Extraction of information related to the medication is an important task within the biomedical area. Our method is applied to different types of documents in three languages. The results indicate that our approach can efficiently update and enrich the existing drug vocabularies.