Resource summary for pay-as-you-go dataspace systems

Dataspace systems are typically pay-as-you-go systems compared with traditional integration systems. Integration semantics and relationships among resources can be gradually enhanced over time. A system should support graceful transition between keyword query and structured query. A resource summary can be of great help in this situation, which provides an approximate overview of the entire structure of the dataspace. It can be refined over time when more and more semantics are obtained. We formally define a resource summary in this paper, and develop algorithms that allow us to automatically generate resource summaries and to refine them with the system running. Resource summaries are obviously important for advanced users and could be used in query optimization for pay-as-you-go dataspace systems.