May 2004 | Liangzhao Zeng, Boualem Benatallah, Member, IEEE, Anne H.H. Ngu, Marlon Dumas, Member, IEEE Computer Society, Jayant Kalagnanam, and Henry Chang
The paper presents a middleware platform, AgFlow, designed to address the challenge of selecting Web services for composition in a way that maximizes user satisfaction expressed as utility functions over Quality of Service (QoS) attributes, while satisfying user constraints and the structure of the composite service. The platform employs a multidimensional QoS model to evaluate Web services and uses two selection approaches: local optimization and global planning. Local optimization selects services for individual tasks without considering QoS constraints across tasks, while global planning uses integer programming to optimize the selection of services for the entire composite service, taking into account QoS constraints and preferences. The adaptive execution engine of AgFlow can adjust the execution plan during runtime to ensure optimal QoS given the available information about the component services. The paper also discusses the system architecture, service ontologies, composite service specifications, and the quality criteria used in the QoS model.The paper presents a middleware platform, AgFlow, designed to address the challenge of selecting Web services for composition in a way that maximizes user satisfaction expressed as utility functions over Quality of Service (QoS) attributes, while satisfying user constraints and the structure of the composite service. The platform employs a multidimensional QoS model to evaluate Web services and uses two selection approaches: local optimization and global planning. Local optimization selects services for individual tasks without considering QoS constraints across tasks, while global planning uses integer programming to optimize the selection of services for the entire composite service, taking into account QoS constraints and preferences. The adaptive execution engine of AgFlow can adjust the execution plan during runtime to ensure optimal QoS given the available information about the component services. The paper also discusses the system architecture, service ontologies, composite service specifications, and the quality criteria used in the QoS model.