On Triangular versus Edge Representations -- Towards Scalable Modeling of Networks

In this paper, we argue for representing networks as a bag of triangular motifs, particularly for important network problems that current model-based approaches handle poorly due to computational bottlenecks incurred by using edge representations.