Web Services Classification Using an Improved Text Mining Technique

These days, service-oriented architecture is popular among application developers. So, a variety of web services have been developed for this purpose. The biggest advantage of using Service-Oriented architecture is that it is platform-independent and implementation-independent. Finding relevant web service which matches user need from the web services registry such as UDDI is a cumbersome task. Web services from the UDDI are searched using keywords and company information given in its registry. This approach cannot fulfill user requirements completely and may leave some potential matches. In this paper, an improved text mining-based web services classification technique is proposed for addressing this issue. Two features from WSDL file describing web services namely service name and operations is taken into consideration to prove this technique. An enhanced pre-processing approach having WSDL specific words removed is used. A vector is formed using the preprocessed information and later Maximum Entropy algorithm is applied on this pre-processed information to classify any new web service. This approach can be used to retrieve relevant web services effectively. The proposed technique improved the accuracy of existing classification techniques by 3%. This classification technique will boost the web services discovery by saving time of searching the complete Registry of Web Service.