Imbalanced data preprocessing model for web service classification

Abstract

Web services are a novel method of web application development. They allow business to adapt to a new environment and change quickly according to customer needs. The client requires high-quality web services with minimal response time, more security, and high availability. With the increasing demand for web services, the introduction of web services rapidly in the business environment has influenced rapidly the web service quality. In the present work, a novel model for web service classification is proposed. Three metaheuristic techniques: Whale optimization algorithm, Simulated annealing algorithm, and Ant colony optimization are used to select the best subset of features. Web service-based imbalanced dataset is balanced using SMOTETomek (Synthetic minority oversampling + Tomek link). Ensemble Adaboost and Gradient boosting algorithms are used for the creation of a web service prediction model. The publicly available QWS dataset is used for experimental purposes. The results of the proposed models are compared with machine learning techniques. It was observed that the Ant colony algorithm performed best for relevant feature selection and the Ensemble Adaboost and Gradient boosting algorithm outperformed all other machine learning techniques for web service classification.