A HYBRID APPROACH OF ENGLISH- HINDI NAMED-ENTITY TRANSLITERATION
Vaishnavi Singhal, Neha Tyagi · 2015
In recent years, machine transliteration has gained a center of attention for research. Both machine translation and transliteration are important for e-governance and web based online multilingual applications. As machine translation translate source language to target language which results in wrong translation for named entities. Named entities are required to be translated with preserving their phonetic properties. Thus we need namedentity transliteration is utmost required. In this paper, the main focus is on the English-to-Hindi Named-Entity transliteration and a hybrid approach is proposed. The main problem arises during the transliteration of English-Hindi is the possibility of combination of English alphabets to Hindi akshara. There is no specific rule set has been designed to convert Hindi akshara to particular English syllable yet. To solve this issue a hybrid approach has been proposed where syllabification and the uni-gram model is used. The syllabification is done based on rule-based approach. This approach is first syllabifies the English name into appropriate syllable using rule based approach which is termed as syllabification. Then syllables are matched into particular Hindi akshara on the basis of corpora that is designed on the basis of English-Hindi Name-pairs knowledgebase.