An English-Korean transliteration model using pronunciation and contextual rules

Abstract

There is increasing concern about English-Korean (E-K) transliteration recently. In the previous works, direct converting methods from English alphabets to Korean alphabets were a main research topic. In this paper, we present an E-K transliteration model using pronunciation and contextual rules. Unlike the previous works, our method uses phonetic information such as phoneme and its context. We also use word formation information such as English words of Greek origin. With them, our method shows significant performance increase about 31% in word accuracy.

Compositional Machine Transliteration

“…In fact some of the best performing systems in the workshop were primarily grapheme based systems [Jiampojamarn et al 2009;Jansche and Sproat 2009;Oh et al 2009]. Further, combining any of the grapheme based engines with pre-processing modules like word-origin detection were shown to enhance the performance of the system [Oh and Choi 2002]. While previous research addressed combining evidence from multiple systems [Oh et al 2009], to the best of our knowledge, ours is the first attempt at combining transliteration evidence from multiple languages.…”

“…Under such frameworks, transliteration is treated as a conversion from source grapheme to source phoneme followed by a conversion from source phoneme to target grapheme. Hybrid models either use a combination of a grapheme based model and a phoneme based model [Stalls and Knight 1998] or capture the correspondence between source graphemes and source phonemes to produce target language graphemes [Oh and Choi 2002].…”

English to Kashmiri Transliteration System - A Hybrid Approach

“…Since it is not practically possible to find all Named Entities and OOVs in the English-Kashmiri transliterated word pair or in the CMD or AH Dictionary of English Language for direct or syllable based transliteration, some words may still count as a word-error [4]. For such words, a slower but fairly effective method is used that consists of changing a grapheme to its phoneme based on a syllable mapping.…”

Selection of Korean Proper Translation Words Using Bi-Gram-Based Histograms

“…Automatic transliteration can be implemented by direct and pivot-based translation (Oh & Choi, 2002). Previous studies tried to generate several possible candidate words based on pronunciation derived by dictionaries and statistical approaches such as Markov window and decision tree (Jung et al, 2000) (Lee, 2000) (Oh & Choi, 2002). They considered only English unabbreviated words that generate many possible transliteration candidates.…”

“…In this paper, we introduce the sub-sequent methods of managing translations in a real newspaper corpus of about 30 million Korean words: proper translation word finding and translation clustering. Automatic transliteration can be implemented by direct and pivot-based translation (Oh & Choi, 2002). Previous studies tried to generate several possible candidate words based on pronunciation derived by dictionaries and statistical approaches such as Markov window and decision tree (Jung et al, 2000) (Lee, 2000) (Oh & Choi, 2002).…”