You are here

The Influence of Community Topics on Network Structures in V Kontakte

The final speaker in this Social Media and Society session is Yuri Rykov, whose focus is on the Russian social network site V Kontakte. What is the structure of online communities on this site? Are they flat, inclusive, and egalitarian, or are they stratified and hierarchical, with clear leadership structures emerging, as a power law distribution would expect? What new light can network analytics shed on these questions?

Past studies have examined online communities such as Yahoo! Answers and Slashdot, and (somewhat controversially) researchers at the Pew Center have even proposed the existence of six archetypical structures in online networks. Are such structures also related to the aims and functions of the specific social networking platform in question?

Yuri examined three different types of network communities on VK: online fan communities, acting as audiences to media texts; professional groups, engaging as a community of practice in knowledge sharing; and social movement communities, aiming to promote collective action on specific political issues. The project used the VKminer software, which connects with the VK API, and gathered data from each group's wall and discussion boards, including posts, user information, and friendship relations. This covered some 726,000 users with over 2 million friend relations.

Online fan communities had the highest number of connected components as well as clusters, and the lowest proportion of edges – they are the most fragmented of the three communities studied here. There is also a clear distinction between content contributors, semi-passive 'likers', and inactive followers.

Professional online communities have the highest proportion of active users; the highest clustering coefficient (with strongly tied but separated clusters, and only a few brokers between them); and the broadest distribution of posting and liking activity across the community.

Social movement communities have the lowest proportion of isolated users, the lowest modularity, the highest density, and the highest mean degree (excluding the isolates). At the same time, they are most centralised, with ties distributed least equally across the community.