engleski

Comparative Analysis of Automatic Term and Collocation Extraction

Monolingual and multilingual terminology and collocation bases, covering a specific domain, used independently or integrated with other resources, have become a valuable electronic resource. Building of such resources could be assisted by automatic term extraction tools, combining statistical and linguistic approaches. In this paper, the research on term extraction from monolingual corpus is presented. The corpus consists of publicly accessible English legislative documents. In the paper, results of two hybrid approaches are compared: extraction using the TermeX tool and an automatic statistical extraction procedure followed by linguistic filtering through the open source linguistic engineering tool. The results have been elaborated through statistical measures of precision, recall, and F-measure.

automatic extraction; term and collocation base; English language; evaluation metrics

nije evidentirano