Identifying collocations using cross-lingual association measures
Lis Kanashiro Pereira, Elga Strafella, Kevin Duh, Yūji Matsumoto · 2014
We introduce a simple and effective cross-lingual approach to identifying colloca-tions. This approach is based on the obser-vation that true collocations, which cannot be translated word for word, will exhibit very different association scores before and after literal translation. Our exper-iments in Japanese demonstrate that our cross-lingual association measure can suc-cessfully exploit the combination of bilin-gual dictionary and large monolingual cor-pora, outperforming monolingual associa-tion measures. 1