IRLab-er Maarten de Rijke leads a new consortium, called LESSEN, on low resource chat-based conversational intelligence. Our interaction with devices and access to information is increasingly conversational. People express their questions and needs to a system via a conversation; the system responds in a natural way, possibly by asking questions for clarifications, and meets the needs of its users. How do we ensure that this chat technology is also accessible for languages and applications with limited training data and compute power? LESSEN focuses on developing efficient chat algorithms and methodologies that can make optimal use of written conversational data, in a safe and transparent way.
LESSEN is a collaboration between the University of Amsterdam, Leiden University, University of Groningen, Amsterdam University of Applied Sciences, Radboud University, Achmea, Albert Heijn, Bol.com, KPN, Rasa Technologies, Ahold Delhaize, National Police Force. It will run for five years and employ 12 PhD students, a postdoc, a research engineer and a project manager.