| name | build-home-page |
| description | Rebuild the home.microprediction.org GitHub Pages site (Peter Cotton's publications page) from papers.json. Use when asked to update the home page / publications list, add or refresh a paper, pull in newer drafts or arXiv links from the package repos (schur / humpday / precise / mechanics / skaters .microprediction.org), or regenerate docs/index.html. Run from the microprediction/home repo. |
Build home.microprediction.org
An academic publications page generated from data. Organized by research theme
(related work sits together); within each theme, published work / preprints come
first, then the most recent working drafts. Restrained academic styling — serif,
low bold, abstracts in collapsed <details>.
Files
| File | Role |
|---|
papers.json (repo root) | Source of truth. site, book_length[], themes[] (each with papers[]), software[], talks[], patents[], more[]. Edit this. |
abstracts.json (repo root) | Generated cache of real abstracts keyed by title. Committed so CI needn't extract. |
scan.py | Read-only. Scans ../{schur,humpday,precise,mechanics,skaters} for draft PDFs/TeX + arXiv ids. |
extract_abstracts.py | Local only (needs pdftotext + sibling repos). Pulls real abstracts → abstracts.json. |
build.py | Renders papers.json + abstracts.json → docs/index.html, docs/academic.css, docs/CNAME. Pure stdlib; runs in CI. |
Paper entry
{ "title": "...", "venue": "SIAM J. Financial Math., 2021",
"draft_new": true, "abstract_source": "https://github.com/.../x.tex",
"links": [ {"label":"arXiv","url":"..."}, {"label":"latest draft","url":"..."} ] }
venue present ⇒ published (rendered italic). List the canonical link (arXiv/journal) first.latest draft label is highlighted; use a PDF url, never a raw .tex.draft_in_progress: true shows a muted "draft in progress" note (no link).abstract_source = a local file to extract the abstract from when it isn't one of the entry's links (e.g. a TeX with no public PDF).
Procedure
- Find what's new / verify arXiv ids —
python3 scan.py. To confirm an
arXiv id belongs to Peter (vs. someone citing him), check the author:
curl -s "https://export.arxiv.org/api/query?search_query=au:%22Peter+Cotton%22&max_results=40".
- Edit
papers.json — add/move papers into the right theme, published-first;
add arXiv/journal/draft links. Don't invent abstracts or arXiv pairings.
- Abstracts —
python3 extract_abstracts.py (local). It writes only abstracts
it can extract verbatim; papers without a clean source get none (that's fine —
never fabricate). To add a missing one, paste the real text into abstracts.json.
- Build —
python3 build.py. Then open docs/index.html to review.
- Publish — commit
papers.json, abstracts.json and the skill. The Pages
workflow runs build.py and deploys docs/. DNS already points
home.microprediction.org → microprediction.github.io.
Notes
docs/index.html, docs/academic.css, docs/CNAME are generated and gitignored.- Restyle: edit
CSS in build.py.
