Japanese Term Extraction Toward French-Japanese Bilingual Term Extraction on Wind Power Generation Domain

摘要

Pattern based term extraction is a basic approach to collect terms from texts without learning corpus. We show the experimental results of several frequency-based term extraction approaches on two different corpora: the one is NTCIR-1 term extraction corpus, i.e., abstracts of papers in Japanese artificial intelligence; the other is biomedical corpus constructed for biomedical term extraction. After discussing possibility of term extraction from monolingual corpus, we show preliminary experimental results of French-Japanese bilingual term extraction from comparable corpora on Wind Power Generation and then discuss perspective of our approach.