Abstract
We focus on extended multi-turn multi-topic human-robot dialogues, which can be more challenging than short question-answering interactions. In earlier work we developed two dialogue systems for Nao robots: WikiTalk supporting Wikipedia-based dialogues on open-domain topics, and CityTalk supporting task-based dialogues on restaurant and hotel domains. We used WikiTalk with ERICA at Kyoto University, and now wish to make our systems available on multiple robot platforms. To support this aim we use Rasa open-source conversational AI, which creates transformer-based dialogue models that aim to recognise flexible intents in multi-turn multi-domain dialogues. To improve CityTalk we use Rasa knowledgebase actions backed by Neo4j graph databases which support knowledge graphs for multiple domains. By adding taxonomies and other semantic context to the knowledge graphs we aim to give more intelligent dialogue responses. We also plan to use the large MultiWOZ multi-domain dialogue dataset to support additional domains.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Multilingual WikiTalk: https://www.youtube.com/watch?v=NkMkImATfYQ.
- 2.
ERICA and WikiTalk: https://www.youtube.com/watch?v=Aq4Rfwrktr0.
- 3.
CityTalk Cambridge: https://www.youtube.com/watch?v=zWdd7kv5sX8.
- 4.
CityTalk Tokyo 2020: https://www.youtube.com/watch?v=OhjIJp8XBEA.
- 5.
Rasa architecture figure: https://rasa.com/docs/rasa/arch-overview.
- 6.
Rasa knowledgebase actions: https://rasa.com/docs/action-server/knowledge-bases.
- 7.
MultiWOZ datasets: https://github.com/budzianowski/multiwoz.
- 8.
MultiWOZ to Rasa conversion: https://github.com/RasaHQ/TED-paper.
References
Barrasa, J., Hodler, A.E., Webber, J.: Knowledge Graphs: Data in Context for Responsive Businesses. O’Reilly Media, Newton (2021)
Bocklisch, T., Faulkner, J., Pawlowski, N., Nichol, A.: Rasa: open source language understanding and dialogue management (2017). arXiv:1712.05181
Budzianowski, P., et al.: MultiWOZ – a large-scale multi-domain wizard-of-oz dataset for task-oriented dialogue modelling (2018). arXiv:1810.00278
Bunk, T., Varshneya, D., Vlasov, V., Nichol, A.: DIET: Lightweight language understanding for dialogue systems (2020). arXiv:2004.09936
Eric, M., et al.: MultiWOZ 2.1: a consolidated multi-domain dialogue dataset with state corrections and state tracking baselines (2019). arXiv:1907.01669
Jokinen, K., Wilcock, G.: Multimodal open-domain conversations with the Nao robot. In: Mariani, J., Rosset, S., Garnier-Rizet, M., Devillers, L. (eds.) Natural Interaction with Robots, Knowbots and Smartphones: Putting Spoken Dialogue Systems into Practice, pp. 213–224. Springer, Cham (2014). https://doi.org/10.1007/978-1-4614-8280-2_19
Lala, D., Wilcock, G., Jokinen, K., Kawahara, T.: ERICA and WikiTalk. In: Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI 2019), Macao, China, pp. 6533–6535 (2019). https://www.ijcai.org/proceedings/2019/947
Mosig, J.E.M., Vlasov, V., Nichol, A.: Where is the context? – a critique of recent dialogue datasets (2020). arXiv:2004.10473
Mrkšić, N., et al.: Multi-domain dialog state tracking using recurrent neural networks. In: Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics, pp. 794–799. Association for Computational Linguistics, Beijing (2015). http://aclweb.org/anthology/P15-2130
Robinson, I., Webber, J., Eifrem, E.: Graph Databases, 2nd edn. O’Reilly Media, Newton (2015)
Ultes, S., et al.: PyDial: a multi-domain statistical dialogue system toolkit. In: Proceedings of ACL 2017, System Demonstrations, Vancouver, Canada, pp. 73–78 (2017). https://aclanthology.org/P17-4013/
Vaswani, A., et al.: Attention is all you need (2017). arXiv:1706.03762
Vlasov, V., Drissner-Schmid, A., Nichol, A.: Few-shot generalization across dialogue tasks (2018). arXiv:1811.11707v1
Vlasov, V., Mosig, J.E.M., Nichol, A.: Dialogue transformers (2019). arXiv:1910.00486
Wilcock, G.: WikiTalk: a spoken Wikipedia-based open-domain knowledge access system. In: Proceedings of the COLING 2012 Workshop on Question Answering for Complex Domains, Mumbai, India, pp. 57–69 (2012). https://aclanthology.org/W12-6006/
Wilcock, G.: Using a deep learning dialogue research toolkit in a multilingual multidomain practical application. In: Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI 2018), Stockholm, Sweden, pp. 5880–5882 (2018). https://www.ijcai.org/proceedings/2018/869
Wilcock, G.: CityTalk: robots that talk to tourists and can switch domains during the dialogue. In: D’Haro, L.F., Banchs, R.E., Li, H. (eds.) 9th International Workshop on Spoken Dialogue Systems Technology, pp. 411–417. Springer, Cham (2019). https://doi.org/10.1007/978-981-13-9443-0_37
Wilcock, G., Jokinen, K., Yamamoto, S.: What topic do you want to hear about? A bilingual talking robot using English and Japanese Wikipedias. In: Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: System Demonstrations, Osaka, Japan (2016). https://aclanthology.org/C16-2025/
Acknowledgments
We thank Kristiina Jokinen (AI Research Center, AIST Tokyo Waterfront) for suggesting the use of Rasa conversational AI.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Wilcock, G. (2022). Recognising Flexible Intents and Multiple Domains in Extended Human-Robot Dialogues. In: Takama, Y., et al. Advances in Artificial Intelligence. JSAI 2021. Advances in Intelligent Systems and Computing, vol 1423. Springer, Cham. https://doi.org/10.1007/978-3-030-96451-1_13
Download citation
DOI: https://doi.org/10.1007/978-3-030-96451-1_13
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-96450-4
Online ISBN: 978-3-030-96451-1
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)