Graph databases represent entities as nodes and relationships as explicitly stored edges.

In the graph model, entities are nodes, relationships are edges, and both may carry properties. This makes relationship semantics part of the stored data rather than an indirect consequence of foreign-key references. The distinction matters when queries ask for paths, neighborhoods, or recursively connected entities: traversal follows already represented edges, whereas a relational engine must resolve references through join operations. Graph structure therefore matches data whose meaning resides in connectivity, not merely in records.