Context Modeling For Supporting Uncertainty Reasoning And Context Reasoning Based On Bayesian Network

Context-aware is a key technology in the realization of intelligent environment. A lot of context data which are derived from sensors, are imprecise or uncertain. And the context of a higher level which is inferred from this uncertain data is still imprecise. A majority of context model are constructing by OWL based on the basic binary logic. But OWL does not support uncertainty reasoning. Therefore, how context-aware applications modeling based on ontology to resolve the uncertainty of context information and the uncertain reasoning, has increasingly become a context-aware research point.In response to this problem, a construction approach for context model supporting uncertainty reasoning is designed in this paper. In this method, the context model has been expanded to support uncertainly reasoning. This paper designs a two-level framework of the ontology model. Using the "domain-independent ontology" describe the general concept, and using the "domain ontology" describe the domain-concept. This paper selects the Bayesian networks to achieve uncertainty reasoning. According to the structural characteristics of Bayesian networks, this paper expands the context model, and introduces the meta-information to describe the uncertain information. The methods of selecting and sorting the nodes of Bayesian network from the expansion model are given. The experiments, learning the structure and parameters of Bayesian network from the samples, and using this Bayesian network in the uncertainty reasoning, simultaneously completing the certain reasoning based on the rules, show that the usability of the expansion context model.