A Survey on Embedding Dynamic Graphs

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