Unsupervised Constraint Driven Learning for Transliteration Discovery

Ming-Wei Chang     Dan Goldwasser     Dan Roth     Yuancheng Tu    
North American Chapter of the Association for Computational Linguistics (NAACL), 2009
[pdf]

Abstract

This paper introduces a novel unsupervised constraint-driven learning algorithm for identifying named-entity (NE) transliterations in bilingual corpora. The proposed method does not require any annotated data or aligned corpora. Instead, it is bootstrapped using a simple resource – a romanization table. We show that this resource, when used in conjunction with constraints, can efficiently identify transliteration pairs. We evaluate the proposed method on transliterating English NEs to three different languages - Chinese, Russian and Hebrew. Our experiments show that constraint driven learning can significantly outperform existing unsupervised models and achieve competitive results to existing supervised models.


Bib Entry

  @InProceedings{CGRT_2009,
    author = "Ming-Wei Chang and Dan Goldwasser and Dan Roth and Yuancheng Tu",
    title = "Unsupervised Constraint Driven Learning for Transliteration Discovery",
    booktitle = "North American Chapter of the Association for Computational Linguistics (NAACL)",
    year = "2009"
  }