The use of Description Logics (DLs) in modeling various real-world domains, and reasoning about them, has wellknown benefits. We believe that the same framework can be used for representing a user-centric view of usage contexts. A DL-reasoner can then be used for organizing context definitions, merging domain knowledge into these definitions, and performing recognition of contexts from sensor inputs. A pure DL-based approach, however, has certain limitations in a context environment; hence, a hybrid reasoning approach is proposed.

To ensure syntactic and semantic interoperability, we adopt a Semantic Web -compliant approach that uses the OWL-DL variant of the OWLWeb Ontology Language. This allows us to aggregate various heterogeneous data sources, both directly in Semantic Web formalisms, and indirectly via the use of a transformation framework capable of using data from legacy applications.

Context and domain models, along with associated reasoners, are themselves context, user and/or organizationspeci fic; hence multiple systems should be maintained and connected into a distributed architecture. In order to deploy this architecture in a mobile environment, an underlying SIP-based provisioning framework has been experimented with, enabling device, network and location transparency.