Effect of the Topic Dependent Translation Models for Patent Translation - Experiment at NTCIR-7
Takeshi Ito, Tomoyosi Akiba, Katunobu Itou · 2008
In this paper, we investigate the effect of the topic dependent translation model for patent translation. We employ clustering technique to estimate topics in the training corpus and document retrieval to identify the topic fitting to the source sentence. In our exper-imental evaluation, we investigate the contribution of our topic dependent models to phrase-base Statistical Machine Translation.