Quality and Relation Driven Service Selection for Web Services Composition

In service oriented application, value-added service is composed of some component services which are selected from candidate services. However, in the current service selection models, services in the composite service are independent from each other. In fact, there are relations (e.g. data interaction, condition dependency) between them. Both QoS and service relation should be regarded as criterions for selecting a set of services to construct a composite service. This paper presents a semantic relation model for the service composition. Based on the model, service selection is formulated as a multi-objective optimization problem. Genetic algorithms (GA) is used to search for a solution. Experimental results show that exploiting relations between services can further enhance the interoperability and usability of produced compositions.