Transform mathematical papers into interactive visualizations that illuminate the logical structure underpinning the research.
Each paper is rendered as a constellation:
- theorems, lemmas, definitions, and remarks are the stars
- logical dependencies are the edges between them
Status: research prototype / experimental
- Landing page:
http://localhost:3000/ - Example constellation:
http://localhost:3000/1705.03104/
- Interactive graph (D3): pan/zoom and drag nodes
- Dynamic node sizing based on connectivity
- Typed dependency edges (normalized for readability)
- Focus mode: select a node to isolate its neighborhood
- Info panel: read an artifact’s preview and prerequisites
- Legend filters: toggle node types to declutter
- Proof Path Explorer: expand/collapse prerequisite depth
- Review mode (prototype): guided review of theorems
- Node.js (LTS recommended)
npm installnpm run devThen open http://localhost:3000/.
index.html– landing page1705.03104/– an example constellation (static page + embedded data)assets/style.css– shared site stylingassets/modules/– graph rendering, interactions, UI, and proof/review logic
The interactive constellations are generated from an analysis pipeline (developed as part of ongoing research) that:
- identifies logical artifacts (theorems, lemmas, definitions, …)
- extracts/normalizes definitions and prerequisites for richer tooltips
- infers citations and dependency types from context, then normalizes edges for visualization
Because this is a research system, extractions and inferred relationships may be incomplete or imperfect.
The included example constellation page (1705.03104/index.html) defines window.graphData:
nodes:{ id, type, display_name, content_preview, prerequisites_preview, ... }edges:{ source, target, dependency_type, context, ... }
MIT © Turnstile Labs — see LICENSE.