Dataset & exports
The research graph is the product; the presentation exists to make it legible. So the underlying data is public and downloadable — 211 records, 198 typed dependency edges, and 15 model families.
Downloads
- papers.json — Every record with all fields, as JSON.
- papers.csv — Every record, flattened for spreadsheets.
- relationships.json — The typed dependency edges, as JSON.
- relationships.csv — The typed dependency edges, flattened.
- bibliography.bib — BibTeX for every record (arXiv / DOI in the note field).
Schema
Each record carries identity (stable id, slug, title, authors, institutions, year, venue, arXiv ID / DOI, canonical URL, artifact type), an openness classification, a controlled status, a difficulty rating, six influence dimensions, an estimated citation figure with its source and retrieval date, and a verification tier. Each edge carries a source, target, typed relationship, explanation, and evidence tier. Field-level definitions and the controlled vocabularies are described on the methodology page.
Licence & citation
The editorial dataset is offered for research and educational use. Underlying papers remain the property of their authors and publishers; always cite the primary source (linked on every record), not this archive, for the research itself. When referring to this archive’s analysis or structure, cite The Intelligence Papers with the record id and retrieval date.