- From: Gavin Nicol <gtn@ebt.com>
- Date: Fri, 30 Dec 1994 10:03:26 -0500
- To: glenn@stonehand.com
- Cc: www@unicode.org, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
>Unless you store bunsetsu boundaries and yomi (phonetic reading) along >with the original text, you can pretty much forget about automatic >transliteration of Japanese (or Chinese for that matter). Well, 100% accurate transliteration, or transcription, like 100% accurate machine translation, is one of those "hard" problems. I have seen systems that can take a paragraph of kana and convert it into kanji, or take a paragraph of kanji, and convert it into kana. The accuracy is about 98%. I have also seen machine translation systems working on PC's that can achieve readable, if not grammatically correct English and Japanese. Resource usage for the above systems is still fairly high: about on par with a typical Microsoft application in fact :-) Not impossible, not impractical, but not 100% perfect either. Hmm, sounds familiar...
Received on Friday, 30 December 1994 07:03:04 UTC