Modelling Formation of Online Temporal Communities

Contemporary social media networks can be viewed as a break to the early two-step flow model in which influential individuals act as intermediaries between the media and the public for information diffusion. Today's social media platforms enable users to both generate and consume online contents. Users continuously engage and disengage in discussions with varying degrees of interaction leading to formation of distinct online communities. Such communities are often formed at high-level either based on metadata, such as hashtags on Twitter, or popular content triggered by few influential users. These online communities often do not reflect true connectivity and lack the cohesiveness of traditional communities. In this study, we investigate real-time formation of temporal communities on Twitter. We aim at defining both high and low levels connections and to reveal the magnitude of clustering cohesion on temporal basis. Inspired by a real-life event center sitting arrangement scenario, the proposed method aims to cluster users into distinct and cohesive online temporal communities. Membership to a community relies on intrinsic tweet properties to define similarity as the basis for interaction networks. The proposed method can be useful for local event monitoring and clique-based marketing among other applications.