Using a Machine Learning Approach To Support an Intelligent Cooperative Multi-Agent System

In this paper, we describe a machine learning approach, ID3 Decision Tree Induction Algorithm, to analyzing and predicting learning style of learners on-line. Our goal is to adapt the interaction by choosing an appropriate presentation for the learners. One way to make a good adaptation is by extracting some knowledge about each learner such as learners’ behavior during a learning session, knowledge level, and learning styles. We have developed a Confidence Intelligent Tutoring System (CITS), which is based on a multi-agent approach, in order to manage negotiations within a community of learners. The main goal of CITS is to adapt intelligent distance learning environments interactions among the participants to be more cooperative. This paper focuses on how CITS can extract the knowledge from learners. An experiment shows that this approach can determine learning style with 78% accuracy. Using this way, CITS can predict learning style instead of using a long questionnaire.