communitydetektion

Communitydetektion, or community detection, is the task of identifying groups of vertices in a network that are more densely connected to each other than to the rest of the network. The resulting groups, called communities or modules, reveal the network's mesoscale structure and can indicate functional units or social divisions. The concept is central in network science and applies to social, biological, information, and communication networks.

Algorithms for communitydetektion fall into several families. Modularity optimization seeks partitions that maximize the difference between observed and expected within‑community edges; common implementations include the Louvain and Leiden methods. Stochastic block models view the network as a probabilistic division into blocks. Information-theoretic approaches such as Infomap use random walks to compress a network's flow. Other methods include spectral clustering, label propagation, and approaches for overlapping or hierarchical communities.

Evaluation uses metrics such as modularity, normalized mutual information, and adjusted Rand index, and benchmarks include synthetic graphs like the Girvan-Newman and LFR models. Challenges include the resolution limit, degeneracy of solutions, and sensitivity to network density. Dynamic networks require methods for tracking communities over time, and many algorithms allow nodes to belong to multiple communities.

Applications span sociology, biology, marketing, and information science. In biology, communities can correspond to functional modules in metabolic or protein networks; in social networks, they reflect groups or interests; in information networks, they reveal topic or user segments. Ongoing research seeks scalable, robust, and interpretable methods that handle overlapping, dynamic, and multi-layer networks.