Individual needs for sharing information take more and more importance. The easy access to social media services encourages people to produce informal communication. Many research efforts are devoted to the analyze of such kinds of data, potentialy imprecise or ambiguous. In this paper we propose a way to help people to write applications for gathering and indexing information produced on Twitter, mainly into tweets. This approach involves two elements. The first one is the description of a system able to semantically index tweets in a distributed structure. The second element is the definition of the category of semantic tweets which contain semantic references. The last section of this paper describes an application able to insert semantic hash tags into tweets in a manner completely transparent to users.