Design Tools for Japanese-to-English Machine Translation Dictionaries

Hajime UCHINO*    Masahiko HARUNO**    Yamato TAKAHASHI*    Satoshi SHIRAI*


Abstract

Large bilingual corpora are useful for improving the quality of machine translations, but only small bilingual corpora exist. This paper describes a method of building a large bilingual corpus by aligning Japanese sentences with English sentences from newspaper articles which can be easily and continually collected. A corpus analysis method by using n-gram statistics and computer-aided design tools of machine translation dictionaries are also described.



[ NTT R&D, Vol.46, No.12, pp.1425-1432 (December, 1997). ]



* NTT Communication Science Laboratories
** ATR Human Information Processing Research Laboratories