CHAI Overview - Pervasive Personalised Interaction based on Lifelong User Models

Prof Judy Kay

University of Sydney

Tue Apr 17 10:00:00 NZST 2007 in Room 101, Erskine Building (MSCS)

Abstract

This talk overviews the research vision of the CHAI (Computer Human Adaptive Interaction) lab at University of Sydney. CHAI aims to address the range of challenges that underpin adaptive, personalised, interaction that ensures the user can always maintain control. CHAI research tackles key systems and user interface challenges, from the foundations such as the ontologies defining the meaning of what is modelled about the user, through the systems architectures that can support pervasive user modelling to the interface issues of user control and including exploration of novel user interfaces. Demonstrators are in teaching systems and pervasive computing. The foundation of the research lies in lifelong user models that are scrutable. The talk explains the motivation for this notion of scrutability. Then, to give a sense of the directions this is taking CHAI, this talk presents a selection of our recent systems, showing how they address key parts of the long term vision. These include: the Just-in-time Training system (JITT); the Unix tutor with web-based teaching where the personalisation is scrutable; and MyPlace, a pervasive personalised service that informs the user about relevant people, things and services. The talk also overviews some of CHAI projects in novel pervasive interfaces: CRUISER and OnTop tabletop interaction projects; new interfaces to help groups learn to collaborate more effectively in long term projects; and the Keep-In-Touch message based communication system.

Biography

Judy Kay is a principal on the Computer Human Adaptive Interaction, CHAI lab, which conducts both fundamental and applied research in personalisation and pervasive computing. A driving goal of her personalisation research is to ensure the user can maintain control, being able to scrutinise and control the whole process of personalisation: the user can determine what is modelled about them, how this is managed and how it is used. This is particularly important in pervasive computing. The core technologies to come from this are the Personis user modelling server and Personis-Lite. This provides fundamental technologies for personalisation. The testbed areas are in ubiquitous, pervasive computing as well as intelligent teaching systems. The latter reflect the research group's work in teaching computer science and in building teaching systems that help develop reflective, deep learners.