CAJ | 학술논문

以中国最大的交易型网站淘宝网为平台，选取不同活跃度的交易型社区作为研究对象，从社会网络分析视角探讨不同社区活跃度下交易型社区成员之间关系网络的闭包机制。基于一般网络闭包相关理论，进一步考虑关系的方向以及交易型社区本身的信息性和功能性特征对其成员之间关系构建的影响。借助以Selenium为核心工具编写的爬虫程序，获取淘宝帮派网页上在同一主题下规模相似的4个具有不同社区活跃度的社区成员相互之间关系构建（相互关注和成为粉丝）的数据，运用图布局算法的Gephi软件，将所有样本社区的整体关系网络结构进行可视化呈现，通过构建关系形成概率的风险模型和基于SAS 9．2工具的参数检验得到研究结论。研究结果表明，活跃度较低的交易型社区内成员之间的关系网络闭包主要受选择性影响机制的驱动，而基于互惠性和传染性的社会性影响对成员的关系构建都具有负向的影响；活跃度较高的交易型社区内成员之间的关系网络闭包会同时受到社会性影响和选择性影响的驱动，并且社会性影响比选择性影响具有更强的作用；不论是在活跃度较低还是活跃度较高的交易社区，社会性影响和选择性影响的交互影响对社区成员的关系网络闭包具有显著的正向影响。这些研究结果可以帮助交易型社区的管理者和参与者从宏观角度观察社区的整体结构及其动态演化过程，同时也能从微观角度了解社区中成员之间相互关系构建的规律。
이중국최대적교역형망참도보망위평태，선취불동활약도적교역형사구작위연구대상，종사회망락분석시각탐토불동사구활약도하교역형사구성원지간관계망락적폐포궤제。기우일반망락폐포상관이론，진일보고필관계적방향이급교역형사구본신적신식성화공능성특정대기성원지간관계구건적영향。차조이Selenium위핵심공구편사적파충정서，획취도보방파망혈상재동일주제하규모상사적4개구유불동사구활약도적사구성원상호지간관계구건（상호관주화성위분사）적수거，운용도포국산법적Gephi연건，장소유양본사구적정체관계망락결구진행가시화정현，통과구건관계형성개솔적풍험모형화기우SAS 9．2공구적삼수검험득도연구결론。연구결과표명，활약도교저적교역형사구내성원지간적관계망락폐포주요수선택성영향궤제적구동，이기우호혜성화전염성적사회성영향대성원적관계구건도구유부향적영향；활약도교고적교역형사구내성원지간적관계망락폐포회동시수도사회성영향화선택성영향적구동，병차사회성영향비선택성영향구유경강적작용；불론시재활약도교저환시활약도교고적교역사구，사회성영향화선택성영향적교호영향대사구성원적관계망락폐포구유현저적정향영향。저사연구결과가이방조교역형사구적관리자화삼여자종굉관각도관찰사구적정체결구급기동태연화과정，동시야능종미관각도료해사구중성원지간상호관계구건적규률。
This research, based on the most active communities in the largest platform of e-business in China ( Taobao.com) , an-alyzed the differences of transactional community in the network closure mechanisms from different levels of community activities. Based on the network closure literatures, the authors take the directions of relationships and the informational and functional char-acteristics in transactional communities into considerations and explore their effects on the network closure among the members in transactional community.By using the Web-crawler tool and from the perspectives of social network analysis, the authors collect-ed the network data from four sampled transactional communities with the similar community size under the same community is-sue.The results show that the major differences among the sampled transactional communities rely on the levels of their communi-ty activities.The authors apply Selenium, open source software programmed, to collect the static online information from the sampled four transactional communities.After the authors visualized the general structures of all the sampled transactional com-munities, which is based on the software of Gephi, we build up the hazard model to account for the probabilities of established re-lationships from a dynamic view.Then, the authors apply the statistics tool of SAS 9.2 to test the hazard model built up previous-ly.Hence, the authors get the major results as follows: ①in the less active transactional communities, the network closure a-mong the members in the communities is mainly driven by the mechanism of selection effect, while both the reciprocity and conta-gion in social influence have negative effects on network closure among the members in transactional communities;②in the more active transactional communities, the network closure among the members in the communities is mainly driven by the mechanisms of both selection effects and social influences, while the social influence, such as reciprocity and contagion, has stronger effects once compared with the mechanism of selection effects;③no matter it is less active transactional community or more active trans-actional community, the interaction between the mechanisms of social influences and selection effects have positive effects on the network closure among the members in a transactional community.The key findings from this research can help not only the man-agers in the transactional communities but also the participants of those communities to better understand the evolution process of the transactional communities.From the macro-level perspective, they can monitor the general structures during the process of a community evolution, and from the micro-level perspective, they can get more insights on the patterns of relationship formation a-mong members in a transactional community.