Evaluation of characteristics of a community for recommending dissimilar communities in SNS

Currently, SNSes (Social Networking Services) are widely available on the Internet. In a SNS, users can communicate with the users who have a similar interest in a community. To activate communications drastically in SNS, users are encouraged to join other dissimilar communities. In this paper, utilizing activities of communities and relationships with communities is proposed to realize a dissimilar community recommendation. However, such communities tend to be useless for the users, so that investigating characteristics of communities that are selected by the methods is necessary to recommend. By utilizing real data in the largest SNS in Japan, the correlations between the users' subjective estimations and the characteristics of communities are evaluated. As a result, it is clarified that effective recommendation of dissimilar communities will be possible by integrating activities of communities and relationships with communities.