A Tag-Based Network Evolution Mechanism for Online Communities

An online community involves many interactions among members. When the interactions are frequent and important enough, they are usually explicitly shown as "friend links " among community members. In this paper, a tag-based network evolution model is proposed to simulate the growth and evolution of social networks of online communities. Our model is an agent-based model in which community members are simulated as agents who can decide whether to link to other agents as friends or not, according to their tag similarity. A tag is a value that represents an agent\´s total trait. Agents\´ traits could be their interests, locations or ages. We demonstrate the emergent properties of online communities generated by our tag-based evolution model, and compare these properties with those of other three network evolution models: BA model, small world network and random network model. Furthermore, two real-world online communities are studied empirically and compared with the online communities evolved from our model. Experimental results show that our evolution mechanism can facilitate the formation of a higher-quality community in which distances between members are smaller and members are more closely clustered.