Please click or tap on the image, or click here for a high resolution version for printing.
This graphic is my attempt to give a data-driven representation of the structure of recent philosophy. It is, at least for the moment, the final version of this project. The general method for its production is described in this previous post, where also all code is available. It was generated in the same fashion: using umap and hdbscan on WOS-data extracted with metaknowledge, plotted with ggplot and then reworked with Inkscape.
The Text in the graphic reads like this:
For this map I parsed 55327 papers in philosophy from the Web-Of-Science-Collection. The papers were determind by snowball-sampling: I started with a small sample (a few thousand papers), and extended from there by repeatedly looking at the most cited publications. For each paper I determined the specific works and authors it cited. Each of these features of the papers is a dimension in the dataset, which I then embedded into two dimensions, using Uniform Manifold Approximation and Projection, a dimension reduction algorithm by McInnes & Healy(2018). These two dimensions form the basis of the scatterplot you see above. The reduced data was then clustered with the hdbscan-algorithm into 42 clusters. The hdbscan finds clusters above a specified size, so we could have found far more smaller clusters, if we were looking for a more detailed view of the data, or less, but larger clusters. Data that was too sparse to be clustered is marked grey in the graphic. I then labeled the clusters by hand, identifying them by looking at the most frequent words and bigrams in the abstracts, the authors of the most cited works, and the most prolific authors in the field. To give an idea of the contents of each cluster, I added names of the most cited and most prolific authors, identifying the latter by mentioning them with their initials. While the most prolific authors can not always be understood to have shaped the field in a deep fashion, they anchor the clusters in more recent debates, and give me an opportunity to mention more women in the graphic. The clusters are a bit heterogenic in their nature: while some are thematic, others are determined strongly by specific persons or eras, which seems to be an interesting observation about the structure of the literature. But there is more that we can discover: the cleft between philosophy of science and epistemology, for example, or how the various historical clusters group themselves around moral philosophy. We can also observe that continental philosophy is a distinct cluster, that seems to split into two halves, but is still well formed, and that it is not that far away from the rest of philosophy, which might serve as a reality check for some debates. I hope you enjoy the visualization and find some more informative patterns in there. If you find a mistake or are interested in discussing the graphic further, please let me hear from you.