Constraint Optimization Approach to Context Based Word Selection
Jun Matsuno and Toru Ishida
Consistent word selection in machine translation has been realized by resolving word sense ambiguity based on the context of a single sentence or neighboring sentences. However, consistent word selection over the whole article has not been achieved yet. Consistency over the whole article becomes quite important when applying machine translation to collectively developed documents like Wikipedia. In this paper, we propose to consider constraints between words in the whole article based on their semantic relatedness and contextual distance. The proposed method has been successfully implemented in both statistical and rule-based translators. We evaluated those systems by translating 100 articles in the English Wikipedia into Japanese. The results showed that the ratio of appropriate word selection for common nouns increased to around 75% with our method, while it was around 55% without our method.