A comparative evaluation of semantic Web service discovery algorithms and engines
MetadataShow full item record
As Web service technology becomes more prevalent, effective discovery becomes more important. Discovering Web services using UDDI can be difficult, since the discovery mechanism mostly takes into account the syntactic aspect of Web services by providing an interface for keyword and taxonomy based searching. Due to this, semantics implied by the information provider may not be explicitly represented, leading to possible misinterpretation by others. Therefore, in the last several years there has been significant research on semantic web service discovery. In this work we analyze and compare four different algorithms for discovery in Semantic web services: OWL-S MX, WSMO, MWSDI and TVERSKY. Furthermore, an empirical evaluation is given for the last two algorithms. Comparing the former two with the latter two is somewhat problematic, since they select services based on two separate criteria. The former two use a logic based degree of match that includes a syntactic similarity score in some cases, while the latter two produce a single match score based on semantics (although syntax is considered as a minor factor). When comparing the TVERSKY Algorithm (property based comparison) and MWSDI algorithm (based on taxonomy and properties), we find them to be in general agreement, although the similarity scores given by the MWSDI algorithm to be closer agreement with the human evaluators, but more time consuming. The TVERSKY algorithm did better than MWSDI in some cases and performed similar in some cases.