Exploring terms and their linguistic environment in text

The proliferation of specialist texts over recent decades has exacerbated the need for term extraction software to assist terminologists in compiling terminology collections. To this end, an automated approach to English term extraction is presented, which, in keeping with the multidisciplinary working environments of many contemporary terminologists, is designed to be domain independent. Based on observations made of the linguistic features of terms and their linguistic environment in text, this approach identifies single- and multi-word terms spanning a range of word classes. An implementation of the approach (denoted ‘Textprobe’) is described and evaluated by measuring its term extraction efficiency against the manual scanning output of both domain experts and terminologists. Results obtained in the evaluation suggest that a high proportion of single-and multi-word terms can successfully be extracted from special language texts. It is anticipated that the approach will be portable to other European languages.