This project will create an architecture for a system that provides comprehensive support for context-aware applications. This architecture will make use of dynamically evolving data such as measurement streams collected by sensors, or Web services that execute requests on behalf of transactions. Data integration tools and ontologies will be developed and then applied to actual source data, to user profiles, and to the diverse problems of matching users with data sources and relevant courses of action. This activity will be augmented by the use of data mining techniques for dynamically constructing spatio-temporal user profiles and for profile classification. Methodologies will be developed to ensure scalability of the system. For example, the use of data cubes for caching user profiles and retrieved data, and the use of request aggregation. An authentication model will be developed using a distributed trust framework that will be applied to spatio-temporal context and data streams. The architecture will also incorporate a component for services arbitration that can manage on-demand resource allocation in a competitive environment. A system prototype will be tested against the operational scenario of a public health management application, provided by the Alliance of Chicago Community Health Services.
The architecture of the proposed system is shown in the following figure: