Up

A Reliable Punjabi to English Machine Transliteration System for Proper Noun

File Size:
1.85 MB
Volume:
Volume 3, Issue 3 (March, 2017)
Publication No:
IJTC201703005
Author:
Atul Katoch, Dr.Raman Chadha
Downloads:
9 x

Abstract:
Machine Transliteration has come out to be an emerging and very important part of NLP, which is not only concerned with representation of sounds in original, the characters, but optimally accurately and explicitly. Transliteration means to represent the characters of source language by the characters of target language, although keeping the action reversible. The main target of transliteration is to perpetuate the linguistic structure of the words. Appropriate transliteration of name entities play integral role in upgrading the characteristics of machine translation. Language Transliteration plays a significant role in various research areas like machine transliteration (MT) and cross-language information retrieval (CLIR) processes. The design of transliteration model is such like the articulate structure of words should be conserved as closely as possible. Two contrasting languages are considered in this case, one is Punjabi and another is English. There are numerous machine transliteration models used for Transliteration. Each model has peculiar requirements for implementation. After studying number of work done by various researchers in the   area, we   have   developed algorithm based on statistical machine transliteration from Punjabi to English and the accuracy appeared to be approximately 95.82%.

Keywords:
Rule based approach, Statistical Machine Translation approach (SMT), Transliteration, Natural Language Processing (NLP).