Expanding the Language model in a low-resource hybrid MT system

Hits: 4451
Research areas: Year: 2014
Type of Publication: In Proceedings
Editor: D. Wu, M. Carpuat, X. Carreras, E.M. Vecchi
Book title: Proceedings of the Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation (SSST-8)
Pages: 57-66
Address: Doha, Qatar
Organization: SSST-8 Workshop [held within EMNLP2014] Month: October 25
ISBN: 978-1-937284-96-1
The present article investigates the fusion of different language models to improve translation accuracy. A hybrid MT system, recently developed in the European Commission-funded PRESEMT project that combines example-based MT and Statistical MT principles is used as a starting point. In this article, the syntactically-defined phrasal language models (NPs, VPs etc.) used by this MT system are supplemented by n-gram language models to improve translation accuracy. For specific structural patterns, n-gram statistics are consulted to determine whether the pattern instantiations are corroborated. Experiments indicate improvements in translation accuracy.