An automatic approach for tagging Web services using machine learning techniques

Web services have become popular and increasingly important in e-business and e-commerce applications especially in large scale distributed systems. As a result, an increasing number of Web services have been developed. However, this huge collection of Web services makes the task of locating a suitable one more challenging as well as more difficult. Automatic clustering of Web services groups services with similar functions together. Clustering could greatly boost the power of Web service search engines and generate tags to improve the search accuracy of tag-based service recommendation. In this paper, we propose a Web service clustering technique based on Carrot search clustering and K-means to group similar services together. These clustered groups are then tagged. We also develop a tag-based service recommendation for WSDL documents using naive bayes algorithm to classify Web services into different tags. We demonstrate that the proposed clustering approach is effective for Web service discovery through two sets of real data.