Research On Mining Relationships Between Web Services And Its Applications

With the development of service computing and cloud computing, the number of Web services on the Internet are growing rapidly. By invoking different Web services to quickly build innovative Mashup application, has aroused widespread interest in the service developers, and how to help users to quickly locate the Web services which the users needed is still a challenging problem in the field of service computing. According to the latest statistics, the usage of Web services is considerable low and the reuse rate is rather low, it's not benefit to the creation of Mashups and the sustainable development of Web services eco-system. Therefore, improving the usage of Web services is also a key problem to be resolved.In order to solve the above problems, this paper mainly mines the various relations between Web services and applys them to the search and recommendation of Web services, and the classification of Web services is also studied. The specific contents are as follows:Firstly, this paper presents an effective and efficient approach to mine the similar relation, composition relation and potential composition relation between Web services.After mining the similarity relation and composition relation between Web services,then applying the ideology of link prediction to mine the potential composition relation between Web services. Case studies and experiments based on real datasets demonstrate that the approach can guide users to find desired services, and thus improves efficiency of the service discovery process. Furthermore, the approach can find new services that Mashup developers never used and improve the useage of Web services.Secondly, this paper presents a multi-relation based manifold ranking algorithm for API recommendation. The approach exploits the similar relations between Mashup clusters and between Web services, the composition relation and potential composition relation between Web services, and the inclusion relations between Mashup clusters and Web services. Then integrating these various ideologies to original manifold ranking algorithm, and recommend a series of relevant Web services to Mashup developers, according to the requirement descrption of them. Experimental results demonstrate that the approach can not only recommend services that are popular in the Mashups, but also recommend services that have similarity relations, composition relations, and potential composition relations with each other. Furthermore, the approach can satisfy the requirement of the Mashup developer well and improve the usage of Web services.Finally, this paper proposes a Web services classification via exploiting the LDA topic model. Which takes a Web service's name, description and tags into consideration, and use the LDA(Latent Dirichlet Allocation) topic model to model text description of Web services. After obtaining the topic distribution of Web services,then it make use of a couple of automated classification methods to classify the Web services, namely, K-Nearest Neighbor and Support Vector Machine. The experimental results demonstrate that the proposed classification method based on the LDA topic model significantly outperforms those based on traditional word statistical models.LDA topic model can capture the semantic information of the Web services textual descriptions, make the results more accurate and improve the efficiency of services discovery.