Korean Adverb Ordering in English-Korean Machine Translation Using Clustering

This paper proposes an approach to determine the ordering of Korean adverb by using clustering method for making sentences more natural at the generation stage of English-Korean machine translation system. After observing the feature information of Korean adverb classified by scholars of Korean literature, we analyze an adverb ordering about the feature information. Afterwards, we extract consecutive adverbs and their frequencies from KAIST corpus and then provide feature information. With feature information, we calculate adverb-adverb similarity and re-classify Korean adverbs based on these similarities. Finally, we show the ordering rate of adverbs extracted in a corpus, and their detailed adverb ordering through the clustering method.