A Framework For Phrase Based SMT In The Technical Translation Domain

Abstract: This thesis is interested in researching the statistical machine translation approach and trying to apply it to the problem of the technical translation domain. We proposed a statistical machine translation system and our experiments using a small corpus of size 20,000 sentences suggested that this system outperforms the well-established word based statistical machine translation system. In a small experiment, we showed that the output of the proposed system is better than the suggestions CAT tools supply to the human translator, and we suggest a new architecture to replace the fuzzy match suggestions found in the available commercial CAT tools. In future work we need to do more experimentation using the new suggested architecture for the CAT tools. Also we need to enhance the statistical model used in our machine translation system by adding a syntax language model and to experiment the effect of this language model on the performance of the system.
URL
Publication year 2004
Organization Name
External authors (outside ARC)
    Hisham Kodeir
Agris Categories Documentation and information
Proposed Agrovoc Statistical Machine Translation;
Publication Type Master Thesis