforms other competitive methods.

PDF AAAI Semantic Scholar

Cite

Text

Wang et al. "Topic Enhanced Controllable CVAE for Dialogue Generation (Student Abstract)." AAAI Conference on Artificial Intelligence, 2020. doi:10.1609/AAAI.V34I10.7250

Markdown

[Wang et al. "Topic Enhanced Controllable CVAE for Dialogue Generation (Student Abstract)." AAAI Conference on Artificial Intelligence, 2020.](https://mlanthology.org/aaai/2020/wang2020aaai-topic-a/) doi:10.1609/AAAI.V34I10.7250

BibTeX

@inproceedings{wang2020aaai-topic-a,
  title     = {{Topic Enhanced Controllable CVAE for Dialogue Generation (Student Abstract)}},
  author    = {Wang, Yiru and Si, Pengda and Lei, Zeyang and Yang, Yujiu},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2020},
  pages     = {13955-13956},
  doi       = {10.1609/AAAI.V34I10.7250},
  url       = {https://mlanthology.org/aaai/2020/wang2020aaai-topic-a/}
}