Deep Reinforcement Learning For Modeling Chit-Chat Dialog With Discrete Attributes

Chinnadhurai Sankar, Sujith Ravi


Abstract
Open domain dialog systems face the challenge of being repetitive and producing generic responses. In this paper, we demonstrate that by conditioning the response generation on interpretable discrete dialog attributes and composed attributes, it helps improve the model perplexity and results in diverse and interesting non-redundant responses. We propose to formulate the dialog attribute prediction as a reinforcement learning (RL) problem and use policy gradients methods to optimize utterance generation using long-term rewards. Unlike existing RL approaches which formulate the token prediction as a policy, our method reduces the complexity of the policy optimization by limiting the action space to dialog attributes, thereby making the policy optimization more practical and sample efficient. We demonstrate this with experimental and human evaluations.
Anthology ID:
W19-5901
Volume:
Proceedings of the 20th Annual SIGdial Meeting on Discourse and Dialogue
Month:
September
Year:
2019
Address:
Stockholm, Sweden
Venues:
SIGDIAL | WS
SIG:
SIGDIAL
Publisher:
Association for Computational Linguistics
Note:
Pages:
1–10
Language:
URL:
https://aclanthology.org/W19-5901
DOI:
10.18653/v1/W19-5901
Bibkey:
Cite (ACL):
Chinnadhurai Sankar and Sujith Ravi. 2019. Deep Reinforcement Learning For Modeling Chit-Chat Dialog With Discrete Attributes. In Proceedings of the 20th Annual SIGdial Meeting on Discourse and Dialogue, pages 1–10, Stockholm, Sweden. Association for Computational Linguistics.
Cite (Informal):
Deep Reinforcement Learning For Modeling Chit-Chat Dialog With Discrete Attributes (Sankar & Ravi, 2019)
Copy Citation:
PDF:
https://aclanthology.org/W19-5901.pdf