Math Research Compass

An in-progress dashboard for visualizing mathematical research trends across the arXiv. This project uses topic modeling, NLP, and arXiv metadata to map collaborations and thematic structure within the mathematical sciences.

The goal is to identify:

• Emerging topics across subfields (e.g. algebraic geometry, representation theory)
• Key research clusters and coauthor networks
• Thematic overlaps between adjacent arXiv categories (e.g. math.AG and math.RT)

The current pipeline uses:

• BERTopic with sentence-transformer embeddings
• Dimensionality reduction via UMAP
• Clustering with HDBSCAN
• Metadata aggregation from arXiv API
• Interactive visualizations powered by Plotly and Pandas

Planned features include:

• Implement collaboration network analysis and visualization
• Incorporate citation data (if accessible) for impact analysis

GitHub:
MathResearchCompass