부사의 통계 정보를 이용한 한국어 의존구문분석

The adverbs are constituents among the sentence causing syntactic ambiguity in syntactic analysis. This paper progress a method that using statistical information of the adverbs extracted from a corpus when syntactic analyzing. Statistical information of the adverbs is used to adjust weight of dependents and heads. Online learning algorithm was used for machine learning and 50,000 of 53,000 sentences from Sejong tree tagged corpus is used for training. The accuracy was measured by 3,000 of 53,000 sentences which is rest of training data used. The result of experiment was that 87.9% accuracy showed without applying the adverb informations. On the other hand, the case of applying the adverb was seen 88.2% accuracy that the accuracy rose compared with no adverb informations.