Kyonghee Paik
Using Language-specific Characteristics to generate a
Korean-to-Japanese Dictionary
Abstract:For machine translation, a bilingual transfer dictionary is a most important
resource. There are many ways of dictionary building. In this paper, we propose
a simple method to make a dictionary using intermediate means, taking
advantage of accessible resources such as English-to-X dictionaries and
X-to-English dictionaries. Also, we focus on how we can improve the existing
method to generate Korean-to-Japanese dictionary automatically. Since Korean
and Japanese share Chinese characters for the majority of words, we use them
as well. We argue that this "multi-pivot criterion" is useful to build dictionaries
especially for the languages using Chinese characters. However, this method can
be applied to building dictionaries of other language combinations with
consideration of language-specific characteristics.