“…Pattern, or subgraph, identification is one means to identify chemically relevant species within a large ensemble present in a graph representation. Locally, algorithms that include edge distribution, spectral graph theory, and clustering have been employed to study chemical heterogeneity, for example, in multicomponent solutions and crystalline systems. , In some cases, neural networks have been implemented to learn about local environments of nodes (specifically atoms in crystals) and predict potentially new organizational structures within materials design (Figure ). However, the limited information content in traditional chemical graphs highlights an arbitrary nature of many of the choices associated with such algorithms.…”