Energy-based Neural Modelling for Large-Scale Multiple Domain Dialogue State Tracking

Anh Duong Trinh, Robert J. Ross, John D. Kelleher


Abstract
Scaling up dialogue state tracking to multiple domains is challenging due to the growth in the number of variables being tracked. Furthermore, dialog state tracking models do not yet explicitly make use of relationships between dialogue variables, such as slots across domains. We propose using energy-based structure prediction methods for large-scale dialogue state tracking task in two multiple domain dialogue datasets. Our results indicate that: (i) modelling variable dependencies yields better results; and (ii) the structured prediction output aligns with the dialogue slot-value constraint principles. This leads to promising directions to improve state-of-the-art models by incorporating variable dependencies into their prediction process.
Anthology ID:
2020.spnlp-1.5
Volume:
Proceedings of the Fourth Workshop on Structured Prediction for NLP
Month:
November
Year:
2020
Address:
Online
Editors:
Priyanka Agrawal, Zornitsa Kozareva, Julia Kreutzer, Gerasimos Lampouras, André Martins, Sujith Ravi, Andreas Vlachos
Venue:
spnlp
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
33–42
Language:
URL:
https://aclanthology.org/2020.spnlp-1.5
DOI:
10.18653/v1/2020.spnlp-1.5
Bibkey:
Cite (ACL):
Anh Duong Trinh, Robert J. Ross, and John D. Kelleher. 2020. Energy-based Neural Modelling for Large-Scale Multiple Domain Dialogue State Tracking. In Proceedings of the Fourth Workshop on Structured Prediction for NLP, pages 33–42, Online. Association for Computational Linguistics.
Cite (Informal):
Energy-based Neural Modelling for Large-Scale Multiple Domain Dialogue State Tracking (Trinh et al., spnlp 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.spnlp-1.5.pdf
Video:
 https://slideslive.com/38940154
Data
MultiWOZ