Web services clustering for improving services discovery and selection / Abdelmoniem Helmy Ismail Abdelhafez ; Supervised Mervat Hassan Gheith , Akram Ibrahim Salah

Thesis (Ph.D.) - Cairo University - Institute of Statistical Studies and Research - Department of Computer and Information Science

The increasing usage of Web services on the Internet has led to much interest in service discovery. In industry, many applications are built by calling different web services available on internet. These applications are highly dependent on discovering correct and efficient web service. Clustering of web services is one methodology that can be used to enhance the speed of web service discovery process. Classifying Web services and labeling them based on their functional features have played a major role in several fundamental service management tasks, such as service discovery, selection, ranking, and recommendation. This doctoral thesis focuses on identifying and understanding the various cluster analysis methods for web services. The work deeply examined most of the available methods and techniques for web services analysis and clustering from different dimensions. The research conducted has involved an evolutionary process that starts from the investigation of the concepts of web services matching up to proposing a framework for web services discovery that adopt the approach of services clustering before matching with user requests. In this research, we presented an enhanced approach for service classification that combines text mining and machine learning technology. The method only uses text description of each service so that it can classify different types of services. This approach provides better performance in terms of service discovery efficiency and effectiveness. The approach automatically classifies services to specific domains and identifies key concepts inside service textual documentation. A comprehensive experimental study on real-world service data to demonstrate the effectiveness of the proposed approach was conducted in this research. An experiment made using supervised machine learning techniques such as Support Vectors machine (SVM) Naïve Bays (NB), K-Nearest Neighbors (K-NN IS), and Random Forest (RF) classification methods

Issued also as CD