A cognitive service adaptation decision algorithm: XCSCAP

Personalization and self-adaptation are two very import characteristics of the cognitive service. They are essential for the reducing of the user interaction and the improving of the user experience. In order to provide personalized and self-adaptable service to different people in different environments, a context aware user preference model need to be constructed, which will be used to determine the adapted action according to the current context. However usually the actions to be chosen have various parameters, and this will cause the XCS algorithm have a very long convergence time. Therefore we proposed a modified XCS (eXtend Classifier System) algorithm called XCSCAP (XCS with Complex Actions and Parameters). The experiments result showed that XCSCAP can have a much shorter time to converge when dealing with complex actions.