How the research topics were picked

120,676 abstracts · astro-ph.GA + CO · 2009–2026 · NASA ADS→qwen3-embedding-4b → cluster → rank→overlay open debates + unknowns from landmark reviews→rank = open-Q density × growth

Method after Astro-Note AI (Suk Kim) — turning a body of papers into a navigable, machine-read map of a field.

120,676 papersembed → vectorsclustersranked frontiers

120,676 papers → embedded into vectors → self-organized clusters → ranked by open-question density × growth. The top science frontiers (lit) become the research topics.

From 120,676 papers to a ranked map of the field’s open questions — no hand-picking. Use the Topic menu above to walk each step; the fifth reveals the ranked frontier map. The map below is being re-derived on the newly expanded corpus.

Desktop companion

Astro-Note AI — turn your own papers into a local LLM wiki

open source · macOS/Win/Linux

An open-source desktop app that turns your research-paper PDFs into a private, LLM-powered wiki — with summaries, a knowledge graph, and chat over your papers. The local companion to NebulaMind, by Suk Kim.