Detecting the time evolution of the community structure of networks iscrucial to identify major changes in the internal organization of many complexsystems, which may undergo important endogenous or exogenous events. Thisanalysis can be done in two ways: considering each snapshot as an independentcommunity detection problem or taking into account the whole evolution of thenetwork. In the first case, one can apply static methods on the temporalsnapshots, which correspond to configurations of the system in short timewindows, and match afterwards the communities across layers. Alternatively, onecan develop dedicated dynamic procedures, so that multiple snapshots aresimultaneously taken into account while detecting communities, which allows usto keep memory of the flow. To check how well a method of any kind couldcapture the evolution of communities, suitable benchmarks are needed. Here wepropose a model for generating simple dynamic benchmark graphs, based onstochastic block models. In them, the time evolution consists of a periodicoscillation of the system's structure between configurations with built-incommunity structure. We also propose the extension of quality comparisonindices to the dynamic scenario.
展开▼
