Abstract
This work investigates the task-oriented dialogue problem in mixed-domain settings. We study the effect of alternating between different domains in sequences of dialogue turns using two related state-of-the-art dialogue systems. We first show that a specialized state tracking component in multiple domains plays an important role and gives better results than an end-to-end task-oriented dialogue system. We then propose a hybrid system which is able to improve the belief tracking accuracy of about 28% of average absolute point on a standard multi-domain dialogue dataset. These experimental results give some useful insights for improving our commercial chatbot platform FPT.AI, which is currently deployed for many practical chatbot applications.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Liu, B., Tur, G., Hakkani-Tur, D., Shah, P., Heck, L.: Dialogue learning with human teaching and feedback in end-to-end trainable task-oriented dialogue systems. In: Proceedings of NAACL (2018)
Serban, I., Sordoni, A., Bengio, Y., Courville, A.C., Pineau, J.: Building end-to-end dialogue systems using generative hierarchical neural network models. In: Proceedings of AAAI (2016)
Li, J., Galley, M., Brockett, C., Spithourakis, G.P., Gao, J., Dolan, B.: A persona-based neural conversation model. In: Proceedings of ACL (2016)
Wen, T.H., et al.: A network-based end-to-end trainable task-oriented dialogue system. In: Proceedings of EACL (2017)
Bordes, A., Boureau, Y.L., Weston, J.: Learning end-to-end goal-oriented dialogue. In: Proceedings of ICLR (2017)
Eric, M., Manning, C.D.: A copy-augmented sequence-to-sequence architecture gives good performance on task-oriented dialogue. In: Proceedings of EACL (2017)
Liu, F., Perez, J.: Gated end-to-end memory networks. In: Proceedings of EACL (2017)
Seo, M.J., Hajishirzi, H., Farhadi, A.: Query-regression networks for machine comprehension. In: Preprint ArXiv (2016)
Lei, W., Jin, X., Ren, Z., He, X., Kan, M.Y., Yin, D.: Sequicity: simplifying task-oriented dialogue systems with single sequence-to-sequence architectures. In: Proceedings of ACL (2018)
Ramadan, O., Budzianowski, P., Gasic, M.: Large-scale multi-domain belief tracking with knowledge sharing. In: Proceedings of ACL (2018)
Kelley, J.F.: Iterative design methodology for user-friendly natural language office information applications. ACM Trans. Inf. Syst. 2(1), 26–41 (1984)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
A Example Dialogues
A Example Dialogues
The following is three example dialogues generated by our system. The first dialogue is in single-domain.
Turn | Dialogue 1 | Domain |
|---|---|---|
Human | What time is my next dinner event and who will be attending? | Calendar |
Machine | Your dinner is on MONDAY at 5 PM | Calendar |
Human | That will do, thanks | Â |
Machine | You are welcome | Â |
The next two dialogues are in mixed-domains.
Turn | Dialogue 2 | Domain |
|---|---|---|
Human | Please give me the address and directions to home within 5 miles of here, avoiding all heavy traffic | POI |
Machine | HOME is located at 5671 BARRINGER STREET | POI |
Human | What is the forecast for the next 7 days? | Weather info |
Machine | What city are you interested in? | Weather info |
Human | Cleveland, will it be windy? | Weather info |
Machine | It will not be windy in Cleveland this week | Weather info |
Human | Thank you | Â |
Machine | You’re welcome |  |
Turn | Dialogue 3 | Domain |
|---|---|---|
Human | Are there any chinese restaurants nearby? | POI |
Machine | The closest is TAI PAN restaurant, and TAI PAN is 5 miles away with NO TRAFFIC | POI |
Human | What’s the weather like today? | Weather info |
Machine | What city are you wanting to know the weather for? | Weather info |
Human | for Newyork city please? weather info | Â |
Machine | Today is CLEAR SKIES, and WARM | Weather info |
Human | Ok, good, set the direction to there | POI |
Machine | The address is 830 Almanor Ln, i sent it on your screen | POI |
Human | Ok thanks | Â |
Machine | You’re welcome! |  |
Rights and permissions
Copyright information
© 2020 Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Luong, T.C., Le-Hong, P. (2020). Towards Task-Oriented Dialogue in Mixed Domains. In: Nguyen, LM., Phan, XH., Hasida, K., Tojo, S. (eds) Computational Linguistics. PACLING 2019. Communications in Computer and Information Science, vol 1215. Springer, Singapore. https://doi.org/10.1007/978-981-15-6168-9_22
Download citation
DOI: https://doi.org/10.1007/978-981-15-6168-9_22
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-15-6167-2
Online ISBN: 978-981-15-6168-9
eBook Packages: Computer ScienceComputer Science (R0)