Automatic Annotation for Korean--Approach Based on the Contextual Exploration Method

We present an automatic semantic annotation system for Korean based on the contextual exploration method. Creating a morphological analyzer and part-of-speech tagger for the Korean language is difficult as it is a highly agglutinative language. Accordingly, processing Korean in the same order as inflectional languages - morphological analysis, then syntactical and then semantic - has not yielded satisfactory results. Our new method identifies semantic information in Korean text without going through the morphological and syntactical analysis steps. Our initial system properly annotates approximately 88% of standard Korean sentences, and this annotation rate holds across text domains. Previously, the contextual exploration method has been applied successfully to languages as diverse as French and Arabic. Given our success with Korean, we believe that this method can be applied to other agglutinative languages such as Japanese, Turkish and Finnish.