A network is a visualization of connections between entities, such as people or places. In Networks, Crowds, and Markets: Reasoning About a Highly Connected World, Easley and Kleinberg define networks as “a pattern of interconnections among a set of things,” explaining that networks are adaptable to any set of things and links, such as a kinship network or the spread of an epidemic. In Networks: An Introduction, Newman interprets networks as a “collection of points joined together in pairs by lines.” Elijah Meeks defines a network as “a primitive object capable of and useful for the modeling and analysis of relationships between a wide variety of objects,” also noting the flexibility of this method of analysis. Scott Weingart defines networks as an “interlocking system” of “stuff and relationships.” He states that networks are integral to the research conducted, because the relationships between entities render them interdependent and necessary for comprehension of the topic. Network analysis is not merely a visual.
As Weingart notes, this flexibility is both an advantage and disadvantage. Networks can be used in a variety of ways, but scholars should think about the methodology carefully before undertaking such a task. One particular measure of networks to think about is centrality degree. This measure shows the entity, or node, most connected within the graph. This can be misleading, as “importance” can mean many things depending on what you’re visualizing. Another factor to consider is the type of data you want to represent. A network of a fraction of interactions experienced by jazz musician Roy Haynes is not useful. It becomes a Roy Haynes network graph, evolving around him. The visualization should represent the larger social network, such as the professional and social network of jazz musicians in Linked Jazz. However, the data should not be too much, or else the network graph will be unreadable. Along with that point, you should consider the reduction of the data necessary to render such a readable graph. You have to negotiate between the entire set of data and the amount required, often reduced, to produce a useful network graph.
Despite those points of wariness, network analysis is a very useful methodology for history. One of the most important features is the ability to see the whole network and the individual nodes and relationships. With networks, you can “distantly read” the information and the overall structure of the connections. When studying the people of a small town in eighteenth-century Virginia, you can use network analysis to analyze the structure of the community. With textual data, you can graph the network of newspaper and magazine reprinting in nineteenth-century to study print culture and public consumption, as in Viral Texts. You can also focus on individual nodes within the graph and how they relate to each other and how they are affected by certain factors.
Network graphs also apply to other methodologies in digital humanities. You can employ mapping to the network analysis, as in Stanford’s Orbis. You can also apply network analysis to research generated from topic modeling, creating powerful and useful visualizations. Scott Weingart’s insightful suggestion about contextualizing networks with textured base maps is even more involved. He proposes the use of space, demographics, networks, and other relevant data to enhance and contextualize a digital humanities project, akin to deep mapping. Network analysis, along with text mining and mapping, are very powerful tools for historical research.