포털 뉴스 사이트의 한국 총선 유형변화: 개체명 인식을 적용한 포털 뉴스 이슈의 종단적 변화를 중심으로

초록

The purpose of this study is to analyze how Naver and Daum portals constructed major issues in their political news sections and how the composition of the issues changed from the 17th to 21th general election periods. In order to analyze about 1,750,000 news articles, this study, using Korean Named Entity Recognition and Classification method, tried to reveal changes in the appearance of major people, organizations and local names, and to identify the similarities and differences between Naver and Daum. The results of the analysis are summarized as follows. In the case of the name of persons, the incidence of incumbent presidents has been decreasing, the incidence of non – politicians increasing. Regarding the name of organizations, the proportion of exposure of government agencies has continuously increasing, while the exposure of civilian organizations decreasing. With regard to the name of the regions, the proportion of the local names and the names of the places related to the region has been steadily decreasing and this result indicates that the importance of regional issues has decreased in the general election processes. Finally, the comparative analysis results of the two portals showed that the two portals were not able to secure differentiation in the agenda setting by exposing similar entity names.

키워드