Computing and Informatics

Context prediction is a promoting research topic with a lot of challenges and opportunities. Indeed, with the constant evolution of context-aware systems, context prediction remains a complex task due to the lack of formal approach. In this paper, we propose a new approach to enhance context prediction using a probabilistic temporal logic and model checking. The probabilistic temporal logic PCTL is used to provide an efficient expressivity and a reasoning based on temporal logic in order to fit with the dynamic and non-deterministic nature of the system's environment. Whereas, the probabilistic model checking is used for automatically verifying that a probabilistic system satisfies a property with a given likelihood. Our new approach allows a formal expressivity of a multidimensional context prediction. Tested on real data our model was able to achieve 78 % of the future activities prediction accuracy.