Automating the process of term recognition and classification is important for digital libraries. Automatic Term Recognition (ATR) has many applications in areas related to dig- ital libraries, e.g. information retrieval and extraction from the web, summarisation, machine translation, dictionary construction etc. Automatic term classification attracts the interest of researchers following the steps of ATR. By automatic term classification we mean the grouping of terms into sets whose elements share conceptual properties. In this paper we present C-value/NC-value a method for the automatic extraction of terms, and its further use for the recognition of similarities between them, to be used for them classification.