1st Semester 2021/22: Dialogue Modelling
- Raquel Fernández (TA: Mario Giulianelli)
Conversation or dialogue is the most natural way in which we humans use language, and arguably the holy grail of language-enabled AI systems. With machine learning advancing at such a rapid pace, we now have powerful tools for modelling interacting agents. Yet, to model human-like conversational abilities remains remarkably difficult. This project will examine what makes dialogue so challenging, delving into classic and contemporary research in linguistics, cognitive science, Natural Language Processing and AI.
The precise organisation will depend on the number of students. In principle, there will be a few initial lectures during the first week. In the following weeks, students will give presentations about related literature and take part in discussions. By the end of the project, students will be expected to hand in a short paper (for example, outlining a research proposal with some initial results).
The project will only go ahead if there are at least 4 students registered. Given the current uncertainty with the Covid-19 pandemic, it is unclear what the format will be. There may be sessions online and sessions in person, or we may have to resort to a fully online setup. I will not be able to accommodate a hybrid setup though (with both online and in-person students within the same session).
An interest in language and NLP. NLP skills and experience will be valuable, but are not absolutely necessary -- we are willing to adapt the learning goals and assessment of the presentations and the short paper to different backgrounds.
The assessment will be based on the in-class presentation(s) and the final short paper. Participation, both in the sense of being present during class and in the sense of actively contributing to the discussion, will be taken into account.
Some relevant background literature (other papers to be discussed in class will be given at the beginning of the project):
- Bisk et al. Experience Grounds Language, in Proceedings of EMNLP, 2020.
- Brennan & Clark, Conceptual pacts and lexical choice in conversation. Journal of Experimental Psychology, 22:1482–1493, 1996.
- Brown-Schmidt et al., People as Contexts in Conversation, in The Psychology of Learning and Motivation (pp. 59–99). Academic Press: Elsevier Inc. 2015.
- Clark, Using Language, CUP, 1996.
- The Wikipedia entry for "Grounding in Communication" is reasonably good.
- Frank & Goodman, Predicting Pragmatic Reasoning in Language Games, Science, vol. 336, 2012.
- Levinson, Pragmatics, CUP, 1983.
- McTear, Conversational AI: Dialogue Systems, Conversational Agents, and Chatbots, Morgan & Claypool Publishers, 2020. [eBook available at the UvA online library]
- Pickering & Garrod, Towards a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27:169-225, 2004.
- Sankar et al. Do neural dialogue systems use the conversation history efficiently? An empirical study, in Proceedings of ACL, 2019.