Ontology-independent and QOS-enabled dynamic composition of web services in business domains

Abstract

This thesis proposes a novel and high-performance ontology-independent approach and methods for Quality of Services (QoS)-enabled dynamic web services discovery and composition. One proposed method uses Google distance for calculating semantic similarities instead of using the state-of-the-art ontological-based approaches in the semantic matching stage. A further new method is architected for the QoS operational matching stage of web services discovery. Moreover, the thesis proposes a hybrid approach to dynamic web services composition, called FOIQOS, consisting of using a prescriptive system for web services discovery and composition. Another problem the thesis addresses is the absence of comparisons of existing QoS-enabled composition approaches in the literature. To compare the new methods proposed in the thesis, FOIQOS and three other approaches for QOS-enabled dynamic web services composition were implemented. Experimental results show that the proposed FOIQOS approach significantly outperforms its ontology-based and heuristic-based method counterparts, in terms of both increased accuracy and reduced overhead.