Technical Report - Scalable querying of service contracts based on their temporal behavior

Service contracts (construed broadly, beyond web services and software, e.g. airline tickets and insurance policies) exhibit behavior that is rarely modeled completely in IT systems. This prevents consumers from automatically finding the contracts they need as well as providers from publishing the contracts they offer. We believe that the source of difficulty resides in the fact that these contract agreements tend to contain a great deal of information regarding the temporal interaction among various events (e.g. ”No refunds are allowed after a reschedule of the flight, which can be requested only before any flight leg has been used”). Such agreement clauses cannot be succinctly captured using a set of predefined attributes, ontology concepts or keywords, as one would need a name for each relevant sequence of events; and their natural language descriptions are not amenable to semantic search. In this paper, we present a broker that enables providers to register service contracts and consumers to query them, in both cases based on the contracts’ temporal behavior. We define what it means for a contract to permit a query, and describe an algorithm for checking this permission. We develop an indexing scheme for querying large collections of service contracts. We evaluate experimentally our broker implementation, showing that it scales well with both the number and the complexity of the contracts.