Analyzing Reference Networks Across Philosophical Texts

Over the past year, I worked with Aron Culotta, a well-known professor in AI and natural language processing, to conduct what is currently the largest computational analysis of philosophical texts within the English language. Our work is currently undergoing peer review from the Oxford Journal, "Digital Scholarship in the Humanities."

In our study, we downloaded & processed over 2,000 philosophical texts, and used Python to track how often each author references other authors within the dataset. With these results, we constructed graph representations—with authors as nodes and references as edges—to represent the reference networks mathematically. We also applied transformer models to classify each reference by its surrounding contexts, allowing us to identify specific threads in how philosophical influence has spread.

We applied multiple algorithms from graph theory to analyze these networks—clustering them, identifying paths between nodes, and determining the influence of individual authors. We combined our data analysis with a deep dive into philosophical literature, spanning across both primary and secondary sources to inform our analysis. Finally, we created a visualization tool with the D3 javascript library to allows anyone to interact with and visualize our networks for their own research.

Our paper first analyzes the overall structure of our network. For instance, we use total incoming references and in-degree centrality (incoming references from unique authors) to quantify the level of influence of various philosophers:

Figure 1: Distribution of incoming references across authors in our dataset.

Figure 2: In-degree centrality of top 15 authors in our dataset.

We also make similar calculations of subsets of our data which only include references classified into various topics. This allows us to make more precise estimates on the level of influence across different areas within philosophy:

Figure 3: Percentage of incoming references across topics in philosophy. Shows how the most influential philosophers have impacted different areas.

Finally, we explore the positions of various key figures with our visualization tool, providing additional context on how individuals relate to the wider network:

Figure 4: Reference networks of top authors with our visualization tool.

Figure 5: Kant's position in our reference network

If interested, feel free to check out the full paper here, our GitHub repository here, our visualization tool here, or some of my previous work in this area here. Special thanks to Aron Culotta for his guidance and the Tulane computer science department for their support!