View article

(in English) Wikipedia is a large-scale multilingual encyclopedia and therefore a promising resource for bilingual terminology extraction. However, current approaches translate only terms that are represented by a Wikipedia article. We believe that even if the article is missing, Wikipedia still contains valuable translation information. We propose a method that analyzes Wikipedia article texts and links to extract translation candidates, and uses Web search engine results to evaluate them. An experiment using terms extracted from a Japanese-English medical dictionary shows that our approach achieves high accuracy and coverage.