English generation in Japanese-to-English machine translation

被引：0

作者：

NTT Communication Science Lab, Japan ^{[1
]}

机构：

来源：

NTT R&D | / 12卷 / 1419-1424期

关键词：

Linguistics;

D O I：

暂无

中图分类号：

学科分类号：

摘要：

We have been focusing research on two areas:word order, particularly of adverbial expressions; and the generation of number and determiners. To order adverbs correctly, we have classified them into 41 different types, based on function, meaning and usage and developed 7 ordering principles. Testing revealed a success rate of 98% (up from 86%). To generate determiners and number, we deduce information about a noun phrase's meaning in context, combined with default information about English noun countability. We have improved our accuracy to 87% (up from 65%).

引用