Network analysis based on big data in social media of Korean adolescents’ diet behaviors

Adolescents are increasingly interested in weight control; hence, proper health education is important for helping them control their weight properly. This study was designed to pick out social media words that express adolescents’ diet behaviors, and identify the associations and types between such words and the behaviors. It used text-mining techniques and semantic network analysis for related big data collected from the Internet on adolescents’ diet behaviors. Text mining was used to extract meaningful information from unstructured text data, whereas semantic network analysis was used to understand the relationships between keywords. The top five keywords were “obesity,” “health,” “exercise,” “eat,” and “increase” in online news, and “exercise,” “eat,” “weight loss,” “obesity,” and “health” in blogs. The betweenness centrality of “appearance” was particularly higher than that of other centralities in online news. As a result of the CONCOR analysis, eight clusters each were identified in online news and blogs. This study’s results will serve as a basis for weight management-related intervention strategies, reflecting the perspectives of adolescents. It also has significance as basic data to provide correct information, and establish desirable weight control in the future.