Code
A proof of concept for science maps that visualize individual documents. It uses text embeddings of Leiden University publications in OpenAlex that include a title and abstract, then projects them into two-dimensional space with UMAP. The fast-loading visualization includes layouts generated with different UMAP hyperparameters, document search, and coloring by publication year.
An open-source web dashboard for exploring research institutions and funders from the CWTS OpenAlex dataset — no SQL knowledge required. Supports filtering, basket building, co-occurrence network visualization in VOSviewer, and CSV export.
An interactive science map that clusters networks using the Leiden algorithm. Allows selecting clusters to create new maps for iterative refinement. Comes with demo data, supports uploading your own data, and allows downloading data from selected clusters.
A free web app for creating science maps from paper titles and abstracts — no citation links required. Generates text embeddings with SPECTER2, builds a cosine similarity network or a UMAP dimensionality reduction, and exports directly to VOSviewer Online.
Papers