In the field of Semantic Web Services, the automatic composition of services has been a challenging research problem. At the same time, mainly due to increasing number of available services and their heterogeneity, the industry has demanded for techniques that automate the process of selection and composition of services. Thus, several techniques for semantic service composition have been developed, but these techniques have shown to considerably hinder the overall performance of the application, making their use very expensive. One of the factors that contribute to this performance problem is the way that planners check the similarity between the ontologies concepts, i.e. the Concepts Matching. Thus, the choice of a proper technique for Concepts Matching directly influences the overall performance of the system and the choice of a proper technique is considered a key design decision in semantic web services. In this context, we propose the use of a Concepts Matching technique based on a Synonyms Dictionary approach aiming to reduce the performance issues found in service composition process. To validate our proposal, we present a systematic experimental analysis which compares this technique with other two approaches. The results show that the Synonyms Dictionary technique has the best performance when compared to other techniques.