Functional Expansion of Morphological Analyzer For Efficient Korean Parsing

Abstract

Korean has the advantage of omission of sentence structure and its range of formulas, so it is better to process it in morpheme analysis stage than parsing. In this paper, we propose a functional expansion method of morpheme analyzer which can reduce the burden of parsing. This method combines them into the longest matching method when a plurality of morpheme columns have a syntactic category by the estimation of unknown words, processing of compound nouns and compound verbs, processing of numbers and symbols, The proposed morpheme analysis method improves accuracy by eliminating unnecessary morphological analysis results. Experimental results show that the average parse tree is 73.4% and the parse time is 52.9%.