Targoman is a bilingual statistical machine translation system, established as a research program supported by
Iran Telecommunication Research Center
and represented by
Human Language Technology and Machine Learning lab of Amirkabir University of Technology
in 2011.
Targoman is capable to translate English texts to Persian and vice versa. Targoman is based on a dedicated SMT engine which is comparable to the famous OpenSource engine 'moses'. Targoman has been trained on a corpora of 38-Million words. Compared to other English-Persian machine trnalsation systems including Google, Bing, Farazin, etc, Targoman shows higher BLEU score as of 170% better BLEU score in literature domain compared to Google translation which is the second best.
Targoman system is continuously being updated and different features such as sentence suggestion, word pronunciation, and picture matching will be added to it in the near feature. In addition, the system’s translation quality and other related areas are under development, and it will gradually gain more coverage over both English and Persian languages.
Targoman SMT core has been published as an OpenSource with LGPLv3 Licence. Review and fork it on
https://github.com/Targoman