Graph Neural Network Policies and Imitation Learning for Multi-Domain Task-Oriented Dialogues

Thibault Cordier, Tanguy Urvoy, Fabrice Lefèvre, Lina M. Rojas Barahona


Abstract
Task-oriented dialogue systems are designed to achieve specific goals while conversing with humans. In practice, they may have to handle simultaneously several domains and tasks. The dialogue manager must therefore be able to take into account domain changes and plan over different domains/tasks in order to deal with multi-domain dialogues. However, learning with reinforcement in such context becomes difficult because the state-action dimension is larger while the reward signal remains scarce. Our experimental results suggest that structured policies based on graph neural networks combined with different degrees of imitation learning can effectively handle multi-domain dialogues. The reported experiments underline the benefit of structured policies over standard policies.
Anthology ID:
2022.sigdial-1.10
Volume:
Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue
Month:
September
Year:
2022
Address:
Edinburgh, UK
Editors:
Oliver Lemon, Dilek Hakkani-Tur, Junyi Jessy Li, Arash Ashrafzadeh, Daniel Hernández Garcia, Malihe Alikhani, David Vandyke, Ondřej Dušek
Venue:
SIGDIAL
SIG:
SIGDIAL
Publisher:
Association for Computational Linguistics
Note:
Pages:
91–100
Language:
URL:
https://aclanthology.org/2022.sigdial-1.10
DOI:
10.18653/v1/2022.sigdial-1.10
Bibkey:
Cite (ACL):
Thibault Cordier, Tanguy Urvoy, Fabrice Lefèvre, and Lina M. Rojas Barahona. 2022. Graph Neural Network Policies and Imitation Learning for Multi-Domain Task-Oriented Dialogues. In Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 91–100, Edinburgh, UK. Association for Computational Linguistics.
Cite (Informal):
Graph Neural Network Policies and Imitation Learning for Multi-Domain Task-Oriented Dialogues (Cordier et al., SIGDIAL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.sigdial-1.10.pdf