Source documents are not committed to this repository — neither the
original PDFs (data/*.pdf) nor the extracted text the pipeline archives
alongside them (data/*.txt). Both are gitignored. We don't have
redistribution rights to the source itself, only the right to read and
extract from a copy legitimately obtained.
This holds for automatically discovered open-access sources too, and the
distinction is worth stating plainly: extracting text changes the format,
not the copyright. "Open access" spans everything from CC-BY (redistributable
with attribution) through to free-to-read-only (not redistributable), and
this pipeline records no per-source license field — sources.json carries
title, authors, year, publication, DOI and URL, but nothing about
redistribution rights. Publishing the text would need that evidence
per source; until it's collected, the text stays local.
Keeping it locally is not incidental — data/<slug>.txt is what the
grounding check re-verifies quotes against, so it has to survive after the
PDF is deleted.
What is committed, and what actually matters for reproducibility, is
the citation: every source a subject draws from is fully described in
that subject's own sources.json — title, authors, year, publication,
DOI, and URL where available — plus page-level provenance on every
individual event and relation (sources[].page; see
schema/README.md). That's what lets a reader
verify a claim without the PDF needing to sit in this repo.
This file is a quick index of what's been handled, without opening every subject folder.
| Subject | Title | Authors | Year | DOI |
|---|---|---|---|---|
suntola |
A Short History of Atomic Layer Deposition: Tuomo Suntola's Atomic Layer Epitaxy | Riikka L. Puurunen | 2014 | 10.1002/cvde.201402012 |
aleskovskii |
From V. B. Aleskovskii's "Framework" Hypothesis to the Method of Molecular Layering/Atomic Layer Deposition | Anatolii A. Malygin, Victor E. Drozd, Anatolii A. Malkov, Vladimir M. Smirnov | 2015 | 10.1002/cvde.201502013 |
Add a row here (and the full entry in the subject's sources.json) any
time a new source is added, whether found by hand or by a future
discovery pipeline.
scripts/build_site.py --pdf expects it at data/<slug>.pdf (or
wherever --pdf points) and records that path in sources.json's
file field. That field documents a local convention, not a
guarantee the file is present in the repo — re-running extraction, or
checking a sources[].page citation against the original, both need
your own copy of the PDF; neither needs it to be in git.