Epiphyte advisor systems for collaborative learning

Abstract

Today's information technologies support interactions at distance. Taking advantage of them, Télé-université has integrated collaborative learning activities into the learning set-up it proposes to learners. A typical learning environment provides learners with many information systems that help them to perform their tasks, to communicate with other people, and to manage their on-going learning processes. But collaborative learners still need resources capable to start again a stalled learning process. Furthermore group work within a class has to be managed: how many groups exists, who are the members of these groups, what is the state of the collaborative process... These models must follow closely the evolution of the groups. Such ever up-to-date models are very useful to help human and artificial tutors to balance the work load, to adjust their advising strategies, and to direct learners towards appropriate resources. In this paper, we address these issues related to collaborative learning by means of epiphyte advisor information systems. As the term epiphyte suggests, the advisor information systems are grafted onto existing arbitrary information systems. We show how to build distributed advisors; how they can take advantage of uptodate models to manage groups, and how they can use models of groups to produce advice for a collaborative learning process. These advisors rely on agents, organizations and hierarchical knowledge graphs. Hierarchical knowldege graphs help to design and structure descriptions of an advisor. At run-time, agents provide autonomy to the advisor's components, and organizations link and co-ordinate agents.