This repository contains the code used for our analysis in The Geometric Structure of Topic Models. The topic model was computed by Topic space trajectories based on https://github.com/allenai/s2orc. Detailed the used data and pre-processing can be found in the article.
To run the experiments, you first have to retrieve the raw data (as
discussed in Topic space trajectories), apply the topic model and save
the results in data/author_paper_topics.csv for the author
embeddings and data/venue_paper_topics.csv for the venue
embeddings. After that, you can either use org-tangle to generate the
scripts from the org files or start a python/clojure REPL and copy
paste each code block.
The computed contexts that were used for the concept lattice diagrams
are included in the data directory.
The code in this repository falls under EPL-2.0 license.
Please cite the following article when using the presented methods:
T. Hanika and J. Hirth. The Geometric Structure of Topic Models. 2023