근대계몽기 잡지 자연어 전처리 작업에서 문체 분류와 형태분석 방안에 대한 모색

Abstract

This paper classifies the style and discusses natural language processing problems presented by eleven academic journals published during the modern Enlightenment period which have been digitized to the Korean history database. This is a basic step in studying these journals, which deploy various types of stylistic formats, including Chinese, Korean, and Korean written in Chinese characters. These stylistic complexities are significant obstacles which prevent accurate data extraction by means of digital humanities methodologies based upon word processing. Journal articles were initially classified into ten types, and those articles which predominantly used Chinese words with Korean grammatical patterns were selected for further research. This style was typically used by articles which translated or transcribed academic content, and also for articles which described new or revised legal codes. Thus, if digital humanities methodology is applied to articles selected by this stylistic criterion, research results tend to be more tightly focused upon the introduction of modern education and knowledge, and upon new concepts related to the legal, political, and economic fields. The method of discourse analysis used by this study, in which stylistic criteria are used to segment the texts found in journals of modern Enlightenment for further analysis, is a useful and appropriate tool, since the various types of modern Enlightenment material are readily and recognizably separable on the basis of their writing style: each type tends toward a characteristic format. This segmentation also allows a clearer and more focused discussion of how to process the target data for each style, resulting in an increased accuracy for the data extracted, and a more effective understanding of the way in which the worldview of the authors is revealed by their stylistic choices. In addition, this paper discusses the morphological analysis of the Chinese words, especially the processing of vocabulary in complex relations and derivative relations. The general criteria used in Korean linguistic analysis are inadequate for this kind of morphological analysis, which needs to be adapted to the specific research aims of a particular study.