A Primer on Temporal Graph Learning

Abstract

This document aims to familiarise readers with temporal graph learning (TGL) through a concept-first approach. We have systematically presented vital concepts essential for understanding the workings of a TGL framework. In addition to qualitative explanations, we have incorporated mathematical formulations where applicable, enhancing the clarity of the text. Since TGL involves temporal and spatial learning, we introduce relevant learning architectures ranging from recurrent and convolutional neural networks to graph neural networks. We develop a foundational understanding through intuitive examples and formal definitions, covering temporal graph representations, neural architectures adapted for dynamic graphs, and key learning tasks including prediction, generation, classification, and compression. Rather than surveying the full breadth of the literature, our goal is to build a conceptual framework that prepares readers for further study and application of TGL methods.