Kussul O. Information technology for service selection in service-oriented systems using reputation

The thesis is dedicated to the service selection problem in service oriented systems using reputation. A review on existing service selection methods is given and the efficiency of the reputation model usage for this purpose is considered. The probabilistic estimation of service reputation in service-oriented system is proposed for the accuracy of the utility-based reputation model assessment. To tackle threats scenarios the utility-based reputation model was further extended for by incorporating several additional components, in particular: assigning initial reputation to a new entity in VO, capturing alliance between consumer and resource, determining time decay function, and score function. Also the nonlinear compromise scheme is proposed for the reputation integration into service selection subsystem. Unlike the other known approaches the usage of the nonlinear compromise scheme provides better multicriteria evaluation that leads to the reject level reduce and to improvement of the system response speed. The efficiency of the proposed approach was analyzed with simulations that were run using data collected from the EGEE Grid-Observatory project. The information technology for service selection in service-oriented systems with the respect to reputation has been developed. The implementation of this information technology in Ukrainian National Grid resource center reduced the rejection rate by 10% and increased overall task completion. The implementation of this information technology in Ukrainian Education and Research Institute of Dataware and Telecommunication Support of Agroindustrial and Nature Conservative Branches of Economy essentially improved the QoS provided to consumers of service-oriented agromonitoring system. The scientific results acquired are also been used in several student courses at the NTUU "KPI".