GIATI: A general methodology; for finite-state translation using alignments

Abstract

Statistical techniques for machine translation have experienced an increasing interest by the natural language research community in the last years. Both statistical language modeling and statistical machine translation are now well-established disciplines with solid basis and outstanding results. On the other hand, finite-state transducers have revealed as an efficient and flexible formalism for the representation of a wide range of the kind of information that arises in natural language processing.This paper presents a powerful general framework for combining statistical techniques with grammatical inference and finite-state traducers. The GIATI methodology proposed here provides a schema for building inference algorithms that are able to generate finite - state transducers from parallel corpora of text making use of information supplied by robust statistical techniques such as n-grams and alignments. Here, the general method is presented together with two concrete inference algorithms and some experiments that show the validity of the GIATI framework for real - world translation tasks. © Springer-Verlag 2004.