1. The underlying Scopus data for scholarly searches were obtained via the procedure described in Step 1 here.
2. The Scopus data were uploaded to "Get Networks" option of ScienceScape.
3. We used the default filtering setting of "Remove disconnected nodes". This procedure removes any data points for attributes such as author, keyword, or source title that appears in the data set but is not connected to any other of those attribute nodes.
4. We allowed the ScienceScape network layout to run and then downloaded the resulting data file for use with Gephi.
5. Using Gephi we ranked node size for given attributes of interest (e.g., keywords, authors, source titles) based on citation numbers so that more highly ranked attribute nodes would appear larger.
6. We ran the "Noverlap" layout to adjust node placement so that no nodes would visually obscure one another (i.e., every node remains visible).
7. We labeled each node using font size proportional to node size (i.e., larger font size corresponds proportionally to node size).
8. Once labeled, we ran the "Label Adjust" layout to further adjust node and label placement to avoid visual overlaps and to enhance readability of the graph.
9. Gephi's Sigma.js export feature (developed by Oxford's Internet Institute) was used to export the graph to a file format compatible for viewing in a web browser.
10. The config.json file was edited to achieve a balance between the legibility of the graph and visually highlighting prominent nodes.
Users who might wish to replicate our procedure can download the underlying Scopus data here and upload it to the ScienceScape option for "Get Networks" for subsequent use with Gephi. Alternatively, users can download the file ready for use in Gephi here. A tutorial for using Gephi can be found here.