To support automated Web service composition, it is compelling to provide a template, or model that dictates the ways in which services can be composed. In this paper, a novel composition model based on the relative vector is proposed, where individual services, composite services, and user objectives are described with the relative vectors, through a series of operators defined on the relative vector, available composite services can be found, and how much they meet user objectives can be evaluated. A significant advantage of our approach is that many existing optimization methods can be used to search optimized compositions, where parallel or choice structure is enabled, individual service with multiple input or output parameters is allowed.
