Automatic construction of a bilingual thesaurus using citation analysis

Hidetsugu Nanba, Saori Mayumi, Toshiyuki Takezawa · 2011

We propose a method for constructing a bilingual thesaurus automatically from patents. First, we extract hypernym-hyponym relations from Japanese and US patents by using the pattern "A such as B". Second, we align terms between these thesauri by combining statistical machine translation and citation analysis techniques. To confirm the effectiveness of our method, we conducted some experiments. The results showed that our best method obtained Recall of 79.4%, Precision of 77.5%, and F-measure of 78.3%.

Read the paper · More papers on PaperTik