文献类型：期刊文献

中文题名：在线健康社区中用户回帖行为影响机理研究

英文题名：Research on Mechanisms of User Replying Behaviors in Online Health Communities

作者：刘璇[1];汪林威[1];李嘉[1];张朋柱[2]

机构：[1]华东理工大学商学院,上海200237;[2]上海交通大学安泰经济与管理学院,上海200052

年份：2017

卷号：30

期号：1

起止页码：62

中文期刊名：管理科学

外文期刊名：Journal of Management Science

收录：CSTPCD;;国家哲学社会科学学术期刊数据库;北大核心:【北大核心2014】；社科基金资助期刊;CSSCI:【CSSCI2017_2018】；

基金：国家自然科学基金(71371005;71471064;91646205);上海市浦江人才计划项目(15PJC019);中央高校基本科研业务费(WN1522008)~~

语种：中文

中文关键词：在线健康社区;回帖行为机理;指数随机图模型;网络结构;节点属性

外文关键词：online health communities ; mechanisms of reply behaviors ; ERGM ; network structure ; node attributions

摘要：在线社交网络的持续发展和公民健康意识的不断增强促进了在线健康社区的兴起和繁荣。健康社区为用户提供了一个获取医疗资源并分享知识、经验和情感的开放式平台。由于其在为用户和患者提供丰富的医疗信息和知识资源、满足用户间的社交需求等方面表现突出,近年来,国内外在线健康社区都出现了蓬勃发展的态势。在社区的各种行为中,发帖和回帖是社区中用户社交行为的重要体现,是维系社区繁荣的根本,因此对社区中用户回帖行为的影响机理进行研究具有重要的意义。以糖尿病在线健康社区甜蜜家园网站为研究对象,基于该网站2015年1月~6月的社区发帖和回帖数据以及所有用户的个人信息数据,构建健康社区的用户回帖有向网络,该网络包含边和节点两部分属性信息;运用指数随机图模型,探讨该网络中网络结构和节点属性对回帖网络形成的影响机理。研究结果表明,回帖网络同时具有互惠性和传递性特征,互惠性表明社区中的用户倾向于互相回帖,传递性表明该网络有很好的发展潜力。节点属性对用户回帖行为的影响机制较为复杂,其中,用户倾向于对同质性(类型同质性)用户回帖;社会资本(如好友数量、活跃度)高的用户能获得更高的回帖概率;大量给他人回帖的用户和新用户更易获得他人的回帖。研究结果进一步丰富了在线健康社区中用户行为机理的研究范式,以及电子健康、在线社交网络、健康社区、论坛回帖行为等领域的研究,并对社区的管理有一定的启示作用;有利于指导社区的设计机制,如鼓励新人多发帖、引导有较大影响力的用户(如管理者和活跃度高的用户)发帖互动、进一步促进同质用户间的信息交流等,这些机制进一步促进和引导论坛中用户间的信息交互,对论坛持续繁荣发挥重要作用。
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 net-works 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 in- dicates 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 superi- or 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 e- lectronic 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 influ- ence, such as those managers, active actors, to post and reply; as well supply efficient channels to facilitate information ex- change 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.