SURVEY OF MULTI-AGENT BASED SOLUTIONS FOR COLLABORATIVE E-LEARNING AND NEW GENERIC E-PEDAGOGY BASED ARCHITECTURE PROPOSAL

In recent years, multi-agent based solutions have evolved rapidly affecting their application in many disciplines. One of the sciences in which they are more frequently being applied is education (eLearning). This application of Multi-Agent Systems (MAS) brings to eLearning many advantages compared to other architectures ( peer to peer and Server/Client ) thanks to its adaptability, interactivity, distribution, collaboration, personalization, security and intelligence. But one of the big points of debate is the dramatic student dropout rates from eLearning courses (typically 85% - 95% [1]), this is why applying the traditional metrics of higher education to eLearning is misleading. eLearning is a very different kind of educational package and needs different metrics. We can talk about traditional pedagogy and eLearning pedagogy.

ePedagogy refers to strategies of instruction or a style of instruction to support eLearning. A range of design strategies can be used to help build quality online programs using Multi-Agents Systems.

Many eLearning MAS architectures have been developed in recent years each one proposes a module composed of different layers with multiple agents. But most of these architectures don’t consider all the features in a single system. So there is a need for a generic architecture that should support all the features especially the new pedagogies to make the eLearning system more efficient and to help learners to complete their courses with success.

In this paper we will present a survey of existing MASs in collaborative eLearning and their pedagogy implementation and limitations then we are going to introduce a new generic eLearning MAS architecture that integrates new ePedagogical theories with implementation of adapted pedagogical agents , semantic agents to analyze interactions between system actors, social networks agents to understand the background of eLearners and Learning Object (LO) agents to find the LOs that the given learner is really interested in. All these agents in addition to the basic agents (eLearner, eTeacher, Quality, Content, Database, Interface ) will be categorized in layers.

To demonstrate the interactions between all agents in deferent layers we use a communication diagram that shows roles and actions of each agent and its position in the system.
As result of this paper we will have a generic model that combines the latest information technology and new pedagogical theories, making the eLearner in the centre, trying to find the best path for him, encouraging him to accomplish the course.

References:
[1] Jordan, K. (2014). Initial trends in enrolment and completion of massive open online courses. International Review of Research in Open and Distance Learning, 15(1), 133-160. (http://www.katyjordan.com/MOOCproject.html)