To provide effective decision support for online community management, this study proposes new methods for formulating community dynamics and identifying influential members through the animal flocking model. We considered the online community as an animal flock, where members of the community are individual animals. The change in the direction of community discussions is analogous to the migration of an animal flock, although this occurs in virtual rather than physical space. To accurately formulate the community dynamics, we established a virtual space by using the ratios of articles in various categories as coordinates. In addition, the influence of members was defined according to their ability to change the direction of community discussions in the virtual space. On the basis of this concept, a novel algorithm was developed to identify the influential members. The evaluation results showed that the influential members identified using conventional measurements of influence often have no substantial impact on changes of the discussion direction. By contrast, the method proposed in this study clearly marked the influential members in the community. In addition, the results also demonstrated that the virtual space construction method can appropriately formulate community dynamics.
Electronic Commerce Research and Applications,49, 101085