Government of Canada
Symbol of the Government of Canada

Common menu bar links

 
 
 

The trouble with SMT consistency

 
 
Affiliation:
Information and Communication Technologies; National Research Council Canada
Language:
English
Type:
Conference publication
Conference:
7th Workshop on Statistical Machine Translation, June 7-8, 2012, Montreal, Quebec
Proceedings
Title:
Proceedings of the 7th Workshop on Statistical Machine Translation
Date:
2012
Pages :
442-449
NPArC #:
20794314
Abstract:
SMT typically models translation at the sentence level, ignoring wider document context. Does this hurt the consistency of translated documents? Using a phrase-based SMT system in various data conditions, we show that SMT translates documents remarkably consistently, even without document knowledge. Nevertheless, translation inconsistencies often indicate translation errors. However, unlike in human translation, these errors are rarely due to terminology inconsistency. They are more often symptoms of deeper issues with SMT models instead.
 
Bookmark and Share:
 
 
 
 
Link:
HTML Link: