Network Theory

Also known as: Graph Theory

framework · mathematics · formal-scientific

Study of networks as nodes and edges, applied across many domains.

Network Theory (or Graph Theory in pure-mathematical contexts) is the study of the mathematical structures used to model pairwise relations between objects — networks composed of nodes (vertices) and edges (links) connecting them. The mathematical foundation was established by Leonhard Euler's 1736 solution to the Königsberg bridges problem, with substantial subsequent development through 19th and 20th century mathematics. The empirical-network science revolution of the late 1990s and 2000s — driven principally by Duncan Watts and Steven Strogatz's small-world networks (1998), Albert-László Barabási and Réka Albert's scale-free networks (1999), and the broader network-science movement — substantially extended graph theory into analysis of real-world networks. Network theory provides foundational vocabulary and mathematical tools: degree (number of connections per node), degree distribution, clustering coefficient, path length, betweenness centrality, eigenvector centrality, modularity (community detection), assortativity, percolation, network dynamics, network formation models (Erdős-Rényi random graphs, preferential attachment, configuration models). Applications span sociology (social networks — separately enriched), epidemiology (disease transmission), neuroscience (brain connectomics), genetics (protein-protein interactions), transportation (road, rail, airline networks), the World Wide Web, internet topology, financial networks, ecology (food webs), and many others. The discipline of 'network science' as integrative empirical-mathematical study has emerged largely since 2000.

Originators

Leonhard Euler (foundational, 1736 Königsberg bridges); 19th-20th century graph theory mathematicians; modern network science figures including Duncan Watts, Steven Strogatz, Albert-László Barabási, Réka Albert, Mark Newman high

Year / Decade

1736 (Euler foundational); ongoing graph-theory development; late 1990s onward (network-science revolution) high

Primary sources

Euler, L. (1736). 'Solutio problematis ad geometriam situs pertinentis', Newman, M.E.J. (2010, 2nd ed. 2018). Networks: An Introduction, Barabási, A.-L. (2016). Network Science, Watts, D.J. (2003). Six Degrees: The Science of a Connected Age high

Core components

Nodes (vertices) and edges (links)
Directed vs undirected networks
Weighted vs unweighted
Degree and degree distribution
Clustering coefficient
Path length and shortest paths
Betweenness, closeness, eigenvector, PageRank centralities
Modularity and community detection
Assortativity (degree correlation)
Network formation models (random graphs, preferential attachment)
Specific network classes (small-world, scale-free)
Network dynamics (epidemics, synchronization, opinion dynamics)
Connection to statistical physics, computer science, sociology

Primary use case

Mathematical foundation for analysis of networks across many domains; foundation for modern network science (Barabási, Newman, Watts); social network analysis (sociology, organizational research, marketing); epidemiology (disease transmission models); neuroscience (brain connectomics); biology (gene regulatory networks, protein-protein interaction networks); internet and web science; transportation and infrastructure analysis; financial network analysis; foundation for graph databases (Neo4j, Amazon Neptune) and graph machine learning (Graph Neural Networks).

Common criticisms

The 'network science' enthusiasm of 2000s sometimes outpaced the empirical and analytical rigor — many 'scale-free' networks turn out not to be cleanly scale-free on careful analysis (Aaron Clauset and colleagues' 2009 'Power-Law Distributions in Empirical Data' provided substantial methodological correction)
network analysis can produce metrics whose meaning isn't always clear (what does 'betweenness centrality' mean in this specific context?)
network models often abstract from temporal dynamics (most real networks are evolving, not static) — temporal network theory addresses this incompletely
cross-domain application of network frameworks sometimes elides domain-specific structure
commercial 'network analysis' platforms produce visualizations that can be misleading
integration with non-network features of phenomena (where the network is one aspect of a richer system) is often weak
some critics argue 'network science' has produced more hype than substance for some application areas, particularly when network metaphors substitute for substantive disciplinary engagement.

Lineage

Parent of: Scale-Free Networks, Small-World Networks
Siblings: Information Theory