Service-oriented computing provides an evolving paradigm for flexible and scalable applications of open systems. Web services and their automatic composition are in the mainstream of the evolution of new value-added services. Functional and non-functional aspects are considered together for automatic service composition (ASC). After locating suitable functionality for the required composition, non-functionalities are considered to select the final set of services. Non-functional properties (NFPs) obtained from users or identified during or discovery usually have abstract concepts that cannot be identified at the selection stage. In this paper, we propose a transformation technique for automatic composition that identifies binding information (service I/O) in the n stage from intermediate-abstraction NFPs. The classification of ion level in NFPs, a model to define abstract and concrete NFPs, and an algorithm for transformation from intermediate to concrete level are presented. The iden...