Abstract:The persistent development of online social networks and citizens' health awareness increasingly promote the emergence and boom of online health communities(OHCs).The online health communities provide a platform where users are able to obtain medical resources,share knowledge,experiences,and emotions with others.Recently,the OHCs is developing rapidly because of the advantages in providing abundant medical resources and health related knowledge as well as meeting users' social needs.Among various actions proceeding in communities,posting and replying embody the establishment of social relationships,and they are fundamental for keeping boom of the communities.Thus it is worthwhile investigating the post-reply behaviors and the underline mechanisms in the communities.We selected "Tianmijiayuan",an online diabetes health community,as our research testbed.We establish the reply networks based on the posting and replying data as well as users' personal information on this community from January,2015 to June,2015.The networks contain both the edges and nodes information.Then we utilize Exponential Random Graph Model(ERGM) to explore how network structure and node attributions affect the establishment of reply networks.The results indicate the reply networks exhibit reciprocity and transitivity characteristics simultaneously.The reciprocity indicates that users tend to reply between each other,and the transitivity shows the reply networks own great intention to develop.The influences of node attributions on reply network are much more complex.Users tend to reply those who share homophily with themselves,specifically such as the users in closed regions;users with high social capitals,such as more friends and superior activeness,would have higher possibilities to receive others' replies;those who reply others frequently and the new users are more likely to get others' replies.This study will enrich the research paradigm of user behavior mechanisms in the communities and enrich the literatures of electronic health,social networks,online health communities and reply behaviors.In addition to the above,it will also provide enlightenment for the reply management in online communities.The results can help improve the platform design mechanism,for instance,the communities should stimulate the new users to post courageously;encourage the users who have significant influence,such as those managers,active actors,to post and reply;as well supply efficient channels to facilitate information exchange among homogenous users.Those mechanisms would further promote users' information exchange in the communities and finally determine flourish the communities.Considering that the reply networks could be affected by other networks(such as friendship network),further research can explore the relationships among multiple networks and investigate those networks' formation mechanism.Meanwhile,replying behavior would also be affected by post's specific information,thus future research can also incorporate text mining techniques to explore the effects of post contents on replying behaviors.

Research on Mechanisms of User Replying Behaviorsin Online Health Communities