소셜 빅데이터를 활용한 북한 관련 위협인식 요인 예측

Abstract

The purpose of this study is to develop a prediction model about threatfactors related to North Korea by applying network analysis and decisionmaking tree analysis to the social big data collected through online news sites,blogs, internet cafes, social network service and internet message boards. Through network analysis and opinion mining, classification of emotions towardNorth Korea into three categories was possible, referred to as ‘threat’, ‘praise’,and ‘agreement.’ A combination of low ‘elements of sympathy for NorthKorea’, high ‘elements of conservatism’ and high ‘elements of progressivism’had the most influence on threatening North Korea Policies. High ‘elements of NLL’ combined with high ‘elements of propagandaleaflets’ yielded significant impact on threatening national security related toNorth Korea. Additionally, high ‘elements of North Korean human rights’, low‘elements of South-North Korean Summit’ and low ‘elements of dispersedfamily’ was the most influential combination in threatening key issues on NorthKorea. Lastly, a combination of high ‘elements of China’, high ‘elements ofJapan’ and low ‘elements of the U.S.’ had the greatest influence on threateningmajor countries related to North Korea. In order to successfully promote thecurrent government’s national security strategy of Trustpolitik, and to grasp thetrend of threat factors related to North Korea, there must be cooperationbetween the government and private sectors in collecting information along withthe development of systematic method to manage information on North Koreathrough continuous use and analysis of social big data.