Learning to Find Transliteration on the Web

This prototype demonstrate a novel method for learning to find transliterations of proper nouns on the Web based on query expansion aimed at maximizing the probability of retrieving transliterations from existing search engines. Since the method we used involves learning the morphological relationships between names and their transliterations, we refer to this IR-based approach as morphological query expansion for machine transliteration. The morphological query expansion approach is general in scope and can be applied to translation and transliteration, but we focus on transliteration in this paper. Many texts containing proper names (e.g., “The cities of Mesopotamia prospered under Parthian and Sassanian rule.”) are submitted to machine translation services on the Web every day, and there are also service on the Web specifically target transliteration of proper names, including CHINET (Kwok et al. 2005) ad Livetrans (Lu, Chien, and Lee 2004). Machine translation systems on the Web such as Yahoo Translate (babelfish.yahoo.com) and Google Translate (translate.google.com/translate_t.g) typically use a bilingual dictionary that is either manually compiled or learned from a parallel corpus. However, such dictionaries often have insufficient coverage of proper names and technical terms, leading to poor translation due to out of vocabulary problem. The OOV problems of machine translation or cross language information retrieval can be handled more effectively by learning to find transliteration on the Web. Consider Sentence 1 containing three place names. 1. The cities of Mesopotamia prospered under Parthian and Sassanian rule. 2. 城市繁榮下parthian 達米亞、sassanian統