Using Semantics to Discover Web Services Based on Partial Data

Abstract

The authors developed an algorithm and a proof-of-concept tool to help business analysts search a pool of hundreds or thousands of services to find the closest ones to their needs when they are composing business processes, even when an exact match does not exist. The authors achieve this by semantically annotating the services with properties from an ontology, while, traditionally, only the concepts are used. This enables them to generate finer-grained partial semantic mappings between a query and the signature of the services published in a registry, despite potentially different parameter cardinalities and types. Their searches return a ranked list of those services available in the registry that most closely match the query specification. The analyst can then study the hits to see which of the suggested alternatives is more convenient. Such an approach is fundamental to effectively manage reuse in large service-oriented deployments, where the number of services in the registry becomes too voluminous for browsing individually or even syntactic searches that rely on some degree of memorization by the analyst.