Summary: Term extraction: A Review
Draft Version 091221
Department of Computer and Information Science
This is a review of the current state-of-the-art in term extraction. We consider
both mono-lingual and multi-lingual term extraction, and single-word as well as
multi-word terms. The review is restricted to methods, however, and so does not
make any attempt to cover commercial systems. Readers interested in commercial
systems are referred to Zielinski & Ramirez Safar (2005) and the continuously
updated Compendium of Translation Software1.
First, however, we delimit the scope of the paper and propose a working defi-
nition of term extraction.
2 On terms and term extraction
In classical terminology a term is defined as the expression (or label, or represen-
tation) of a concept2. This seems to imply that we cannot approach the problem of
term extraction without knowing what a concept is, or whether a certain text seg-
ment is the expression of a concept or not. However, this is a situation we certainly