Understanding characteristics of semantic associations in health consumer generated knowledge representation in social media

This study explores knowledge organization behavior on the Web with respect to identifying the semantic relationships of health‐related concepts. In particular, this study aims to investigate the potentials of imparting richer collective intelligence to existing knowledge representation systems in health. The study focuses on detecting semantic relationships between semantic groups of major concepts mined from health consumers' descriptions of health issues and associated user‐generated metadata (i.e., tags). A total of 50,263 blogs and associated 341,720 tags were collected from Tumblr, a blogging social networking site. Text mining and semantic network analysis methods were used to explore the usage patterns at semantic type levels of the identified medical concepts in tags, in blogs, and between tags and blogs. More various associations among semantic types were identified both in tags and in blogs. These associations were more diverse and complicated than the relationships in the Unified Medical Language System Semantic Network. Among the groups of concepts in tags and blogs, groups showed relatively stronger and more diverse relationships with other groups of concepts. In addition, many direct and close relations were found between tags and blogs.