Given a corpus of documents in which terms have been annotated and a classification scheme, the goal of term classification is to map individual term occurrences to the corresponding classes in the given scheme. We, therefore, do not classify terms in general, but rather their specific realisations in the given documents. Such an approach is context-sensitive and as such could also be used to perform term sense disambiguation. On the other hand, the results of classifying individual term occurrences may be used for general term classification by assigning a term to all or the most frequently predicted classes (the ones inferred for its individual occurrences).