Web Service Ranking Using Semantic Profile Information

The promise of dynamic binding and the ability to dynamically and seamlessly move between service providers can only be realized through the path of semantic expressibility. Once we describe the semantics of a service in itpsilas advertisement, a semantic matchmaker can match a query with the set of advertisements that satisfy the query conditions. The state of the art today uses the IOPE (Inputs Outputs Preconditions and Effects) form of advertisements with languages such as OWL-S being used to represent the semantics. Inputs and outputs usually refer to concepts in an ontology. Many semantic matchmakers exist today and they focus on matching the IOPE form of the advertisement to the query. This may result in many matches and they return a set of matched advertisements without rank ordering them. We believe itpsilas important to rank the returned set of services in order to choose the best service. In this paper we present such a ranking algorithm that uses the IOPE information present in the each of the services relative to the query. Our algorithm can complement other approaches that use the history of issued queries and data mining. We have evaluated the effectiveness of the ranking algorithm against a benchmark ranking that is done manually and shown that our ranking scheme is close to the best ranking scheme possible.