This work package aims at implementing a tool that categorizes texts in the target languages and can be tuned to heterogeneous domains. The ultimate goal of the automatic categorization is a qualitatively better and more effective organization of content in ATLAS and respectively in the online services. It is also important to minimize the effort to add a new language to the categorization component, as well as minimize the requirement of linguistic analysis components (i.e. lemmatizers, thesaurus based hypernym lookup, general and domain specific dictionaries, etc.)