The Mass Media involves information and communication products targeted to a wide audience. Today such communications products are also available on Internet where people can react to a given information by posting critics, congratulations, opinions or whatever they want via social networks. Such reactions are considered valuable information for instance to government and companies. However, this information is hard to automatically process as people commonly use ironies, stereotypes, metaphors expressed in informal writing plenty of chat abbreviations, emoticons, and slang words. In this paper, we illustrate how tools based on discursive categories can be used to analyze such reactions and thus to process and understand the information behind them.