Virtual assistants
Goal
Develop and test a chat-bot agent for one of the domains from our external partners: (i) Stadsbibliotek: Gothenburg City Library, and (ii) the SKANDIBOT project which focuses on the usage of chat-bots in second language learning of Danish and Swedish.
Background
Chat-boats are currently used in a variety of applications. In this project we examine a development of a chat-bot system in either the domain of (i) the patron conversations with a librarian or (ii) interactions of second language learners of Danish or Swedish with a tutor.
Problem description
- T1: Extracting and modelling interactive strategies for a particular domain from user data and implementing these in one of the chat-bot platforms (e.g. RASA, Chatfuel).
- T2: Adaptation of the chosen chat-bot platform for Swedish (and Danish!), e.g. finding the correct NLU models.
- T3: Expansion of the limited annotated data using techniques such as dataset boosting, creation of artificial datasets, and applications of transfer learning from other domains and others.
- T4: Application and testing of ML models on the datasets from the previous point.
- T4: Testing of the resulting system in the target domain.
Recommended knowledge and skills
For students in Masters in Language Technology (MLT), the master thesis would build on the following courses
- Computational semantics
- Dialogue systems
- Machine learning
- AI: Cognitive systems
or equivalent knowledge and skills for external students.
A potential candidate should be comfortable in programming with Python, machine learning and text and image processing tools and techniques.
Supervisors
Simon Dobnik and external partners