—The amount of information exchanged over the Internet is continuously growing, taking the form of short text messages on microblogging platforms such as Twitter. Due to the limited size of these types of messages, their understanding may require to know the context of their occurrence. In this paper, we propose a higher-level representation of short text messages based on a thematic model obtained by a Latent Dirichlet Allocation (LDA). We propose to evaluate the effectiveness of this short text message representation by using it in the experimental setup of the INEX 2012 tweet contextu-alization task. This topic-based representation allows to extend the message vocabulary by searching a set of thematically-related words. Results demonstrated the interest of this topic-space based approach for the tweet contextualization task.