network science

network science
Established	Late 20th century
Founders	Influenced by Paul Erdős, Alfréd Rényi, Stanley Milgram, Mark Granovetter
Parent fields	Graph theory, Statistical physics, Computer science, Sociology

network science is an interdisciplinary academic field which studies complex networks such as telecommunication networks, computer networks, biological networks, cognitive and semantic networks, and social networks. The field draws on theories and methods including graph theory from mathematics, statistical mechanics from physics, data mining from computer science, and social network analysis from sociology. Its cross-disciplinary nature aims to uncover common principles, algorithms, and tools that govern the structure and dynamics of networks across physical, biological, and social domains.

Overview

The field seeks to model real-world systems as networks of nodes connected by edges, moving beyond simple regular structures to analyze the intricate patterns of random graphs, small-world networks, and scale-free networks. Researchers analyze properties like degree distribution, clustering coefficient, and betweenness centrality to understand phenomena such as robustness, contagion processes, and the flow of information. Foundational work by figures like Albert-László Barabási and his research on preferential attachment has been instrumental in shaping modern understanding. Major institutions advancing the field include the Santa Fe Institute and the Network Science Center at the United States Military Academy.

History

Early mathematical foundations were laid in the 18th century with Leonhard Euler's solution to the Seven Bridges of Königsberg problem, a precursor to graph theory. The mid-20th century saw the development of random graph theory by Paul Erdős and Alfréd Rényi, providing probabilistic models for network connectivity. In the 1960s, Stanley Milgram's small-world experiment demonstrated the concept of "six degrees of separation" within social networks. The 1970s brought Mark Granovetter's seminal work on "The Strength of Weak Ties" in sociology. The field coalesced in the late 1990s and early 2000s with papers by Steven Strogatz, Duncan J. Watts, and Albert-László Barabási, who identified ubiquitous scale-free properties in real networks, leading to textbooks and dedicated journals like Network Science (journal).

Key concepts

Central to the discipline are several structural and dynamical metrics. The degree of a node counts its connections, while the degree distribution often follows a power law in scale-free networks. The clustering coefficient measures the density of triangles in a local neighborhood, a hallmark of small-world properties. Centrality measures, including betweenness centrality, closeness centrality, and eigenvector centrality, identify the most influential nodes. Dynamical processes studied include percolation theory for network resilience, epidemic models like SIR model for contagion spread, and synchronization phenomena in coupled oscillators, often analyzed using methods from statistical physics.

Types of networks

Networks are categorized by their domain and structural properties. Social networks map relationships between individuals or organizations, studied in platforms like Facebook and Twitter. Biological networks include metabolic networks, protein–protein interaction networks, and neural networks of the brain. Technological networks encompass the Internet, the World Wide Web, power grids, and transport networks. Information networks involve citations between academic papers in the arXiv repository or links between web pages. Other classifications include bipartite networks, multilayer networks, and temporal networks, which account for evolving connections over time.

Applications

Applications are vast and impact numerous sectors. In public health, network models track and mitigate the spread of diseases like COVID-19 through contact tracing. The National Security Agency and cybersecurity firms use it to analyze malware propagation and protect critical infrastructure. Google's PageRank algorithm, based on eigenvector centrality, revolutionized web search. In finance, it models systemic risk in interbank networks to prevent cascading failures like the 2007–2008 financial crisis. Biologists use it to identify key proteins in cancer pathways, while transportation planners optimize routes for entities like the London Underground and FedEx.

Tools and methods

Analysis relies on a combination of theoretical models, computational algorithms, and software tools. Common software packages include Gephi for visualization, NetworkX in Python (programming language), and igraph for large-scale analysis. Data is often gathered from APIs of social media platforms like LinkedIn or from repositories like the Stanford Large Network Dataset Collection. Algorithmic techniques involve community structure detection using the Girvan–Newman algorithm, link prediction, and network motif discovery. High-performance simulations may run on supercomputers at institutions like the Massachusetts Institute of Technology to model global-scale networks.

Category:Interdisciplinary fields Category:Network theory