Background Research Reports

Visit the report calendar

PM Research Watch — an automated literature-surveillance pipeline for MINTS Research that tracks new publications on particulate matter (PM) monitoring (sensing, calibration, exposure modelling) and PM health effects, and publishes them as dated PDF digests on a GitHub Pages calendar site.

How it works

1. A daily job (claude/harvest.py) queries PubMed, Europe PMC, OpenAlex, and arXiv for new records since the last run.
2. Candidates are screened and deduplicated against claude/state/seen.json, then stored as structured records in claude/state/corpus/.
3. claude/plots.py and claude/mktable.py generate figures and tables, which are assembled into a LaTeX digest (claude/templates/daily.tex) and rendered to PDF.
4. The PDF is copied into Reports/daily/, and claude/build_manifest.py rebuilds Reports/reports.json — the manifest the website reads.
5. Periodic rollups (weekly, monthly, yearly) aggregate the daily records into Reports/weekly/, Reports/monthly/, and Reports/yearly/.

See claude/PIPELINE.md for the full pipeline contract, run order, and per-source notes.

Repo layout

Reports/reports.json                      # manifest the website reads
Reports/{daily,weekly,monthly,yearly}/    # generated PDF digests
claude/                                   # pipeline scripts + persistent state
index.dc.html, index.html, support.js     # GitHub Pages calendar site
assets/                                   # site images/logo

index.html redirects to index.dc.html, which renders the report calendar and an in-browser PDF reader driven by Reports/reports.json.