English-Hindi Transliteration by Applying Finite Rules to Data before Training Using Statistical Machine Translation

In this paper, we have presented that by applying finite rules of transliteration to the data before training, the transliteration result get remarkably increased. The rules are applied to bilingual corpus of Indian Name Entities with source in English and target in Hindi (Devnagri) script. We have incorporated this concept to an existing machine translation system name Anusaaraka. The efficiency is increased by 29.1% compared to the system without applying finite rules before training.