Ground-truth communities are groups of nodes in a network whose shared memberships or functional affiliations are known, observed, or explicitly established rather than inferred purely from the connectivity structure. In network science and graph analysis, these predetermined groupings serve as empirical baselines or gold standards used to benchmark, validate, and evaluate the accuracy of community detection algorithms. While algorithmic methods identify communities based on topological patterns, such as the density of links between nodes, ground-truth communities reflect authentic functional or social groupings, such as explicitly declared user interest groups, institutional departments, or biological modules. They provide a critical reference point for assessing how effectively mathematical and structural definitions of network clustering correspond to real-world relationships and functional organization.
