Transliteration Verification of English-Hindi Words using Machine Learning Approach

Automatic transliteration and back-transliteration across languages with different phonemes and alphabets inventories such as English-Hindi, English-Chinese, and English-Korean and few more have practical importance in cross lingual information retrieval, machine translation and automatic bilingual dictionary compilation. Proper name pronunciation based translation termed as transliteration, for many multilingual natural language processing task such as cross lingual information retrieval and statistical machine translation is very important in this modernised world. Due to the pronunciation difference between the source and target language, this task is very challenging. In past for transliteration, research efforts has demonstrated a 30 40 percent error using the top-1 reference. For many applications, this error leads to performance degradation. In asian countries especially in India, there has been very less work which has been done in the field of English-Hindi transliteration. The motive of this research, is to develop a novel and an efficient model , to verify a given proper name transliteration using different machine learning approach