Practical Issues in e-Learning Multi-Agent Systems

Abstract

Multi-Agent Systems (MAS) seem, in theory, an excellent way to integrate the various component systems used in an e-Learning platform.

Why are they not used more broadly?

I suspect the main reason is that using a multi-agent architecture requires solving many issues from the beginning that can be ignored at the prototype stage in other approaches, like fine-grained security and fault tolerance.

To solve that I am developing the e-Learning Multi-Agent System (eLMAS) that will be used as integration platform in the European project Allegro.

In eLMAS there are two kinds of agents: “liaison” agents that connect existing e-Learning systems to the MAS, and learner agents that represent the user, storing the learner model and planning the learning activities.