A new approach to compositional adaptation based on optimizing the global distance function and its application in an intelligent tutoring system

In this paper, we propose a new approach to compositional adaptation based on the idea of constituting the final solution in a way that its global difference with a set of solutions belonging to the retrieved cases can get minimized. Within this respect, the normalized distance between the current problem and each retrieved case is taken into account, using a global distance function, which makes use of the normalized local distances between the candidate final solution and the retrieved cases' solutions as the variables, and some coefficients as its parameters. Here, an approach based on secondary CBR can be used to determine the optimal values of these coefficients based on their past experiences in characterization of the global distance function. An example is illustrated in the paper, which shows the utility of this approach fro rearranging the necessary coursewares for students in the realm of intelligent tutoring systems.