Communities are one of the most important struc- tural elements of a network. They frequently influence network behavior, which makes their identification especially useful. As a result, community detection has been a popular topic within network science in recent decades. Even more recently, fostered by an increasing availability of time stamped datasets and a pressing realization that most empiric networks are dynamic in nature, temporal networks have attracted increased attention. The time dimension introduces new network constructs and communities are not immune. A community is no longer just a bunch of fixed nodes tightly clustered, but have a life and activity of itself, shedding and gaining nodes, appearing and disappearing on the network. We believe that these dynamic constructs are still lacking a formal, consensual definition. In this article we propose a robust taxonomy of life events for communities and a rules based methodology to clearly parse these events.

--

Complex systems,Networks,Clustering

• Physical Sciences - Natural Sciences

2nd best paper award