There has been an increasing interest in Semantic Web services (SWS) as a proposed solution to facilitate automatic discovery, composition and deployment of existing syntactic Web services. Successful implementation and wider adoption of SWS by research and industry are, however, profoundly based on the existence of effective and easy to use methods for service semantic description. Unfortunately, Web service semantic annotation is currently performed by manual means. Manual annotation is a difficult, error-prone and time-consuming task and few approaches exist aiming to semi-automate that task. Existing approaches are difficult to use since they require ontology building. Moreover, these approaches employ ineffective matching methods and suffer from the Low Percentage Problem. The latter problem happens when a small number of service elements - in comparison to the total number of elements – are annotated in a given service. This research addresses the Web services annotation problem by developing a semi-automatic annotation approach that allows SWS developers to effectively and easily annotate their syntactic services. The proposed approach does not require application ontologies to model service semantics. Instead, a standard query template is used: This template is filled with data and semantics extracted from WSDL files in order to produce query instances. The input of the annotation approach is the WSDL file of a candidate service and a set of ontologies. The output is an annotated WSDL file. The proposed approach is composed of five phases: (1) Concept extraction; (2) concept filtering and query filling; (3) query execution; (4) results assessment; and (5) SAWSDL annotation. The query execution engine makes use of name-based and structural matching techniques. The name-based matching is carried out by CN-Match which is a novel matching method and tool that is developed and evaluated in this research. The proposed annotation approach is evaluated using a set of existing Web services and ontologies. Precision (P), Recall (R), F-Measure (F) and Percentage of annotated elements are used as evaluation metrics. The evaluation reveals that the proposed approach is effective since - in relation to manual results - accurate and almost complete annotation results are obtained. In addition, high percentage of annotated elements is achieved using the proposed approach because it makes use of effective ontology extension mechanisms.