옛한글 형태소 사전 구축을 위한 기법 연구

Abstract

Morphological analysis of document written in a natural language provides information for syntactic analyzing and semantic analyzing of them, which is essential process to construct various applications based on natural language process such as information retrieval, machine translation, and semantic analysis. For that reason, many morphological analyzers and morphological dictionaries for processing document written in modern korean have been constructed and have been used in industry and academia. However, since the necessary of processing documents written in old-korean is limited, morphological analyzer and morphological dictionary for old-korean has not been developed yet. Especially, constructing a old-korean morphological dictionary is known as a difficult task that takes a lot of time and effort because of the complex transformation of korean language. To solve the problem, in this paper we devise a technique to construct a old-korean morphological dictionary by using old-modern korean parallel documents and modern korean morphological analyzer and modern korean morphological dictionary. In the proposed technique, after morphological analyzing a document written in modern-korean, a produced modern-korean morphology is searched in a parallel document written in old-korean by using similarity measure. And if a old-korean morphology corresponding the modern-korean morphology is found, the morphology is registered in a old-korean morphological dictionary. In the process, the difference of writing a character between modern-koran and old-korean makes it difficult to measure similarity. In this paper, we showed a method that solves the problem by using Levenshtein distance and consonant-vowel transformation table. The result of conducting an experiment with Hong Gildong jeon written in old-korean showed that 78.3% of old-korean morphologies were extracted automatically.