Web Services Discovery Based On Semantic Similarity

Web services play an increasingly important role in e-commerce and EAI (Enterprise Application Integration) application. Service discovery is the key problem of Web services technology and also the precondition of service call and composition. Web service discovery is the processes that service requestor find matching services in the candidate service set. Traditional service discovery technologyes are mainly based on keywords and UDDI(Universal Description,Discovery and Integration) framework. Keywords-based searching can only provide precise matching other than fuzzy matching, while UDDI framework-based searching need user to paticipate in filtering results, which leading to low efficiency.The emergence of semantic Web makes discorvey model for semantic Web service a hot issue. However, most studies focused on the service's input and output matching instead of precondition and effects matching, besides, they neglected non-functional factors like service's quality and reputation.With analysis of the existed methods of Web service discovery, this paper designs a new discorvey model for semantic Web service named DMSWS(Discovery Model for Semantic Web Service) and proposes a multi-level matching algorithm based on similarity. We focus on our work as follows:(1) Analysis the shortcomings of existed methodes of service description and put ontology and semantic information into it for semantic-based discovery.(2) A new discovery model named DMSWS is presented, which consisting four subsystemes like service registery center, document translator, matching center and reputation manager.(3) A similarity-based filtering algorithm is proposed for multi-level matching. The algorithm consider all elemets like service name, description information, input and output, reputation and effects respectively, and algorithms like levenshtein distance, cosine theorem and symtanic distance are given in this paper to calculate their similarity.(4) Introduce reputation into DMSWS model, list the factors which affect reputation and provide a method to measure service's reputation.(5) Design DMSWS model and implement its subsystem briefly.In this thesis, we analyse multi-level matching algorithm qualitatively and quantitatively, and by experiments, we find that the precision and recall of the algorithm are higher than existed ones, which show its superiority.