Please note: This seminar will be given online.
Aida Sheshbolouki, David R. Cheriton School of Computer Science, University of Waterloo
Bipartite graphs have high representational capacity to capture pairwise and high-order interactions among entities in a wide variety of real-world contexts such as social networks, web-based services, and financial/transportation/communication/Information systems. Despite the richness, ubiquitous applications, and critical role of bipartite structures in driving the topological graph properties, there is a lack of studies on modeling their generative process, particularly in the streaming contexts where the underlying distributions of data experience non-stationarity. Current graph generative models focus on static or aggregated temporal graphs ignoring (1) the meso-scale patterns, (2) the edge attributes such as weights and timestamps, and (3) non-uniform inter-event statistics. We study the temporal organizing principles of mesoscopic building blocks of weighted bipartite streaming graphs and unveil the “Scale-invariant strength assortativity of streaming butterflies (2,2-bicliques)” and its origins. We use these realistic patterns to build a streaming growth model and validate it.
To join this seminar on Zoom, please go to https://us02web.zoom.us/j/83326411204?pwd=Z3dNVUxIK01PMXY3MTlXaHNVckJqdz09