Multi-criteria decision making for broker agents in eLearning environments

In this paper we deal with the problem of enhancing the decision making abilities of broker agents in digital learning resources brokerage systems. Following a multi-criteria decision making approach, we provide broker agents with the means to qualitatively represent the user preferences and efficiently evaluate the proposed alternatives. Learner modelling and digital learning resources description parameters are inferred by a classic cognitive style modelling method, the Honey and Mumford model. We study two different aspects in modelling the learning resources recommendation problem: first formulated as a decision problem for the agent representing the learner by using a unique synthesis criterion approach; second, as a decision problem for the agent representing the content provider by using an approach based on the UTA method. Both aspects are studied in the context of a generic agent-based brokerage architecture, with broker agents requesting, describing and offering e-learning courses.