This work approaches the problem of discovering atomic web services that will realize complex business processes in an adaptive information system. It is proposed a model for semantic description of web services and user profile and the design of a semantic recommender engine based on this model. The recommender engine performs, during the web service discovery phase, a "similarity evaluation" step in which it can be possible to estimate the similarity between what the service offers and what the user prefers. A semantic algorithm, that measures distance between concepts in an ontology, is used to rank the results of the semantic matching between the user profile and a list of web services, suggesting to the user the most suitable services.