System
IMPACT is a system that evaluates the output of machine translation systems by automatically measuring similarity between them and (one or more) reference translations, and is effective in automatic evaluation of segments (sentences), not only documents. IMPACT shows the scores between 0.0 and 1.0. The evaluation for the output of machine translation system is perfect when the score is 1.0.
Requirements
Windows (DOS prompt) or Linux
IMPACT (include README, System, Sample Data)
papers
- Hiroshi Echizen-ya, Kenji Araki : Automatic Evaluation of Machine Translation based on Recursive Acquisition of an Intuitive Common Parts Continuum, Proceedings of the Eleventh Machine Translation Summit (MT SUMMIT XI), Page.151-158, Copenhagen, Denmark, 2007.9 [PDF]
- Hiroshi Echizen-ya, Terumasa Ehara, Sayori Shimohata, Atsushi Fujii, Masao Utiyama, Mikio Yamamoto, Takehito Utsuro, Noriko Kando : Meta-Evaluation of Automatic Evaluation Methods for Machine Translation using Patent Translation Data in NTCIR-7, Proceedings of the 3rd Workshop on Patent Translation, Page.9-16, Ottawa, Canada, 2009.8 [PDF]
- Hiroshi Echizen-ya, Kenji Araki : Automatic Evaluation Method for Machine Translation using Noun-Phrase Chunking, Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics (ACL 2010), Page.108-117, Uppsala, Sweden, 2010.7 [PDF]
- Hiroshi Echizen'ya, Kenji Araki, Eduard Hovy : Optimization for Efficient Determination of Chunk in Automatic Evaluation for Machine Translation, Proceedings of the 1th International Workshop on Optimization Techniques for Human Language Technology (OPTHLT 2012) / COLING 2012, Page.17-30, Mumbai, India, 2012.12 [PDF]