other,6-3-J05-4003,ak Using this approach , we extract <term> parallel datafrom </term> large Chinese , Arabic , and English <term> non-parallel newspaper corpora </term> .
hide detail