Robust Transliteration Mining from Comparable Corpora with Bilingual Topic Models

We present a high-precision, languageindependent transliteration framework applicable to bilingual lexicon extraction. Our approach is to employ a bilingual topic model to enhance the output of a state-of-the-art graphemebased transliteration baseline. We demonstrate that this method is able to extract a high-quality bilingual lexicon from a comparable corpus, and we extend the topic model to propose a solution to the out-of-domain problem.