WordNet-powered Web Services Discovery Using Kernel-Based Similarity Matching Mechanism

With the rapid growth of web services, web services discovery becomes exceedingly important and challenging. Currently, many discovery approaches have been proposed such as keyword-based or VSM-based syntactic matching and ontology-based semantic matching. Syntactic matching approaches are clearly insufficient due to absence of semantic information. Although ontology-based semantic matching approaches expand the semantics and achieve sufficient precision, the high cost of defining ontologies and the lack of standards to integrate or reuse existing ontologies impede them widespread adoption in practice. To cope with these limitations, by introducing WordNet into VSM-based syntactic matching, a novel WordNet-powered feature vector extraction approach is proposed, which not only add semantic information but also reduce the dimension and sparsity of feature vectors. Furthermore, inspired by popular kernel tricks in machine learning, a set of kernel-based similarity measures are presented to well estimate the similarity of the real web services. Finally, the preliminary experimental results demonstrate the feasibility of the proposed approach.