A Survey on Embedding Dynamic Graphs

Embedding static graphs in low-dimensional vector spaces plays a key role in network analytics and inference, supporting applications like node classification, link prediction, and graph visualization. However, many real-world networks present dynamic behavior, including topological evolution, feature evolution, and diffusion. Therefore, several methods for embedding dynamic graphs have been proposed to learn network representations over time, facing novel challenges, such as time-domain modeling, temporal features to be captured, and the temporal granularity to be embedded. In this survey, we overview dynamic graph embedding, discussing its fundamentals and the recent advances developed so far. We introduce the formal definition of dynamic graph embedding, focusing on the problem setting and introducing a novel taxonomy for dynamic graph embedding input and output. We further explore different dynamic behaviors that may be encompassed by embeddings, classifying by topological evolution, feature evolution, and processes on networks. Afterward, we describe existing techniques and propose a taxonomy for dynamic graph embedding techniques based on algorithmic approaches, from matrix and tensor factorization to deep learning, random walks, and temporal point processes. We also elucidate main applications, including dynamic link prediction, anomaly detection, and diffusion prediction, and we further state some promising research directions in the area.

The Impact of COVID-19 on the Dynamic Topology and Network Flow of World Stock Markets

Studies examining the impact of COVID-19 using network dynamics are scant and tend to evaluate a specific local stock market. We present a thorough investigation of 58 world stock market networks using a complex network approach spanning across the uncertain times that have resulted from the coronavirus outbreak. First, we use the daily closing prices of the world stock market indices to construct dynamic complex networks and sixteen minimum spanning tree (MST) maps for the period from December 2019 to March 2021. Second, we present the topological evolution properties of time-varying MSTs by applying normalized tree length, diameter, average path length, and centrality measures. Moreover, the empirical results suggest that (1) the highest correlation among the world stock markets is observed during the first wave of the COVID-19 pandemic in the months of February–March 2020; (2) most of the MSTs appear lower in hierarchy, and many chain-like structures are formed due to the sheer impact of pandemic-related crises; (3) Germany remained a hub node in many of the MSTs; and (4) the tree severely contracted during the first wave of the COVID-19 outbreak (during the months of February and March 2020) and expanded slightly afterwards. Moreover, the results obtained from this study can be used for the development of financial stability policies and stock market regulations worldwide.

Dislocation Topological Evolution and Energy Analysis in Misfit Hardening of Spherical Precipitate by the Parametric Dislocation Dynamics Simulation

Interaction of a single dislocation line and a misfit spherical precipitate has been simulated by the Parametric Dislocation Dynamics (PDD) method in this research. The internal stress inside the precipitate is deduced from Eshelby’s inclusion theory, the stress of the dislocation line and outside the precipitate is calculated by Green’s function. The influence of different relative heights of the primary slip plane on dislocation evolution is investigated, while the cross-slip mechanism and annihilation reaction are considered. The simulation results show three kinds of dislocation topological evolution: loop-forming (Orowan loop or prismatic loop), helix-forming, and gradual unpinning. The dislocation nodal force and the velocity vectors are visualized to study dislocation motion tendency. According to the stress–strain curve and the energy curves associated with the dislocation motion, the pinning stress level is strongly influenced by the topological change of dislocation as well as the relative heights of the primary slip plane.

Geometric Deep Lean Learning: Evaluation Using a Twitter Social Network

The goal of this work is to evaluate a deep learning algorithm that has been designed to predict the topological evolution of dynamic complex non-Euclidean graphs in discrete–time in which links are labeled with communicative messages. This type of graph can represent, for example, social networks or complex organisations such as the networks associated with Industry 4.0. In this paper, we first introduce the formal geometric deep lean learning algorithm in its essential form. We then propose a methodology to systematically mine the data generated in social media Twitter, which resembles these complex topologies. Finally, we present the evaluation of a geometric deep lean learning algorithm that allows for link prediction within such databases. The evaluation results show that this algorithm can provide high accuracy in the link prediction of a retweet social network.