Policy Analyses
Big Data for Social Risk Prediction
- Author
Tae Min Song
- Publication Date
2013
- Pages
- Series No.
- Language
The use of mobile Internet and Social Networking Service (SNS) has increased remarkably in Korea in line with the greater availability of smart phones. The Internet use rate of the population aged older than 3 years stood at 82.1% as of July 2013, of whom 55.1% of those 6 years old or older were found to have started using SNS within the preceding year (The Ministry of Science, ICT and Future Planning, and the Korean Internet Security Agency, 2013)1). As the volume of data transmitted through SNS has increased exponentially, a greater number of countries and corporations have endeavored to use and analyze big data in order to bring about new economic effects, create jobs, and resolve social issues. In the public sector, big data is utilized for disease prevention, prediction, and treatment and patient management through sharing genes and life research resources, and multinational IT (Information Technology) corporations and web search portal sites are producing various value information by analyzing big data saved in servers (Policy Exchange, 2012)2). SNS is where gloomy feelings, stress and worries that adolescents have in their everyday lives are heard and behaviors thereof are understood, so the analysis of emotional expressions about suicide and psychologically risky behaviors appearing on SNS can bring about the positive effect of preventing suicide by detecting risk signs and meaningful patterns. The suicide rate of Korea, amid remarkable socioeconomic changes, has been the highest level since 2004 among the member nations of the Organization for Economic Cooperation and Development (OECD) and in particular, as adolescent suicide is emerging as a social issue, government-level, pro-active measures are urgently needed. In addition, as adolescents exposed to cyber bullying commit suicide or become the perpetrators of violence, cyber bullying is emerging as a serious social issue. Cross-sectional research and longitudinal research, adopted thus far to investigate the causes and relevant factors of the above-noted social risk, are useful in inquiring into the relationships between an individual and a group with respect to pre-determined factors, but have limitations in determining how and to what extent an individual buzz mentioned on cyberspace is related to a social phenomenon. In this sense, the decision tree analysis of data mining that utilizes social big data can be deemed as a useful tool to analyze effectively the relationships of interaction of various causes arising from a complex and dynamic phenomenon of human behavior such as a social risk factor, by identiIntroduction fying a correlation or patterns according to decision rules without special statistical hypothesis. Accordingly, this research document is intended to propose a method to predict social risk factors of Korea through the decision tree analysis of data mining based upon social big data collected from online news sites, blogs, cafes, bulletin boards, etc.