Content recommendation by analyzing user behavior in online health communities

Online health communities (OHCs) are the platforms for patients and their care-givers to search and share health-related information, and have attracted a vast amount of users in recent years. However, health consumers are easily overwhelmed by the overloaded information in OHCs, which makes it inefficient for users to find contents of their interest. This study proposes a framework for content recommendation by analyzing user activities in OHCs that utilizes social network analysis and text mining technology. We model users' activities by constructing user behavior networks that capture implicit interactions of users, based on which closely related users are detected and user similarities are calculated. Text analysis are performed using topic model to select the threads for final content recommendation. Based on the data collected from a famous Chinese OHCs, we expect that our model could achieve promising results.