One current research topic in Knowledge Discovery is the analysis of the information provided by users in Web 2.0 social applications. In particular, some authors have devoted their attention to the analysis of micro-blogging messages in platforms like Twitter. A common shortcoming of most of the works in this field is their focus on a purely syntactical analysis. It can be argued that a proper semantic treatment of social tags should lead to more structured, meaningful and useful results that a mere syntactic-based approach. This work reports the analysis of a case study on medical tweets, in which the results of a semantic clustering process over a set of hashtags is shown to provide much better results than a clustering based on their syntactic co-occurrence.