User modeling on adaptive web-based learning communities

Nowadays, a great number of adaptive systems are successfully applied in a variety of domains. Educational systems axe no exception. In this context, the boom in so-called learning communities has forced a change in user models, which axe used as the basis for adaptation in these systems. In this framework, to be really adaptive, these models should be dynamically constructed from both user and usage data. To achieve this goal, our approach for user modeling on an adaptive web-based learning community is based on a combination of both knowledge based and machine learning techniques. In particular, in this paper we will focus on some of the main problems that arise from the application of machine learning techniques in user modeling and the convenience of using certain combination of several machine learning techniques.