Semi-automatic Algorithem Based on Web Service Classification

Abstract

The number of web services on the Internet is increasing continuously. Determining a web service's domain is very helpful for many issues like web service discovery, composition and semantic annotation. Based on predefined domains, we implemented semi-automatic web service classification with supervised machine learning technique in the paper. First we exploited the process of web service classification with choosing element from web service description document, preprocessing, mapping to a digital vector and applying classifier. Then service classification accuracy of single element and multiple elements based on four machine learning algorithms were tested. At last, experimental results showed that C4.5 classifier had a higher accuracy and efficiency in web service classification. In addition, multi-element classification did not perform better than single element.