Translation of named entities (Nes),such as person names, organization namesand location names is crucial for cross lingualinformation retrieval, machine translation,and many other natural languageprocessing applications. Newly named entitiesare introduced on daily basis innewswire and this greatly complicates thetranslation task. Also, while some namescan be translated, others must be transliterated,and, still, others are mixed. In thispaper we introduce an integrated approachfor named entity translation deployingphrase-based translation, word-based translation,and transliteration modules into asingle framework. While Arabic based,the approach introduced here is a unifiedapproach that can be applied to NE translationfor any language pair.
展开▼
