Using Numeric Optimization To Refine Semantic User Model Integration Of Adaptive Educational Systems

Nowadays, it is a common practice to use several educational systems in one domain. In this situation, each of the systems should be able to provide the best user modeling based on the integrated data about the user. However, differences in domain conceptualization complicate the ability of the systems to understand each other’s user models and necessitate the use of labor intensive and time consuming alignment procedures that require involvement of knowledge engineers. While the latter are best at detecting associative links between the user model items, they fail to reliably identify the strengths of these associations. In this paper, we are proposing a method to improve the user model mapping by using a numerical optimization procedure. Our results show that numerical weight optimization helped to decrease the amount of manual work and improved the target model accuracy.