Hi All,
The Center for Indian Language Technology, IIT Bombay, is pleased to
announce the availability of *Brahmi-Net*, a transliteration and script
conversion system for languages of the Indian subcontinent. The system
supports transliteration between 18 languages (13 Indo-Aryan, 4 Dravidian
and English). REST and Python API for the transliteration system is also
available. The transliteration systems have been evaluated for their
utility in transliterating OOV words in translations among these languages.
The transliteration system was trained on transliteration pairs mined from
the Indian Language Corpora Initiative (ILCI) corpus, a parallel corpus of
11 Indian languages. *The transliteration mining was done using the
transliteration mining module implemented in Moses**.* The transliteration
corpus, constituting a total of about 1.69 million word pairs across 110
language pairs, is available for research under a Creative Commons
Attribution-NonCommercial-ShareAlike 4.0 International License.
You can access the system, corpora and documentation at:
http://www.cfilt.iitb.ac.in/brahminet/
You can read about Brahmi-Net here:
http://www.cfilt.iitb.ac.in/brahminet/static/publications/brahminet_naacl2015.pdf
We have a demo of *Brahmi-Net* at NAACL 2015.
Regards,
Anoop Kunchukuttan.
--
Department of Computer Science and Engineering, IIT
Bombayhttp://www.cse.iitb.ac.in/~anoopk
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
