Grounded Conversation Generation as Guided Traverses in Commonsense Knowledge Graphs

Houyu Zhang, Zhenghao Liu, Chenyan Xiong, Zhiyuan Liu


Abstract
Human conversations naturally evolve around related concepts and hop to distant concepts. This paper presents a new conversation generation model, ConceptFlow, which leverages commonsense knowledge graphs to explicitly model conversation flows. By grounding conversations to the concept space, ConceptFlow represents the potential conversation flow as traverses in the concept space along commonsense relations. The traverse is guided by graph attentions in the concept graph, moving towards more meaningful directions in the concept space, in order to generate more semantic and informative responses. Experiments on Reddit conversations demonstrate ConceptFlow’s effectiveness over previous knowledge-aware conversation models and GPT-2 based models while using 70% fewer parameters, confirming the advantage of explicit modeling conversation structures. All source codes of this work are available at https://github.com/thunlp/ConceptFlow.
Anthology ID:
2020.acl-main.184
Volume:
Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics
Month:
July
Year:
2020
Address:
Online
Editors:
Dan Jurafsky, Joyce Chai, Natalie Schluter, Joel Tetreault
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2031–2043
Language:
URL:
https://aclanthology.org/2020.acl-main.184
DOI:
10.18653/v1/2020.acl-main.184
Bibkey:
Cite (ACL):
Houyu Zhang, Zhenghao Liu, Chenyan Xiong, and Zhiyuan Liu. 2020. Grounded Conversation Generation as Guided Traverses in Commonsense Knowledge Graphs. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2031–2043, Online. Association for Computational Linguistics.
Cite (Informal):
Grounded Conversation Generation as Guided Traverses in Commonsense Knowledge Graphs (Zhang et al., ACL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.acl-main.184.pdf
Video:
 http://slideslive.com/38928828
Code
 thunlp/ConceptFlow +  additional community code
Data
ConceptNet