Japanese-Hungarian dictionary generation using ontology resources

In this paper we describe an automatized dictionary generation method that can be applied with most language pairs. As an example, we created a Japanese-Hungarian dictionary. Our approach is a pivot language based method, with English as the intermediate language. We use the Japanese-English and English-Hungarian dictionaries to create a set of translation candidates, for correct translation pair selection we use the English WordNet as a main resource. Our purpose is to achieve a high precision dictionary with a maximized recall. Evaluations showed an improved recall and precision when compared with two conventional methods, showing that ontology based selection can produce better results than dictionaries based methods.