Mining New Translation Lexicons from Wikipedia
Jong–Hoon Oh, Daisuke Kawahara, Kiyotaka Uchimoto, Hitoshi Isahara · 2007
In this paper, we propose a method of mining new translation lexicons from Wikipedia. To do this, first a network is constructed with nodes representing Wikipedia articles and links representing the dependencies between the articles. Then we learn a model of translation lexicons by investigating co-occurrence patterns of the existing translation lexicons in a Wikipedia network. Finally, new translation lexicons can be found by estimating the cross-lingual similarities between the nodes in different languages. Experiments show that our method effectively finds new translation lexicons.