Günce Keziban Orman

Şükrü Demir İnan Özer, Günce Keziban Orman, Vincent Labatut

Bipartite networks serve as highly suitable models to represent systems involving interactions between two distinct types of entities, such as online dating platforms, job search services, or ecommerce websites. These models can be leveraged to tackle a number of tasks, including link prediction among the most useful ones, especially to design recommendation systems. However, if this task has garnered much interest when conducted on unipartite (i.e. standard) networks, it is far from being the case for bipartite ones. In this study, we address this gap by performing an experimental comparison of 19 link prediction methods able to handle bipartite graphs. Some come directly from the literature, and some are adapted by us from techniques originally designed for unipartite networks. We also propose to repurpose recommendation systems based on graph convolutional networks (GCN) as a novel link prediction solution for bipartite networks. To conduct our experiments, we constitute a benchmark of 3 real-world bipartite network datasets with various topologies. Our results indicate that GCN-based personalized recommendation systems, which have received significant attention in recent years, can produce successful results for link prediction in bipartite networks. Furthermore, purely heuristic metrics that do not rely on any learning process, like the Structural Perturbation Method (SPM), can also achieve success.

Günce Keziban Orman, Vincent Labatut, Ahmet Teoman Naskali

Dynamic Networks are a popular way of modeling and studying the behavior of evolving systems. However, their analysis constitutes a relatively recent subfield of Network Science, and the number of available tools is consequently much smaller than for static networks. In this work, we propose a method specifically designed to take advantage of the longitudinal nature of dynamic networks. It characterizes each individual node by studying the evolution of its direct neighborhood, based on the assumption that the way this neighborhood changes reflects the role and position of the node in the whole network. For this purpose, we define the concept of \textit{neighborhood event}, which corresponds to the various transformations such groups of nodes can undergo, and describe an algorithm for detecting such events. We demonstrate the interest of our method on three real-world networks: DBLP, LastFM and Enron. We apply frequent pattern mining to extract meaningful information from temporal sequences of neighborhood events. This results in the identification of behavioral trends emerging in the whole network, as well as the individual characterization of specific nodes. We also perform a cluster analysis, which reveals that, in all three networks, one can distinguish two types of nodes exhibiting different behaviors: a very small group of active nodes, whose neighborhood undergo diverse and frequent events, and a very large group of stable nodes.