A Web service is a software functionality accessible through the network.Web services are intended to be composed to realize domain-specific applications. Achieving a required functionality by a service composition necessitates the discovery of a collection of Web services out of the enormous service space. Each service from this service space must be examined to verify its functionality, which makes the discovery task neither efficient nor practical. Moreover, when a service in a composition becomes unavailable or functionally broken, the whole composition may become broken too. Therefore, a functionally equivalent service must be discovered, in order to replace the broken service, thus spending more time and effort. In this paper, we propose an approach that facilitates the discovery of a web service and the identification of its candidate substitutes. We use many-valued concept lattices to classify web services, depending on the similarity estimated between their operations. This classification enables the identification of a needed web service as well as its possible alternatives. Thus, a Web service composition can be achieved more easily and can be supported with backup services, to recover the functionality of a broken service.