In this paper, we describe the process by which web services ontologies are populated from a web services collection. The general approach relies on a global ontology model that is used to represent automatically web services. The model is enriched with web service instances classified into a taxonomy. The main idea is to extract taxonomic relations (isTypeOf) from web services using a supervised classifier of textual descriptions attached to web services. The entire process for ontology population involves the following tasks: text extraction from web service descriptions, classification of text descriptions and extraction of taxonomic relations (instances of classified web services). An experimentation was carried out with a collection of web service, which shows promising results and the feasibility of our approach.