An Empirical Analysis on the Role of WSDL Metrics in Web Service Anti-Pattern Prediction

Service-Oriented Architecture (SOA) is one of the most well-known models for designing web systems. SOA system evolution and maintenance is challenging because of its distributive nature and secondly due to the demand of designing high-quality, stable interfaces. This evolution leads to a problem called Anti-patterns in web services. It is observed that these anti-patterns negatively impact the evolution and maintenance of software systems, making the early detection and correction of them a primary concern for the software developers. The primary motivation of this work is to investigate the relationship between the Web Service Description Language(WSDL) metrics and anti-patterns in web services. This research aims to develop an automatic method for the detection of web service anti-patterns. The core idea of the methodology defined is to identify the most crucial WSDL metrics with the association of various feature selection techniques for the prediction of anti-patterns. Experimental results show that the model developed by using all the WSDL quantity metrics(AM) shows a bit high performance compared to the models developed with the other metric sets. Experimental results also showed that the performance of the models generated using Decision Tree(DT) and Major Voting Ensemble(MVE) is high compared to the models generated using other classifier techniques.