한국어 역사 소설에서 공간적 배경 인식 기법

Abstract

Abstract Submitted：February 1, 2016 1 st Revision：March 6, 2016 Accepted：March 10, 2016*본 연구는 문화체육관광부 및 한국콘텐츠진흥원의 2015년도 문화 콘텐츠산업기술지원사업(자유공모)의 연구결과로 수행되었음.** 단국대학교 대학원 컴퓨터학과 석사과정*** 단국대학교 응용컴퓨터공학과 교수, 교신저자Background in a novel is most important elements with character s and events, and means time, place and situation that characters appeared. Among the background, spatial backgro und can help conveys topic of a novel. So, it may be helpful for choosing a novel that readers want to read. In t his paper, we are targeting Korean historical novels. In case of English text, It can be recognize spatial background easily because it use upper and lower case and words used with the spatial information such as Bank, University and City. But, in case Korean text, it is difficult to recognize that spatial background because there is few information about usage of letter. In the previous studies, they use machine learning or dictionaries and rules to recognize about s patial information in text such as news and text messages. In this paper, we build a nation dictionaries that re fer to information such as 'Korean history' and 'Google maps.' We Also propose a method for recognizing spatial backgro und based on patterns of postposition in Korean sentences comparing to previous works. We are grasp using of po stposition with spatial background because Korean characteristics. And we propose a method based on result of mor pheme analyze and frequency in a novel text for raising accuracy about recognizing spatial background. The reco gnized spatial background can help readers to grasp the atmosphere of a novel and to understand the events and atmo sphere through recognition of the spatial background of the scene that characters appeared.