Network Visualization & Analysis of the Stanford Encyclopedia of Philosophy
The Stanford Encyclopedia of Philosophy (hereafter, SEP) is a "dynamic reference work" that has a significant influence in and on academic philosophy. The SEP is also an enormous resource of data about academic philosophy. This project is an attempt to represent the network structure of the SEP, visualize its global properties, and understand the structure of academic philosophy as represented in the SEP.
The SEP is archived four times per year, essentially freezing the encyclopedia at the time of the archive and creating a collection of snapshots from Fall 1997 until Fall 2018 (as of this writing). The notice on the SEP archive says that "[t]he content of these editions is neither updated nor modified in any way once the archive is made." In total, there are 84 archives of the SEP. A simple Perl script collects the links from each article and identifies the links that target other SEP articles. The script then prints two data files: one that lists every article in the archive and another that lists the links between them.
Once the data is loaded into R, we can use it to create the network objects:
data <- read.csv("data/2000_fall_edge_list.csv", header = TRUE) g <- graph_from_data_frame(as.data.frame(data), directed = TRUE) V(g)$label <- V(g)$name V(g)$degree <- degree(g) V(g)$betweenness <- betweenness(g,V(g),directed=TRUE) V(g)$group <- membership(cluster_louvain(g))
After creating the graph object from the list of edges (in this case with the
igraph package) and computing a few properties, we can use several packages to visualize the graph.
Example Graph: Fall 2000
This graph represents one of the early archives of the SEP. The nodes are sized by centrality in the graph and colored by community detection. You can hover, zoom, click and drag to manipulate the graph.
The community detection shows the early divisions aomong the SEP: broad areas of philosophy like value theory, metaphysics, epistemology, logic, and language are grouped together.
As later years of the SEP bear out, the articles that tend to score highest on centrality measures such as eigenvector centrality and betweenness are disproportionately articles about individual philosophers.
This year: Russell, Frege, Whitehead, Bradley, and Feyerabend.
Later (around 2017): Plato, Aristotle, Kant, Lewis, Locke, and Hume.
Analysis & Discussion
Here are some links to discussion and analysis regarding the SEP data.