A recent trend in the world of Network Science is higher-order network analysis. In higher-order analyses we go beyond graphs by focusing on interactions involving more than two individuals through e.g. hypergraphs, simplicial complexes, motif-derived data, or tensors and higher-order Markov chains. But how do we choose which structures are relevant for which networks?
In my talk, I discuss an (I think) heretofore overlooked aspect of temporal network analysis, namely the constraints imposed by the network generating process. These constraints offer a unifying perspective to organize higher order network models. I discuss how the structure of communication events strongly constrain network configurations and shape network flows in a way that strongly impacts key areas of dynamic network analysis from community detection and null models, to our approach to modeling dynamics.