Declan O'Sullivan, Ruaidhrí Power

Bridging heterogeneous, autonomous, dynamic knowledge at runtime

Knowledge and Data Engineering Group
Department of Computer Science
Trinity College Dublin

The ubiquitous computing vision assumes that a wide range of application/services will interoperate seamlessly. Reaching global agreements on interoperability standards (which is the current assumption) will be impossible. Ubiquitous Computing applications/services will need to utilise knowledge encoded in a variety of formats, gathered and controlled by diverse entities, and which have a dynamic deployment lifecycle. Traditionally providing an integrated view upon such diversity has been achieved at design time. Model heterogeneity has been overcome through common models and expert mappings, coupled with an assumption of a static knowledge source deployment lifecycle. In addition, there had been acceptance that autonomy had to be ceded by knowledge owners in order to participate in an integrated view.

In this paper the authors contend that application interoperability solutions are needed which are metadata and ontology based, loosely coupled and runtime based. In particular two major interoperability use cases for applications in ubiquitous computing are examined: the sharing of context information and the matching of concepts based on semantics. In both cases a solution is needed that will bridge heterogeneous, autonomous, dynamic knowledge sources at runtime. For each case, the paper reports upon initial experiments and results. In conclusion, the experiments demonstrate that a variety of ontology, loosely coupled and runtime approaches show promise in bridging the kind of heterogeneous, autonomous, dynamic situations that will be prevalent in the ubiquitous computing era.