Hybrid Filtering for Extraction of Term Candidates from German Technical Texts

Most of the current methodologies for automatic term extraction rely heavily on morpho-syntactic criterion in identification of term candidates, and are mainly developed for English. This in turn makes it difficult to apply these techniques for German texts directly due to the morpho-syntactic differences between the two languages. In this paper, we present an approach to automatic term extraction which takes into account the characteristics of German language, and attempts to combine different filtering techniques (linguistic and statistical). The current work is part of an on-going research at IAI on the development of multilingual document production/management tool MULTILINT which has already been successfully employed by various international enterprises