Semantic methods for service categorization an empirical study

In this work we provide an initial analysis of service categorization, the process of associating services with ontologies. Service categorization is an important pre-processing step to tasks such as service composition. The absence of semantic understanding of services may yield erroneous compositions and therefore, service categorization can assist in determining the correctness of a composition. We analyze two common methods for text processing, TF/IDF and context analysis. We also test two types of service description, textual and WSDL. Our initial results indicate that context analysis is more useful than TF/IDF and that WSDL description provides better categorization than textual description.