Web Service Selection Using Soft Computing Techniques

Abstract

Web service selection is one of the important aspects of SOA. It helps to integrate the services to build a particular application. Web services need to be selected using appropriate interaction styles i.e., either Simple Object Access Protocol (SOAP) or Representational State Transfer Protocol (REST) because choosing web service interaction pattern is a crucial architectural concern for developing the application, and has an impact on the development process. In this study, the performance of web services for Enterprise Application based on SOAP and REST are compared. Since web services operate over the network, throughput and response time are considered as metrics for evaluation. In the literature, it is observed that, emphasis is given on interaction style for selecting web services. However, as the number of services grows day by day, it is time-consuming and difficult to select services that offer similar functionalities. Web services are often described in terms of their functionalities and set of operations. If a customer chooses an application that is of low quality or have malicious content that can affect the overall performance of the application. Hence, web services are selected based on the quality of service (QoS) attributes. In this proposed work, various models are designed using soft computing techniques such as Back Propagation Network (BPN), Radial Basis Function Network (RBFN), Probabilistic Neural Network (PNN) and hybrid Artificial Neural Network (ANN) for web service selection, and their performances are compared based on various performance parameters.