v0.2.6 errata: corrections to the notes, cards and scripts (no data change)

.agents/skills/source-rights-review/SKILL.md

@@ -98,7 +98,7 @@ Write one row per hop: author -> publisher/platform -> intermediary(ies) -> this
 
 If upstream licenses item by item, the corpus needs a per-row `license` (and author/title/URL where
 available), built from the upstream field. One label over mixed rights is a defect even when every
-individual licence is open (wolne_lektury: PD, FAL 1.3 and CC BY-SA 3.0 PL/4.0 under one label).
+individual licence is an open licence (wolne_lektury: PD, FAL 1.3 and CC BY-SA 3.0 PL/4.0 under one label).
 A licence version is part of the licence: 3.0 to 4.0 is not a free upgrade.
 
 ### 4. Public-domain claims
@@ -183,9 +183,9 @@ scrap(e|ing) | automated (access|means) | bulk download
 3. **`src/sources.py`**: `license`, `license_spdx`, `traceable` (why the licence holds), `provenance`
    (name the intermediary), `upstream`, `legal_note` (caveats, the withdrawal reason), `release`.
    Use `MIXED` plus a per-row field for mixed rights; `LicenseRef-...` for non-SPDX bases.
-4. **Decision record** `polish-dynaword-v2/repo_patch_suggestions/decisions/<key>.yaml`, schema
+4. **Decision record** in the paper repository, `repo_patch_suggestions/decisions/<key>.yaml`, schema
    `decision-record.schema.yaml` (`status: admitted | excluded | held_for_review`; evidence
-   in `polish-dynaword-v2/evidence/`). Example: `1000_novels.yaml`. There is no `partial` status: a partial
+   in the paper repository's `evidence/`). Example: `1000_novels.yaml`. There is no `partial` status: a partial
    admission is `status: admitted` with the kept-out parts listed under `scope.withheld`. No sibling checkout: put the record text in the PR description and flag it.
 5. **Findings log** `artifacts/source_findings.md`: append an entry whether the answer is yes or no.
 
@@ -210,7 +210,7 @@ As done for HPLT (`european_hplt_v3_pl`, removed from main on 2026-09-16):
    `release` value and set `removed` to the first release without it (`european_hplt_v3_pl`: `"0.2.3"` and
    `"0.2.6"`). `verify_paper_claims.py` stops counting the source from `removed` on.
 3. Write a decision record (`status: excluded`, or `held_for_review` while a human decides).
-4. Run `python3 src/verify_paper_claims.py --tex ../polish-dynaword-v2/main.tex --ref worktree`. Its FAILs
+4. Run `python3 src/verify_paper_claims.py --tex ../<paper repository>/main.tex --ref worktree`. Its FAILs
    are the paper's token, document and source counts that the removal breaks.
 5. `make_docs.py` skips sources with no data, so regenerated release totals undercount; see AGENTS.md
    "Known drift" before touching the README tables.

.agents/skills/source-rights-review/references/polish-law.md

@@ -16,7 +16,7 @@ primary text. Read the statute on ISAP/ELI before quoting an article in a datash
   "Prawa autorskie do tekstu wygasają: w 2045 roku w Polsce, w 2021 roku w USA" is a PD-US-only item
   (`wikisource_1295016`). A year after the current year in the "w Polsce" slot means keep out.
 - Europeana's conservative US slice is records created/published before 1929 (`legal_note` in
-  `src/sources.py`); later records stay separately labelled or held pending legal review.
+  `src/sources.py`); later records stay separately labelled or held pending a rights review.
 - Death years need a cited source. Hand-curated years (`scripts/novels_rights_screen.py`) are I.
 
 ## Art. 4 of the Copyright Act (official materials)

.agents/skills/source-rights-review/references/templates.md

@@ -103,7 +103,7 @@ language and per site. Prefer the allowlist argument: pass the exact licence id
 `status` values seen in the repo: `reviewed-open-official-materials`, `documented-statutory-basis-review-needed`.
 Never copy another source's `status`.
 
-## Decision record (`polish-dynaword-v2/repo_patch_suggestions/decisions/<key>.yaml`)
+## Decision record (in the paper repository: `repo_patch_suggestions/decisions/<key>.yaml`)
 
 Schema: `decision-record.schema.yaml` (same directory). Do not add fields it does not have.
 
@@ -134,7 +134,7 @@ verification:
   notes: "<what was NOT checked; hand-curated inputs; open human steps>"
 ```
 
-Per-text evidence (screens, sample classifications) goes in `polish-dynaword-v2/evidence/`
+Per-text evidence (screens, sample classifications) goes in the paper repository's `evidence/`
 (example: `1000_novels_rights_screen.csv`). If you have no checkout of that repo, paste the record into the
 PR description and say so.
 

AGENTS.md

@@ -234,9 +234,9 @@ Two rules about all of the above:
    per-source `stats.json` files. `git diff` the result.
 4. Append the release to `VERSIONS`, `DOCUMENTS` and `TOKENS` in
    `src/plot_version_growth.py`, using the same stats sums as step 3. Then check the
-   paper (polish-dynaword-v2) against the tree:
+   paper repository against the tree:
    ```bash
-   python3 src/verify_paper_claims.py --tex ../polish-dynaword-v2/main.tex --ref worktree
+   python3 src/verify_paper_claims.py --tex ../<paper repository>/main.tex --ref worktree
    ```
    It exits 0 once the paper describes this release. Without `--ref`, it reads
    every repository file at the revision the paper's `\snapshot` pins.

CHANGELOG.md

@@ -4,6 +4,22 @@
 
 Main has the same data as v0.2.6.
 
+Documentation errata (2026-10-30): one commit on main after the 2026-10-29
+squash corrects the release notes, cards, registry entries, scripts and
+reports, and the `v0.2.6` tag was re-pointed to it. No Parquet or JSONL file
+changed; the checksum files and manifests that pin an edited file were
+updated. References to an issue tracker that readers cannot open were removed
+or replaced by the release, date or Hub PR they stood for. The `wikipedia`
+notes give the 44 dropped pages by reason, and the Hub PR lists of
+`wikibooks`, `wikivoyage` and `1000_novels` are completed. The
+`sejm_api_2011_2022` and `sejm_committee_transcripts` cards state their
+`license` value in a **License:** line; the `open_icm_pl` card no longer lists
+an ID or IBAN rule that its scrub does not have; the `plwiki_talk` and
+`plwiki_user_talk` cards say that the contributed builds (Hub PRs 36 and 40)
+labelled every row `CC-BY-SA-4.0`; the `govpl` card notes the 5 rows that its
+count of rows from before 2019-08-05 leaves out. `src/verify_paper_claims.py`
+reads the paper's pinned revision and a card's **Rights status:** line.
+
 Repository history (2026-10-29): the Sejm's interpellations, questions and the
 government's replies came out of `parliamentary` until v0.2.7 (286,135
 documents / 391,602,712 tokens), for the same reason as `sejm_interpellations`:
@@ -81,8 +97,9 @@ tokens.
   v0.2.7 (withdrawn on 2026-10-29). 43,728 documents / 1,112,524,902 tokens
   (Hub PRs #108, #111).
 - `wikipedia` is rebuilt from the official plwiki dump, replacing the
-  SpeakLeash snapshot; each record is labelled by its contribution dates, and 42
-  pages with a copyright-violation notice or deleted for one are dropped.
+  SpeakLeash snapshot; each record is labelled by its contribution dates, and 44
+  pages are dropped: 35 with a copyright-violation notice, 7 deleted for one
+  and 2 that the copyright sample found copied.
   1,468,647 documents / 1,226,906,115 tokens (v0.2.5: 1,171,897 documents; Hub
   PRs #68, #128, #148).
 - `biblioteka_nauki` records carry the licence version and basis the Biblioteka
@@ -109,11 +126,11 @@ tokens.
   chapter titles. 100 documents / 21,504,056 tokens (Hub PR #187).
 - `wikibooks` withholds 54 pages; each record is labelled by its contribution
   dates and the licence of imported text, and markup residue is removed. 9,041
-  documents / 15,220,800 tokens (v0.2.5: 9,112; Hub PRs #127, #134, #139, #142,
+  documents / 15,220,800 tokens (v0.2.5: 9,112; Hub PRs #79, #127, #134, #139, #142,
   #147).
 - `wikivoyage` withholds 6,330 pages, 6,326 of them text from Wikitravel; each
   record is labelled by its contribution dates, and markup residue is removed.
-  7,312 documents / 14,084,102 tokens (v0.2.5: 13,645; Hub PRs #125, #133,
+  7,312 documents / 14,084,102 tokens (v0.2.5: 13,645; Hub PRs #78, #125, #133,
   #138, #141).
 - `wikinews` withholds 1 page; each record is labelled by its contribution
   dates and the licence of imported text, and markup residue is removed. 23,572
@@ -167,7 +184,7 @@ release tag that held them.
   each author and translator died in 1955 or earlier, but no row records the
   edition its text follows, so a protected preface, abridgement or editor's
   corrections cannot be ruled out. Main held 221 documents / 10,759,728
-  tokens, what was left after the rights screen (Hub PRs #49, #55) and the
+  tokens, what was left after the rights screen (Hub PRs #49, #55, #65) and the
   dedup against `wolne_lektury` and `eltec_pol`. The registry keeps its
   `release: "0.2.1"` entry with a `legal_note`, and `src/clean_1000_novels.py`
   stays for a re-admission.

NOTICE.md

@@ -67,10 +67,8 @@ it reaches the text sources here or binds a re-user; users who plan a commercial
 the page and decide for themselves. The page also says that clubs, circles, teams and MPs
 answer for their own pages; we did not take material from those pages.
 
-Archive: a verbatim copy of the terms block, with its date and SHA-256
-(`cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`), is kept as
-`evidence/sis_copyright_2026-10-19.md` (and its `.content.html` and `.meta.json`) in
-[polish-dynaword-v2](https://github.com/slayerlabs/polish-dynaword-v2/pull/121). The page's
+Archive: we keep a copy of the page as read on 2026-10-19 (a verbatim copy of the terms
+block, SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`). The page's
 URL is https://www.sejm.gov.pl/Sejm10.nsf/page.xsp/copyright. An earlier capture of the same
 page by the Internet Archive, 2026-06-03, carries the same source note and clause:
 http://web.archive.org/web/20260603115729/https://sejm.gov.pl/Sejm10.nsf/page.xsp/copyright
@@ -94,24 +92,21 @@ here and is not an acquisition date.
 - **Acquired:** 2026-10-07 (the TEI dump).
 - **Terms:** the only rights statement we found is IPI PAN's on its PPC page: "The
   parliamentary data is public domain." It is a redistributor's statement, not the
-  institutions'. We read the page on 2026-10-18 and archived it in polish-dynaword-v2 (pull
-  request 120). We record the sentence as provenance, not as our legal basis.
+  institutions'. We read the page on 2026-10-18 and keep a copy of it. We record the sentence as provenance, not as our legal basis.
 - **IPI PAN's confirmation, 2026-10-19:** by email, IPI PAN confirmed that its statement also
   covers its right as producer of the database, and agreed to the resource's inclusion in this
   collection under the collection's CC BY-SA 4.0 tag. That is IPI PAN's consent about its own
   rights and about the collection-level tag; it does not give the texts a licence and does not
-  speak for the Sejm or the Senate. The reply is archived as
-  `evidence/ipipan_reply_2026-10-19.md` in polish-dynaword-v2. IPI PAN also confirmed the
-  citation (Ogrodniczuk 2018) as correct for now and said a book on the corpus is in
-  preparation, so the preferred citation may change.
+  speak for the Sejm or the Senate. The reply is kept on file and is not published.
+  IPI PAN also confirmed the citation (Ogrodniczuk 2018) as correct for now and said a book
+  on the corpus is in preparation, so the preferred citation may change.
 - **Credit:** IPI PAN; the speaker turns, speaker ids, roles and dates this source is cut and
   filtered by come from IPI PAN's TEI markup. PPC's CC BY licence covers its annotations; none
   of the annotation layers is shipped. Cite Maciej Ogrodniczuk (2018), "Polish Parliamentary
   Corpus", Proceedings of the LREC 2018 Workshop ParlaCLARIN.
 - **SIS credit:** Źródło: System Informacyjny Sejmu (Kancelaria Sejmu RP), for the Sejm
   material in this source. We give it as a precaution: the source is a copy from IPI PAN's
-  server, not from the Sejm's system, and whether the note applies is open. It does not
-  cover the Senate material.
+  server, not from the Sejm's system. It does not cover the Senate material.
 
 ### sejm_api_2011_2022 and sejm_api
 
@@ -167,8 +162,7 @@ here and is not an acquisition date.
 - **Acquired:** not recorded. The shard was added on 2026-06-15.
 - **Credit:** Dziennik Ustaw; Kancelaria Sejmu RP for the ELI interface. As a precaution
   we also give the System Informacyjny Sejmu note, "Źródło: System Informacyjny Sejmu
-  (Kancelaria Sejmu RP)". The terms page does not mention the ELI interface, so whether the
-  note applies is open.
+  (Kancelaria Sejmu RP)". The terms page does not mention the ELI interface.
 
 ## Takedown and contact
 

README.md

@@ -270,7 +270,7 @@ earlier tags were deleted ([Releases](#releases)).
 |---|---:|---:|---|
 | `european_hplt_v3_pl` | 1,603,974 | 2,555.5M | Licence unverified: HPLT's CC0 covers its packaging, not the crawled pages. It will not be re-added. |
 | `wikisource` | 632,005 | 801.9M | Its single CC BY-SA 3.0 label also covered in-copyright post-war imprints that carry their own per-page licences. A rebuild from the official dump is planned for a later release. |
-| `1000_novels` | 1,000 | 60.5M | Its rights screen ([PR #49](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/49), [PR #55](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/55)) kept the public-domain texts, but no row records the edition its text follows, so a protected preface, abridgement or editor's corrections cannot be ruled out. Withdrawn on 2026-10-17 ([PR #124](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/124)). |
+| `1000_novels` | 1,000 | 60.5M | Its rights screen ([PR #49](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/49), [PR #55](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/55), [PR #65](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/65)) kept the public-domain texts, but no row records the edition its text follows, so a protected preface, abridgement or editor's corrections cannot be ruled out. Withdrawn on 2026-10-17 ([PR #124](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/124)). |
 | `samorzad_gov_pl` | 72,771 | 41.2M | Its CC BY-SA 4.0 is the platform operator's footer, not a grant from the 247 institutions that wrote the texts. Withdrawn on 2026-10-17 ([PR #123](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/123)). |
 | `wikiquote` | 30,363 | 31.9M | Most of its pages quote works still in copyright, and its CC BY-SA label covers only the contributors' selection. |
 
@@ -296,7 +296,7 @@ The 15 v0.2.5 sources in v0.2.6 total 2,297,055 documents / 6,399.8M tokens.
   snapshot. Each record is labelled CC BY-SA 3.0, 4.0 or both by its
   contribution dates. The 35 pages that carry plwiki's copyright-violation notice
   are dropped, and so are 7 pages that plwiki deleted for a copyright
-  violation. 1,171,897 → 1,468,647 documents, 707.2M → 1,226.9M tokens
+  violation and 2 that the copyright sample found copied. 1,171,897 → 1,468,647 documents, 707.2M → 1,226.9M tokens
   ([PR #68](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/68), [PR #128](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/128), [PR #148](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/148)).
 - `biblioteka_nauki` keeps its text from the SpeakLeash redistribution. Each
   record now carries the licence version and licence basis the Biblioteka Nauki
@@ -329,11 +329,11 @@ The 15 v0.2.5 sources in v0.2.6 total 2,297,055 documents / 6,399.8M tokens.
 - `wikibooks` withholds 54 pages: 51 by the rights review and 3 with
   `{{GFDL-tekst}}`. Each record is labelled by its contribution dates and the
   licence of imported text, and markup residue is removed. 9,112 → 9,041
-  documents, 15.6M → 15.2M tokens ([PR #127](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/127), [PR #134](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/134), [PR #139](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/139), [PR #142](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/142), [PR #147](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/147)).
+  documents, 15.6M → 15.2M tokens ([PR #79](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/79), [PR #127](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/127), [PR #134](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/134), [PR #139](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/139), [PR #142](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/142), [PR #147](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/147)).
 - `wikivoyage` withholds 6,330 pages: 6,326 with `{{Atrybucja}}` (text from
   Wikitravel), 2 with `{{Import}}` and 2 by the rights review. Each record is
   labelled by its contribution dates, and markup residue is removed.
-  13,645 → 7,312 documents, 17.1M → 14.1M tokens ([PR #125](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/125), [PR #133](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/133), [PR #138](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/138), [PR #141](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/141)).
+  13,645 → 7,312 documents, 17.1M → 14.1M tokens ([PR #78](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/78), [PR #125](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/125), [PR #133](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/133), [PR #138](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/138), [PR #141](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/141)).
 - `wikinews` withholds 1 page. Each record is labelled by its contribution
   dates and the licence of imported text, and markup residue is removed.
   24,386 → 23,572 documents, 12.1M → 11.0M tokens

artifacts/bartek_source_ingestion_plan_2026-07-02.md

@@ -58,7 +58,7 @@ Release split:
 - `created/published < 1929`: conservative public release slice for US-sensitive
   downstream users;
 - `created/published >= 1929` or unknown date: keep separately labeled, or hold
-  until legal review confirms the release scope.
+  until a rights review confirms the release scope.
 
 Dataset-card wording should say that Europeana rights statements are evaluated
 primarily for Poland/EU and that public-domain status can depend on jurisdiction.

artifacts/kultura_bez_barier_pl_ontology_manifest.json

@@ -118,7 +118,7 @@
         "artifacts/sample.jsonl": "e8b7d7d5962d69b6103ac6359798c2ceb6e72e63c3e933667a8e33d85e23146c",
         "artifacts/stats.json": "87ee0cfa7fbadfc734cf7ebc25be88eea8496cf9924d486db7bb0cc3f41289ad",
         "data/train-00000-of-00001.parquet": "76c36ab04179e6106e97343372f855956503968f6f0bfe1f7949e36995e3b8fb",
-        "src/build_kultura_bez_barier_pl.py": "ee0771bbf46243a8a65b790fe245b468028e61724ea1b7339b5dd2a2206ff4cc",
+        "src/build_kultura_bez_barier_pl.py": "087b314efd4b7b37a85ef60fc0ffc86b323b8eee45991c81bb5625bf6eb95e37",
         "src/requirements.txt": "807ccb72d622706ea3b41d39190c6258a17d4f5472eeab294a9d483eb4c44ac1",
         "src/test_kultura_bez_barier_contribution.py": "0962312fc13fdd8fc33a729b505896311455dcd623dcd543317378efcbfcb523"
       },
@@ -182,11 +182,11 @@
         "artifacts/sample.jsonl": "40f1daf74bf1c3c6ee927306ba19155b36560192fd6e56f190b039271765423e",
         "artifacts/stats.json": "53c0e1236633ba8296aa83887a950c2944ad284fbe74fd8a3ad41b213fa5ea2e",
         "data/train-00000-of-00001.parquet": "ae6126991199dc2a3e7b5627fb6d66b4c29a0179bc2d42bcda5a2dbcc380af4e",
-        "src/build_kultura_bez_barier_pl.py": "ee0771bbf46243a8a65b790fe245b468028e61724ea1b7339b5dd2a2206ff4cc",
-        "src/licence_basis_kultura_bez_barier_pl.py": "5c951168040efcd67e912c4f84fac6874fc047a69cd1a23509a8e7d04589ca7b",
+        "src/build_kultura_bez_barier_pl.py": "087b314efd4b7b37a85ef60fc0ffc86b323b8eee45991c81bb5625bf6eb95e37",
+        "src/licence_basis_kultura_bez_barier_pl.py": "1f16c1ddf569b14ab83cf00105279d7924ad64db35850e2aceb711d49811a960",
         "src/requirements.txt": "807ccb72d622706ea3b41d39190c6258a17d4f5472eeab294a9d483eb4c44ac1",
-        "src/test_kultura_bez_barier_contribution.py": "d04198abd2d2baeccd10a367851606ca148b02a1b1e72449da208a57ac2e6bc9",
-        "src/test_licence_basis_kultura_bez_barier_pl.py": "031225c8c25794db3bbaead2dc84875d4811ec0c9c7b430ef37e78ccd9980336"
+        "src/test_kultura_bez_barier_contribution.py": "58d3cce855ae6503e9117972ae3ed903994457ad22adea9a9efd3d2ba7779c2e",
+        "src/test_licence_basis_kultura_bez_barier_pl.py": "5093353043f75f5d436c81fbff0e390e955ce67050a08bac66e40900691318f8"
       },
       "produced_by": "run:e18ed13a28a81c2f8a8eb248b61794f243187bf3af3ec84067a66b4450a2f9fa"
     }
@@ -204,10 +204,10 @@
   "pending": [
     "cross-source deduplication",
     "benchmark contamination check",
-    "maintainer/legal review of catalogue-level license scope",
+    "maintainer review of catalogue-level license scope",
     "controlled training ablation",
     "replay of the v0.2.6 second PII pass from the raw PDFs (applied offline to the published text)",
-    "scope of the foundation's CC BY-SA 3.0 statement (publications or only the pictograms): to be asked of the foundation (v2#111)"
+    "scope of the foundation's CC BY-SA 3.0 statement (publications or only the pictograms): to be asked of the foundation"
   ],
   "protocols": [
     {

artifacts/open_icm_pl_ontology_manifest.json

@@ -17,7 +17,7 @@
     {
       "falsification_condition": "The pinned Parquet, decisions or overlap report do not reproduce these counts.",
       "id": "claim:open-icm-target-retention",
-      "statement": "The proposed DynaWord slice contains 11 records: 95 passed source QA, 10 exact title/DOI overlaps were excluded and 74 were dropped in the licence review (v2#111).",
+      "statement": "The proposed DynaWord slice contains 11 records: 95 passed source QA, 10 exact title/DOI overlaps were excluded and 74 were dropped in the licence review.",
       "supported_by": [
         "evidence:open-icm-target-stats:fc039f1bf9f46d3962e312f3969e8252b8b39ded2e99442fb0209612ff3918cd",
         "evidence:open-icm-target-overlap:c05fccedde6965de8dcaf62fd1233fe3fa77cdd105c426f9f6d5474684303a61"
@@ -67,14 +67,14 @@
       "observation_type": "checksums",
       "payload": {
         "data/open_icm_pl/open_icm_pl.attribution.jsonl": "4b490d4505f226317435dd11069c0b071a5e58d497e42e151818439a62c4dfa3",
-        "data/open_icm_pl/open_icm_pl.licence-basis.json": "1f3df27ad104feac062fc4264c5393dbcd5f6abc8502ddacd0690fff2fbca6cb",
+        "data/open_icm_pl/open_icm_pl.licence-basis.json": "4a3f97a164ddfc1eabc26f7312a268dee4e0a47db8f4eddc114bb0d40e70eeff",
         "data/open_icm_pl/open_icm_pl.parquet": "8a5e2cdbd3bb5575907d2d14e740316b1df010fd24d8c230bf779b5a6f9aa399",
-        "data/open_icm_pl/open_icm_pl.publisher-check.json": "df988ed8ee92db30c8d39babebf513faa795eed3b902a876e60d65bdf7394039",
+        "data/open_icm_pl/open_icm_pl.publisher-check.json": "a127bace44823cb49ff7bbef57ef2cf2127d9737dc64cff0373dd08cdcb74353",
         "data/open_icm_pl/open_icm_pl.review.csv": "f98a7dc73181b746e91a6033291a944599fc88cb69ce51a2c235c35cf647873a",
         "data/open_icm_pl/open_icm_pl.sample.jsonl": "8f79168ad93142aefeb159f0d527a89ad55e1c21971a0320222c51bdbf839ce0",
         "data/open_icm_pl/open_icm_pl.stats.json": "cbee968cc2368b2185dd56fa661fb2e10c29634bac51d46a0ed2d82e38321f14",
-        "src/licence_basis_open_icm_pl.py": "2f94bea5e89b14223f76fadcb23a37f1a9bb68462821f910a9e26253b666f242",
-        "src/test_licence_basis_open_icm_pl.py": "75a97c39ea3b85ea32363665660df24b7531e04076c5c5292e358921b5d03577"
+        "src/licence_basis_open_icm_pl.py": "ebd36cdd8eda0dbca39c258a4e947b75dae61a7ea0a19f7f13e0641816f0bb49",
+        "src/test_licence_basis_open_icm_pl.py": "a747836007585e665f9e669855ae7839f1861b599c953b72d10fb4dab8355bb7"
       },
       "produced_by": "run:open-icm-licence-basis:cac1f127d9ed0a35c0a39fcb3dfade7ee7a3387274eab224fb0add3894352777"
     }
@@ -94,7 +94,7 @@
     "benchmark contamination check",
     "review of quoted third-party text",
     "controlled training ablation",
-    "licence at the time of publication for the 9 publisher-basis records: the statements were read on 2026-10-20 and a publisher may have applied them to back issues later; 3 are journal-level; 1 names no version (v2#111)"
+    "licence at the time of publication for the 9 publisher-basis records: the statements were read on 2026-10-20 and a publisher may have applied them to back issues later; 3 are journal-level; 1 names no version"
   ],
   "protocols": [
     {

artifacts/rights_audit_2026-10/REVIEW.html

@@ -86,7 +86,7 @@ tr.grp td{background:var(--bg);font-weight:600;font-family:var(--f-display)}
 <section>
   <h2>Why this audit</h2>
   <div class="prose" style="display:flex;flex-direction:column;gap:12px">
-    <p>The paper (<code>polish-dynaword-v2/main.tex:46</code>) says:</p>
+    <p>The paper (<code>main.tex:46</code>) says:</p>
     <blockquote>No source carries a NonCommercial licence, but the rights to HPLT's underlying web pages and to some 1000 Novels texts are unresolved</blockquote>
     <p>That sentence is false for both <code>main</code> and v0.2.5. <code>openstax_pl</code> carries NonCommercial terms and an explicit ban on LLM training. Five sources already shipped in v0.2.5 have the HPLT problem: <code>wikisource</code>, <code>wikiquote</code>, <code>wiktionary_examples</code>, <code>1000_novels</code> and <code>nkjp1m</code>.</p>
     <p><b>Failure classes tested.</b></p>
@@ -183,7 +183,7 @@ tr.grp td{background:var(--bg);font-weight:600;font-family:var(--f-display)}
       <ul>
         <li><b>Registry:</b> <code>src/sources.py</code>. Edit the fields <code>license</code>, <code>license_spdx</code>, <code>traceable</code>, <code>provenance</code>, <code>upstream</code>, <code>legal_note</code> and <code>release</code>. <code>release: None</code> means the source is on main but unreleased.</li>
         <li><b>Per source:</b> <code>data/&lt;src&gt;/&lt;src&gt;.md</code> (card), <code>NOTICE.md</code>, <code>&lt;src&gt;.license-evidence.json</code>, <code>&lt;src&gt;.attribution.jsonl</code>, <code>&lt;src&gt;.stats.json</code>, <code>qa.json</code>.</li>
-        <li><b>Decision records:</b> <code>polish-dynaword-v2/repo_patch_suggestions/decisions/&lt;src&gt;.yaml</code> (schema <code>decision-record.schema.yaml</code>). Per-text evidence goes in <code>polish-dynaword-v2/evidence/</code>.</li>
+        <li><b>Decision records:</b> the paper repository's decision records (<code>repo_patch_suggestions/decisions/&lt;src&gt;.yaml</code>, schema <code>decision-record.schema.yaml</code>). Per-text evidence goes in the paper repository's <code>evidence/</code>.</li>
         <li><b>Paper check:</b> <code>src/verify_paper_claims.py</code> checks token and source counts. Run it after any removal.</li>
       </ul>
     </div>
@@ -255,7 +255,7 @@ tr.grp td{background:var(--bg);font-weight:600;font-family:var(--f-display)}
         <li>Drop PD-US-only pages. Their footer reads "Prawa autorskie do tekstu wygasają: w 20xx roku w Polsce" with a year after 2025.</li>
         <li>Add a per-row <code>license</code> and the source URL. Regenerate the card and remove the false sentence.</li>
       </ol></div>
-      <div><h4>Done when</h4><p>None of the IDs above survive. Every row has a page URL and a death-year basis or licence template. The rights-screen CSV is in <code>polish-dynaword-v2/evidence/</code>.</p></div>
+      <div><h4>Done when</h4><p>None of the IDs above survive. Every row has a page URL and a death-year basis or licence template. The rights-screen CSV (<code>evidence/1000_novels_rights_screen.csv</code>) is among the paper's ancillary files.</p></div>
       <div class="human"><h4>Needs a human</h4><p>Decide whether to withdraw from v0.2.6 until the rebuild lands. Death years for translators may need manual curation.</p></div>
     </div>
   </article>
@@ -304,7 +304,7 @@ tr.grp td{background:var(--bg);font-weight:600;font-family:var(--f-display)}
     <div class="grid">
       <div><h4>Problem</h4><p>CLARIN-PL's CC BY 4.0 covers only the compilation. The card on main now has a rights note, but the parquet is unchanged: all 1,000 rows say <code>CC-BY-4.0</code> and author is empty.</p></div>
       <div><h4>Evidence</h4><ul>
-        <li><span class="tag">V</span> Per-text screen <code>polish-dynaword-v2/evidence/1000_novels_rights_screen.csv</code>: 830 PD by term, 10 PD per Wolne Lektury, 17 in copyright, 143 unverified.</li>
+        <li><span class="tag">V</span> Per-text screen <code>evidence/1000_novels_rights_screen.csv</code>, among the paper's ancillary files: 830 PD by term, 10 PD per Wolne Lektury, 17 in copyright, 143 unverified.</li>
         <li><span class="tag">V</span> The 17 are 2 by Dąbrowska, 12 translated by Michał Friedman and 3 translated by Hōjun. Wolne Lektury publishes them under CC BY-SA 3.0 PL; 5 still carry that footer.</li>
         <li><span class="tag">V</span> Row <em>i</em> of the parquet is the <em>i</em>-th file in the CLARIN listing, which is how rows map to the screen (decision record <code>1000_novels.yaml</code>).</li>
         <li><span class="tag">I</span> The CLARIN-PL licence page wasn't read (timeout). Death years in <code>scripts/novels_rights_screen.py</code> are hand-curated.</li>
@@ -720,7 +720,7 @@ tr.grp td{background:var(--bg);font-weight:600;font-family:var(--f-display)}
         <li>4.4% cite an expert opinion.</li>
         <li>1.2% name natural persons.</li>
       </ul></div>
-      <div><h4>Fix</h4><p>Reword the label to state the review status ("statutory basis, no formal legal review"). If SAOS merges, choose one canonical source for the 4,541 KIO duplicates. Optional: written confirmation from UZP.</p></div>
+      <div><h4>Fix</h4><p>Reword the label to state the review status ("statutory basis, our assessment"). If SAOS merges, choose one canonical source for the 4,541 KIO duplicates. Optional: written confirmation from UZP.</p></div>
     </div>
   </article>
 

artifacts/rights_audit_2026-10/reports/citations/nkjp_novels.md

@@ -59,15 +59,15 @@ CLARIN-PL record (used by several rows): "1000 Novels Corpus", Eder, Rybicki, M
 | M2 | the CLARIN-PL licence page wasn't read (timeout) | gap now closed through the API; the HTML page is still unread | CLARIN-PL deposit licence bitstream `license.txt` (bundle LICENSE, 1,663 bytes, MD5 7670c817844272f0db340269678d2a0e), "NON-EXCLUSIVE DISTRIBUTION LICENSE" | https://clarin-pl.eu/server/api/core/bitstreams/47b881c4-d827-4959-91b5-c453fa33f3cf/content | NONE | "If the submission contains material for which you do not hold copyright, you represent that you have obtained the unrestricted permission of the copyright owner to grant PWR the rights required by this license, and that such third-party owned material is clearly identified and acknowledged" (quote shortened) — the depositor warrants third-party permission and identification. The 1000 file names identify authors and titles; no per-text rights statement appears in the record or the licence. | V |
 | M3 | all 1,000 parquet rows say CC-BY-4.0 and author is empty | supported | Local parquet and card | `data/1000_novels/1000_novels.parquet` (1000 rows; `license` all `CC-BY-4.0`; `author` all empty) | n/a (local) | n/a | V |
 | M4 | the card on main now has a rights note | supported | Local card `data/1000_novels/1000_novels.md`, section "Rights note (per-text screen, September 2026)", lines 16-30 | `data/1000_novels/1000_novels.md` | n/a (local) | "The CC-BY-4.0 licence covers the corpus as compiled by CLARIN-PL. It cannot license third-party rights in the individual novels and translations." (card wording; the sentence is the card's own claim, see M1) | V |
-| M5 | per-text screen: 830 PD by term, 10 PD per Wolne Lektury, 17 in copyright, 143 unverified | supported (counts re-computed from the CSV) | Local evidence file `evidence/1000_novels_rights_screen.csv` in the v2 working tree; column `screen` gives 830 `pd_by_term`, 10 `pd_per_wolne_lektury`, 17 `in_copyright`, 143 `unresolved` (sum 1000). Ids are `1000_novels_NNNN`, 1-based, so id 0042 is 0-based parquet row 41 | `polish-dynaword-v2/evidence/1000_novels_rights_screen.csv` | n/a (local) | n/a | V |
+| M5 | per-text screen: 830 PD by term, 10 PD per Wolne Lektury, 17 in copyright, 143 unverified | supported (counts re-computed from the CSV) | Evidence file `evidence/1000_novels_rights_screen.csv`, among the paper's ancillary files; column `screen` gives 830 `pd_by_term`, 10 `pd_per_wolne_lektury`, 17 `in_copyright`, 143 `unresolved` (sum 1000). Ids are `1000_novels_NNNN`, 1-based, so id 0042 is 0-based parquet row 41 | `evidence/1000_novels_rights_screen.csv` among the paper's ancillary files | n/a (local) | n/a | V |
 | M6 | the 17 are 2 by Dąbrowska, 12 translated by Michał Friedman, 3 translated by Hōjun | supported | CSV ids (column `translator` / `clarin_file`): Dąbrowska 0166 (dabrowska_dzikie-ziele_1956.txt), 0167 (dabrowska_boze-narodzenie.txt); Michał Friedman 0042, 0043, 0044 (anonim5_ze-skarbnicy-midraszy, anonim4_sentencje-ojcow, anonim3_agady-talmudyczne), 0308, 0309, 0310, 0314 (four Kacyzne books), 0451 (Mojcher-Sforim), 0604 (Perec), 0608 (Perle), 0846, 0848 (Alejchem) = 12; Hōjun 0315 (kigen_umysl-drogi), 0316 (kigen_o-przyjmowaniu-wskazan), 0317 (kigen_narodziny-smierc_2013) = 3. WL XML `dc:contributor.translator` confirms the translator for each of the 15 translated slugs; the id-to-slug match was done by file name | CSV path as in M5; WL XML pattern https://wolnelektury.pl/media/book/xml/<slug>.xml for the 17 slugs (e.g. https://wolnelektury.pl/media/book/xml/agady-talmudyczne.xml) | n/a (local); XML archive per row in M7 | WL XML `dc:contributor.translator`: "Friedman, Michał" (12 books) and "Hōjun" (3 books) — the translator tag in the WL metadata. | V |
 | M7 | WL publishes them under CC BY-SA 3.0 PL (Dąbrowska) | supported for Dąbrowska | WL book metadata XML, `dc:rights` and `dc:rights.license`; WL text footer | https://wolnelektury.pl/media/book/xml/dabrowska-boze-narodzenie.xml ; https://wolnelektury.pl/media/book/xml/dabrowska-dzikie-ziele.xml ; footer: https://wolnelektury.pl/media/book/txt/dabrowska-boze-narodzenie.txt ; author page: https://wolnelektury.pl/katalog/autor/maria-dabrowska/ | XML: http://web.archive.org/web/20250416201436/https://wolnelektury.pl/media/book/xml/dabrowska-boze-narodzenie.xml ; http://web.archive.org/web/20221205012452/https://wolnelektury.pl/media/book/xml/dabrowska-dzikie-ziele.xml ; txt: http://web.archive.org/web/20241202160421/https://wolnelektury.pl/media/book/txt/dabrowska-boze-narodzenie.txt ; author page: http://web.archive.org/web/20260105184957/https://wolnelektury.pl/katalog/autor/maria-dabrowska/ | `dc:rights`: "Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0.PL", licence `http://creativecommons.org/licenses/by-sa/3.0/`. Footer: "Ten utwór jest udostępniony na licencji Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0.PL" — This work is available under CC Attribution-ShareAlike 3.0 PL. Source edition `dc:source`: "Maria Dąbrowska, Pisma wybrane … wyd. Czytelnik, Warszawa 1956". | V |
 | M8 | (same) WL publishes the Friedman translations under CC BY-SA 3.0 PL | supported for all 12 Friedman books | WL XML `dc:rights`, one per slug, all "…Na Tych Samych Warunkach 3.0.PL", licence `http://creativecommons.org/licenses/by-sa/3.0/`. Slugs: agady-talmudyczne, alejchem-kasrylewka, alejchem-z-jarmarku, kacyzne-chore-perly-i-inne-opowiadania, kacyzne-dziwny-zyd, kacyzne-remedium-dla-literatow, kacyzne-stare-miasto, perec-opowiadania-chasydzkie-i-ludowe, perle-zydzi-dnia-powszedniego, podroze-beniamina-trzeciego, sentencje-ojcow, ze-skarbnicy-midraszy. Translator page: https://wolnelektury.pl/katalog/autor/michal-friedman/ ("Twórczość Michała Friedmana – Tłumaczenia") | https://wolnelektury.pl/media/book/xml/alejchem-z-jarmarku.xml (and the same path for each slug); footers: https://wolnelektury.pl/media/book/txt/kacyzne-dziwny-zyd.txt , https://wolnelektury.pl/media/book/txt/alejchem-z-jarmarku.txt ; translator page above | XML archives: alejchem-z-jarmarku http://web.archive.org/web/20190724000346/https://wolnelektury.pl/media/book/xml/alejchem-z-jarmarku.xml ; alejchem-kasrylewka http://web.archive.org/web/20260213094002/https://wolnelektury.pl/media/book/xml/alejchem-kasrylewka.xml ; agady-talmudyczne http://web.archive.org/web/20240927092723/https://wolnelektury.pl/media/book/xml/agady-talmudyczne.xml ; kacyzne-stare-miasto http://web.archive.org/web/20220419173712/http://wolnelektury.pl/media/book/xml/kacyzne-stare-miasto.xml ; kacyzne-chore-perly http://web.archive.org/web/20221202002702/https://wolnelektury.pl/media/book/xml/kacyzne-chore-perly-i-inne-opowiadania.xml ; podroze-beniamina-trzeciego http://web.archive.org/web/20220419205049/https://wolnelektury.pl/media/book/xml/podroze-beniamina-trzeciego.xml ; perle http://web.archive.org/web/20220419214430/http://wolnelektury.pl/media/book/xml/perle-zydzi-dnia-powszedniego.xml ; sentencje-ojcow http://web.archive.org/web/20171026060758/http://wolnelektury.pl/media/book/xml/sentencje-ojcow.xml ; ze-skarbnicy-midraszy http://web.archive.org/web/20220419164817/http://wolnelektury.pl/media/book/xml/ze-skarbnicy-midraszy.xml ; kacyzne-remedium and perec: NONE ; kacyzne-dziwny-zyd txt http://web.archive.org/web/20220426102942/https://wolnelektury.pl/media/book/txt/kacyzne-dziwny-zyd.txt ; alejchem-z-jarmarku txt http://web.archive.org/web/20231216130259/https://wolnelektury.pl/media/book/txt/alejchem-z-jarmarku.txt ; lektura page alejchem-kasrylewka http://web.archive.org/web/20260113022216/https://wolnelektury.pl/katalog/lektura/alejchem-kasrylewka/ | `dc:contributor.translator`: "Friedman, Michał"; `dc:rights`: "Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0.PL"; `dc:source` for Z jarmarku: "Szołem Alejchem, Z jarmarku, tłum. Michał Friedman, Wydawnictwo Dolnośląskie, Wrocław 1989." — the Friedman translation was first published commercially (1989); how the translator or publisher granted WL a CC licence is not stated in the metadata. | V |
 | M9 | (same) WL publishes the Hōjun translations under CC BY-SA 3.0 PL | contradicted in detail. The three Hōjun books carry "3.0 Unported", not "3.0 PL". See Corrections D2 | WL XML `dc:rights` for dogen-kigen-umysl-drogi, dogen-kigen-o-przyjmowaniu-wskazan, dogen-kigen-zycie-i-smierc; `dc:date` 2013-05-21 / 2013-08-30; translator page https://wolnelektury.pl/katalog/autor/hojun/ | https://wolnelektury.pl/media/book/xml/dogen-kigen-umysl-drogi.xml ; https://wolnelektury.pl/media/book/xml/dogen-kigen-o-przyjmowaniu-wskazan.xml ; https://wolnelektury.pl/media/book/xml/dogen-kigen-zycie-i-smierc.xml ; footer https://wolnelektury.pl/media/book/txt/dogen-kigen-zycie-i-smierc.txt ; translator page above | umysl-drogi XML http://web.archive.org/web/20201127054312/https://wolnelektury.pl/media/book/xml/dogen-kigen-umysl-drogi.xml ; lektura page http://web.archive.org/web/20260612234912/https://wolnelektury.pl/katalog/lektura/dogen-kigen-umysl-drogi/ ; zycie-i-smierc lektura http://web.archive.org/web/20251119135611/https://wolnelektury.pl/katalog/lektura/dogen-kigen-zycie-i-smierc/ ; zycie-i-smierc XML and txt: NONE ; o-przyjmowaniu-wskazan: NONE | `dc:rights`: "Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0 Unported", licence `http://creativecommons.org/licenses/by-sa/3.0/deed.pl`. Footer: "Ten utwór jest udostępniony na licencji Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0 Unported" — CC BY-SA 3.0 Unported. `dc:source` (Umysł Drogi): "Dōgen Kigen, Umysł Drogi [online], tłum. Hōjun, [dostęp: 09.09.2013]" with the mahajana.net URL. | V |
 | M10 | (new finding, bears on relabelling) licence on the original translator site for the same three texts | conflicts with WL | mahajana.net library pages for the Hōjun translations, project "Tłumaczenia Tekstów Dharmy – 'Kumarajiva'", "Informacje" block | http://mahajana.net/node/414 (Shoji, Życie i śmierć); http://mahajana.net/ttd/Dogen%20Zenji%20-%20Umysl%20Drogi/Dogen%20Zenji%20-%20Umys%C5%82%20Drogi.html ; http://mahajana.net/ttd/Dogen%20Kigen%20-%20Jukai/Dogen%20Kigen%20-%20Jukai.html ; project page http://mahajana.net/informacje/tlumaczenia-tekstow-dharmy-kumarajiva | node/414 http://web.archive.org/web/20170326073032/http://mahajana.net/node/414 ; Umysł Drogi http://web.archive.org/web/20160325104023/http://mahajana.net:80/ttd/Dogen%20Zenji%20-%20Umysl%20Drogi/Dogen%20Zenji%20-%20Umys%C5%82%20Drogi.html ; Jukai http://web.archive.org/web/20160325105832/http://mahajana.net:80/ttd/Dogen%20Kigen%20-%20Jukai/Dogen%20Kigen%20-%20Jukai.html ; project page http://web.archive.org/web/20260618134257/https://mahajana.net/informacje/tlumaczenia-tekstow-dharmy-kumarajiva | "Ten utwór jest dostępny na licencji Creative Commons Uznanie autorstwa-Bez utworów zależnych 3.0 Unported." — available under CC Attribution-NoDerivs 3.0 Unported (BY-ND). Project page: "Teksty będą udostępniane w oparciu o wolne licencje (np. Creative Commons Attribution ShareAlike 3.0)" — intended licences are free ones, e.g. BY-SA 3.0. So the pages read say BY-ND while WL says BY-SA; which one is right is not established. | V |
 | M11 | 5 of the 17 still carry the WL footer | supported. The licence line is present in the corpus text for ids 0042, 0043, 0044, 0315, 0317 (0-based parquet rows 41, 42, 43, 314, 316). The three Friedman footers read "3.0.PL", the two Dōgen footers read "3.0 Unported". The other 12 have no licence line (rows 165 and 166, Dąbrowska, keep only the "Tekst opracowany na podstawie" source line or nothing) | Local check of the parquet text for the footer block; it also matches the WL txt footers | `data/1000_novels/1000_novels.parquet` (rows above); comparison: https://wolnelektury.pl/media/book/txt/kacyzne-dziwny-zyd.txt (line 212), https://wolnelektury.pl/media/book/txt/dogen-kigen-zycie-i-smierc.txt (line 27) | see M8, M9 | In the corpus text (row 42, Sentencje ojców, sic "udostepniony" without the ogonek): "Ten utwór jest udostepniony na licencji Creative Commons Uznanie Autorstwa - Na Tych Samych Warunkach 3.0.PL: http://creativecommons.org/licenses/by-sa/3.0/." — footer text in the corpus itself. Row 41 (Ze skarbnicy Midraszy) adds: "Reprodukcja cyfrowa wykonana przez Fundację Nowoczesna Polska z książki udostępnionej przez Fundację im. Michała Friedmana" — digital reproduction made from a book supplied by the Michał Friedman Foundation (suggests the licence came through that foundation; the text does not say so, so I). | V (footers); I (source of the grant) |
-| M12 | row i of the parquet is the i-th file in the CLARIN listing (decision record 1000_novels.yaml) | partly. This is a validated mapping, not a CLARIN-stated fact: the decision record says 990 of 1,000 texts contain a title word from the paired file name and the 10 exceptions are by the file name's author | Local decision record `repo_patch_suggestions/decisions/1000_novels.yaml` line 14. CLARIN item exposes the 1000 files in the ORIGINAL bundle (local.files.count 1000); the order of the listing was not recorded from a CLARIN page | `polish-dynaword-v2/repo_patch_suggestions/decisions/1000_novels.yaml` ; CLARIN item as in M1 | n/a (local) | "990 of 1,000 texts contain a title word from the paired file name" (decision record) — the mapping was checked by heuristic. | V (record wording); I (that the order is stable) |
-| M13 | death years in `scripts/novels_rights_screen.py` are hand-curated | partly. Author years are hand-curated in the script; translator death years are not (Friedman and Hōjun are `None`, assumed in copyright) | Local script, `CUTOFF = 1955`; translator dict `T` has `'Michał Friedman': None, 'Hōjun': None`; `IN_COPYRIGHT_TRANSLATOR = {'Michał Friedman', 'Hōjun'}`; Dąbrowska entry `('Maria Dąbrowska', 1965, 'pl')` | `polish-dynaword-v2/scripts/novels_rights_screen.py` | n/a (local) | n/a | V |
+| M12 | row i of the parquet is the i-th file in the CLARIN listing (decision record 1000_novels.yaml) | partly. This is a validated mapping, not a CLARIN-stated fact: the decision record says 990 of 1,000 texts contain a title word from the paired file name and the 10 exceptions are by the file name's author | The paper repository's decision record `repo_patch_suggestions/decisions/1000_novels.yaml` line 14. CLARIN item exposes the 1000 files in the ORIGINAL bundle (local.files.count 1000); the order of the listing was not recorded from a CLARIN page | the paper repository's decision records (`repo_patch_suggestions/decisions/1000_novels.yaml`); CLARIN item as in M1 | n/a (local) | "990 of 1,000 texts contain a title word from the paired file name" (decision record) — the mapping was checked by heuristic. | V (record wording); I (that the order is stable) |
+| M13 | death years in `scripts/novels_rights_screen.py` are hand-curated | partly. Author years are hand-curated in the script; translator death years are not (Friedman and Hōjun are `None`, assumed in copyright) | Local script, `CUTOFF = 1955`; translator dict `T` has `'Michał Friedman': None, 'Hōjun': None`; `IN_COPYRIGHT_TRANSLATOR = {'Michał Friedman', 'Hōjun'}`; Dąbrowska entry `('Maria Dąbrowska', 1965, 'pl')` | the paper repository's `scripts/novels_rights_screen.py` | n/a (local) | n/a | V |
 | M14 | death year: Maria Dąbrowska d. 1965 | supported | Wikidata Q242423, property P570 (date of death) 1965-05-19 | https://www.wikidata.org/wiki/Q242423 | http://web.archive.org/web/20251115231935/https://www.wikidata.org/wiki/Q242423 | P570: "19 May 1965" (Wikidata value, preferred rank) | V |
 | M15 | death year: Michał Friedman | found: d. 24 May 2006 (so in copyright to the end of 2076 on a life+70 reading) | Wikidata Q11778742 (P570 2006-05-24; BnF authority 16572553r) and Polish Wikipedia "Michał Friedman". Identity link: pl.wikipedia says his last translation was a collection of Talmudic aggadot (2005); WL lists "Friedman, Michał" as translator of "Agady talmudyczne". A second Wikidata item with the same English label, Q95848015, exists and was not used | https://www.wikidata.org/wiki/Q11778742 ; https://pl.wikipedia.org/wiki/Micha%C5%82_Friedman | Q11778742 http://web.archive.org/web/20250905045440/https://www.wikidata.org/wiki/Q11778742 ; pl.wikipedia http://web.archive.org/web/20251127132228/https://pl.wikipedia.org/wiki/Micha%C5%82_Friedman | pl.wikipedia: "zm. 24 maja 2006 w Warszawie … tłumacz z języka hebrajskiego i jidysz" — died 24 May 2006 in Warsaw, translator from Hebrew and Yiddish. Same article: "Jego ostatnim tłumaczeniem był zbiór kilkuset Agad talmudycznych (2005)." — his last translation was a collection of several hundred Talmudic aggadot. | V (dates); I (the life+70 arithmetic; the statute was not fetched) |
 | M16 | death year: Hōjun | not found | No Wikidata item and no pl.wikipedia article found for "Hōjun" (a Zen name, apparently a pseudonym). WL translator page https://wolnelektury.pl/katalog/autor/hojun/ lists the three texts and no biography. WL `dc:date` 2013-05-21 and 2013-08-30 and `dc:source` "[dostęp: 09.09.2013]" show the translations were published online in 2013, so the translator was alive then | https://wolnelektury.pl/katalog/autor/hojun/ ; https://wolnelektury.pl/media/book/xml/dogen-kigen-umysl-drogi.xml | translator page: NONE (not looked up) ; XML as in M9 | `dc:source`: "tłum. Hōjun, [dostęp: 09.09.2013]" — translated by Hōjun, accessed 9 Sept 2013. A death year is not needed to conclude the translation is in copyright; the conclusion rests on the 2013 date (V) and on the life+70 rule (I). | V (2013 date); I (legal conclusion) |

artifacts/rights_audit_2026-10/reports/citations/parliamentary.md

@@ -135,7 +135,7 @@ C7. Card / registry `created`: "1991-01-01, 2019-12-31" is a dataset-wide consta
 
 C8. Plenary "~14%" is a share of chars; by docs plenary is 0.46% (1,500 docs) (G8).
 
-C9. Dataset NOTICE / card for `sejm_committee_transcripts` omit the Sejm's attribution condition and the commercial-offer bar (T2-T4, T7). Whether they apply is open (T5).
+C9. Dataset NOTICE / card for `sejm_committee_transcripts` omit the Sejm's attribution condition and the commercial-offer bar (T2-T4, T7). Whether they apply is not established (T5).
 
 C10. SpeakLeash manifest has 343,797 docs; the card ships 324,622 (K5). Unexplained here.
 

artifacts/rights_audit_2026-10/reports/citations/sejm_sis.md

@@ -77,7 +77,7 @@ Access notes: www.sejm.gov.pl and orka.sejm.gov.pl return an F5 "Human Verificat
 | L10 | Art. 12 ust. 2-3: conditions set for information made available otherwise than via BIP/portal are an offer; reuse of information made available under art. 5 pkt 1 is acceptance of that offer. Art. 11 ust. 1 requires conditions to be published in BIP category "Ponowne wykorzystywanie". So a website notice can bind as an ODA offer, but only for conditions the Act allows (L5-L9). | V (text); effect is I | `sis_oda_DU2023_1524.txt:497-499, 524-531` | "stanowią ofertę zawierającą te warunki" |
 | L11 | The SIS commercial bar therefore does not fit the closed list in art. 15 ust. 1 and runs against arts. 14 ust. 1, 16 and 17. It is enforceable, if at all, only through rights the Kancelaria holds (art. 14 ust. 2): copyright (but see L13) or the database right (L12). Unverified by any authority; no commentary or case law read. | I | L5-L10 | n/a |
 | L12 | Database Act (DU 2024/1769; earlier text DU 2021/386): producer = who bears investment risk (art. 2 ust. 1 pkt 4); art. 6 ust. 1 exclusive right to extract/re-use all or a substantial part; art. 7 ust. 1 and 4: a lawful user cannot be barred from insubstantial parts for any purpose and contrary contract terms are void; art. 8 ust. 1 pkt 2: illustration/teaching/research use of a substantial part with source, if non-commercial; art. 8a: reproduction for text and data mining allowed unless the rightholder reserved otherwise (machine-readable reservation for publicly accessible online databases); art. 10: 15 years from creation. Whether the Kancelaria is a "producer" of a protected database (substantial investment) is not stated anywhere I read. The SIS commercial sentence is human-readable HTML text; no machine-readable reservation was found on api.sejm.gov.pl (C1, C2). Art. 8a covers reproduction for TDM, stored only as long as needed (ust. 3), not redistribution of a corpus. | V (statutory text); application is I | `sis_dbact_DU2024_1769.txt:45-62, 112-130, 147-175, 194`; `sis_dbact_DU2021_386.txt` | "Postanowienia umowne sprzeczne z ust. 1 i 2 są nieważne" |
-| L13 | Copyright: official documents and materials are outside copyright (Copyright Act art. 4 pkt 2, Dz.U. 2025 poz. 24; earlier pass `parliamentary.md` S1, S11). Committee and plenary records are by statute "urzędowe stwierdzenie przebiegu obrad" (art. 166 ust. 1; art. 176 ust. 1). If no copyright subsists, a notice cannot create an exclusive right in the text; the SIS bar would then rest on contract/ODA offer only (L10, L11). Whether guests' statements keep their own copyright is open (earlier pass S10, S14). | V (statutes); conclusions I | `sis_regulamin_sejmu_MP2026_573.txt:2896-2898, 3074`; `parliamentary.md` S1, S10, S11 | "stanowi urzędowe stwierdzenie przebiegu obrad" |
+| L13 | Copyright: official documents and materials are outside copyright (Copyright Act art. 4 pkt 2, Dz.U. 2025 poz. 24; earlier pass `parliamentary.md` S1, S11). Committee and plenary records are by statute "urzędowe stwierdzenie przebiegu obrad" (art. 166 ust. 1; art. 176 ust. 1). If no copyright subsists, a notice cannot create an exclusive right in the text; the SIS bar would then rest on contract/ODA offer only (L10, L11). Whether guests' statements keep their own copyright is not established (earlier pass S10, S14). | V (statutes); conclusions I | `sis_regulamin_sejmu_MP2026_573.txt:2896-2898, 3074`; `parliamentary.md` S1, S10, S11 | "stanowi urzędowe stwierdzenie przebiegu obrad" |
 | L14 | The SIS page is a unilateral website notice by the Kancelaria: not a statute, a regulation or a licence naming an ODA art. 15 ust. 3 standard open licence. Its binding force on someone who never visited www.sejm.gov.pl (API or intermediary user) is unproven. | I | W1, D4, C1 | n/a |
 | L15 | The attribution clause (source note, "Źródłem tym jest System Informacyjny Sejmu") fits ODA art. 15 ust. 1 pkt 1 and is the lawful kind of condition if it applies at all. The NOTICEs already name "Kancelaria Sejmu RP" and api.sejm.gov.pl (C9) but not the SIS wording. | I | L7, C9 | n/a |
 
@@ -104,7 +104,7 @@ Overall: no source can be said to be subject to the SIS clauses ("yes"), and non
 Audit statement: "Sejm SIS terms require attribution and bar use in a commercial offer (reach to API/PDF unverified)."
 
 1. "Sejm SIS terms": supported for the "System Informacyjny Sejmu" page on www.sejm.gov.pl (V, W1-W4; Sejm8 from 2016, Sejm9, Sejm10). The three-clause wording is not old: the 2015-11-17 and 2015-11-21 pages carried a permissive sentence and no commercial bar (V, H1-H2). The bar appears between 2015-11-21 and 2016-02-04 (H3, H6).
-2. "require attribution": the clause is scoped to material "pobieranych ze strony www.sejm.gov.pl" and prescribes the source "System Informacyjny Sejmu" (W2, W3). Scoped to the website, so its reach to API downloads is the same open question as the bar.
+2. "require attribution": the clause is scoped to material "pobieranych ze strony www.sejm.gov.pl" and prescribes the source "System Informacyjny Sejmu" (W2, W3). Scoped to the website, so its reach to API downloads is as unsettled as the bar's.
 3. "bar use in a commercial offer": the text is narrower than a general commercial-use ban. It bars SIS material from being "elementem komercyjnej oferty" of an agency, TV station, portal or other entity, with iTV Sejm broadcasts as the named example (W4). It says nothing of datasets, open releases or training. Whether a free dataset is a "commercial offer" is interpretation (I).
 4. "reach to API/PDF unverified": now checked. V: API docs, response headers and three sample payloads carry no terms (C1-C3); the Regulamin and Zarządzenie 3/2006 define SIS without naming the API (D1, D4, D5); the Sejm site footer lists the API and SIS as separate links (C4). I: placement of committee transcripts and interpellations in SIS by the Regulamin (D6, D7) is the only textual bridge.
 5. Missing from the statement: enforceability. ODA art. 15 ust. 1 gives a closed list of conditions that does not include a commercial-use bar; arts. 14 ust. 1, 16, 17 point the same way; art. 15 ust. 2 allows commercial-use limits only for museums, libraries, archives (L5-L9, V text; conclusion I). Sui generis database rights are the only plausible hook and only if the Kancelaria holds them (L6, L12); the Database Act voids contract terms that bar insubstantial parts (art. 7 ust. 4). Nothing read says the Kancelaria is a database producer.

artifacts/rock_pollub_pl_ontology_manifest.json

@@ -172,7 +172,7 @@
     "full collection inventory",
     "cross-source deduplication",
     "benchmark contamination check",
-    "legal review of third-party textual excerpts",
+    "rights review of third-party textual excerpts",
     "controlled training ablation"
   ],
   "protocols": [

artifacts/source_candidate_audit_v0_3.json

@@ -696,7 +696,7 @@
     "configured_decision": "maybe_after_legal_review",
     "target_bucket": "web_fallback",
     "priority": "medium",
-    "notes": "CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and legal review.",
+    "notes": "CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and a rights review.",
     "gated": false,
     "private": false,
     "downloads": 1741,

artifacts/source_candidate_audit_v0_3.md

@@ -181,7 +181,7 @@ README/license signals:
 - license tags: `cc-by-4.0`
 - language tags: `none`
 - gated: `False`
-- notes: CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and legal review.
+- notes: CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and a rights review.
 
 README/license signals:
 - license: cc-by-4.0

artifacts/source_findings.md

@@ -387,7 +387,7 @@ legal basis: judgments are *materiały urzędowe* outside copyright under art. 4
 pkt 2 pr. aut., redistributed by SAOS (CeON/ICM UW + Fundacja ePaństwo) through a
 bulk-dump API built for exactly this.
 
-Open question before building: judgments are court-anonymised, but that should be
+To check before building: judgments are court-anonymised, but that should be
 **verified on sampled text**, not assumed — this is the one candidate here where
 personal data is a live risk rather than a formality.
 
@@ -1023,7 +1023,7 @@ say so in the datasheet instead of pretending a before/after can be regenerated.
 - CLARIN-PL (email reply, 2026-10-05): handle `11321/312` holds metadata only;
   CLARIN-PL does not distribute the texts and holds no documentation of rights in
   them. The record's CC BY 4.0 therefore licenses nothing at the text level.
-- Per-text screen (`polish-dynaword-v2/evidence/1000_novels_rights_screen.csv`):
+- Per-text screen (`evidence/1000_novels_rights_screen.csv` among the paper's ancillary files):
   825 public domain by term, 12 by the retained Wolne Lektury footer, 17 in
   copyright, 146 unresolved.
 - Header credit lines beat names found elsewhere in the text. Four rows changed
@@ -1176,7 +1176,7 @@ tag). 45,043 sentences / 994,922 tokens dropped. The data, the registry entry,
 `artifacts/common_voice_pl_ontology_manifest.json` and
 `artifacts/common_voice_pl_language_audit.json` were removed. No legal-note stub was
 kept, because nothing was released. The audit's alternative, dropping only the Europarl
-sentences, was not chosen. The open question of sentence-level units is moot.
+sentences, was not chosen. The question of sentence-level units is moot.
 
 **Reason** — CC0 is Common Voice's platform-level label; the sentences were never shown
 to be CC0. Matching the shard against Common Voice's Polish sentence files gave these
@@ -1187,7 +1187,7 @@ reproduce only with NFKC + casefold + removal of all non-word characters
 - Europarl: the European Parliament legal notice allows reuse with source
   acknowledgement, and asks for the URL of the complete item when only part is
   reproduced. The shard carries neither, and an attribution condition cannot sit under
-  CC0. Mozilla's 2020 legal review of Europarl was never shown as completed. The Common
+  CC0. Mozilla's 2020 rights check of Europarl was never shown as completed. The Common
   Voice README's CC0 sentence does not cover the Europarl file.
 - Sentence Collector: volunteers attest these lines are public domain, but their sources
   are not recorded. `sentence_domain` is "unspecified" for 45,035 of 45,043 rows.
@@ -1232,7 +1232,7 @@ quotation site that is the framing, not the quoted text.
 ## wikisource — withdrawn (2026-10-05)
 
 **Verdict** — `removed`, whole source, from main and from v0.2.6; a rebuild for a later
-release is tracked separately (polish-dynaword-v2 #68). Released in v0.2.0 and in every
+release is planned. Released in v0.2.0 and in every
 tag through v0.2.5; since 2026-10-21 no tag contains it. 632,005 documents / 801,947,427 tokens
 dropped. As with HPLT, the registry keeps its `release: "0.2.0"` entry with a
 `legal_note`. Its `traceable` text no longer claims every text is public domain, and
@@ -1263,7 +1263,7 @@ screened and labelled.
 
 ## biblioteka_nauki — public-domain-label records dropped (2026-10-06)
 
-**Verdict** — `added`, kept for v0.2.6 minus 3 records (polish-dynaword-v2#52). 42,071 →
+**Verdict** — `added`, kept for v0.2.6 minus 3 records. 42,071 →
 42,068 documents, 673,488,831 → 673,112,676 tokens (−376,155), 1,748,333,935 →
 1,747,387,478 characters. `src/clean_biblioteka_nauki.py` filters the v0.2.5 Parquet
 (pinned by SHA-256): it keeps `cc-by`, `cc-by-sa` and `cc0`, drops `public-domain`, and
@@ -1282,7 +1282,7 @@ by Muzeum Niepodległości w Warszawie 2019". That the authors are alive is infe
 **Not done here** — the licence version is still collapsed: rows say `cc-by`/`cc-by-sa`,
 the card and `stats.json` say 4.0. A live re-check (seed 20261006, 24 rows) found 4 of 12
 `cc-by` rows under 3.0 PL or 3.0 Unported; all 12 `cc-by-sa` rows were 4.0. Fixing it needs
-a per-record API re-fetch of name and version (tracked under polish-dynaword-v2#57). The
+a per-record API re-fetch of name and version. The
 card still names SpeakLeash while `src/sources.py` says direct rebuild; `licenses.jsonl`,
 `index.jsonl` and `build_parquet.py` are not in the repo. 128 kept rows (88 `cc-by`, 40
 `cc-by-sa`; 1.56% of tokens) contain "Wszelkie prawa zastrzeżone" or "All rights
@@ -1293,7 +1293,7 @@ the label, and check the date: a 2019 work cannot be public domain by term in Po
 
 ## govpl — PAP reprints dropped (2026-10-06)
 
-**Verdict** — `kept`, minus 188 PAP reprints (slayerlabs/polish-dynaword-v2#51).
+**Verdict** — `kept`, minus 188 PAP reprints.
 88,190 → 88,002 documents, 80,666,129 → 80,445,758 tokens (−220,371). `release` stays
 `0.2.2`. `src/clean_govpl.py` filters the pinned v0.2.5 Parquet, so ids are unchanged,
 and writes `govpl.attribution.jsonl` with each row's gov.pl URL, subsite, title and date.
@@ -1324,7 +1324,7 @@ site, check source lines for agency copy before taking the footer at face value.
 
 ## govpl — other third-party texts dropped (2026-10-19, update to the section above)
 
-**Verdict** — `kept`, minus 186 more rows (slayerlabs/polish-dynaword-v2#106).
+**Verdict** — `kept`, minus 186 more rows.
 88,002 → 87,816 documents, 217,515,240 → 216,869,817 characters (−645,423),
 80,445,758 → 80,204,973 tokens (−240,785). `src/clean_govpl.py` applies
 `data/govpl/govpl.third-party-review.csv` (pinned by SHA-256) after the PAP rule, so ids
@@ -1375,7 +1375,7 @@ of a site publishes, not only who publishes it.
 **Verdict** — kept, relabelled, 144 rows dropped: 72,472 → 72,328 documents,
 40,011,099 → 39,890,244 tokens (baseline: main after #46, not the v0.2.5 tag's
 72,771). The basis is now called an operator-declared platform licence. A question
-to KPRM is pending (polish-dynaword-v2 #49); its answer may still change the
+to KPRM is pending; its answer may still change the
 source's status. The drops are in `src/clean_samorzad_gov_pl.py`
 (`third_party_drop`, applied by `--refilter-release` to the pinned release):
 - 69 "Piątek z poezją" posts (all Powiat Sokołowski): poems credited to their
@@ -1400,7 +1400,7 @@ Credits matter here: 810 documents have an `Źródło:` line, but in 405 every c
 only a de-identified `<URL>`, so their sources cannot be classified; 231 documents
 mention IMGW without the verbatim notice and were not reviewed.
 
-References: polish-dynaword-v2 #49;
+References:
 `rights_audit_2026-10/reports/govweb.md` L8-15; `reports/citations/samorzad_nonsa.md`
 S1, S8, C4; <https://www.gov.pl/web/gov/prawa-autorskie> (read 2026-10-05).
 
@@ -1424,7 +1424,7 @@ Our assessment is that the footer alone does not license the institutions' edito
 texts, which is the HPLT structure: a grant from someone other than the authors.
 Dropping slices cannot fix that, since it covers the whole source.
 
-References: polish-dynaword-v2 #49, #103; `rights_audit_2026-10/reports/govweb.md`
+References: `rights_audit_2026-10/reports/govweb.md`
 L8-15; `reports/citations/samorzad_nonsa.md` S1, S8.
 
 **Lesson** — a platform footer is the operator's statement, not the authors' grant.
@@ -1433,7 +1433,7 @@ confirmation before admitting.
 
 ## global_voices — partner republications dropped (2026-10-06)
 
-**Verdict** — `admitted` with 114 posts withheld (polish-dynaword-v2 #50). The source
+**Verdict** — `admitted` with 114 posts withheld. The source
 goes from 2,040 docs / 10,073,628 chars / 3,715,020 tokens to 1,926 / 9,358,071 /
 3,453,918. The posts come from 51 outlets. The new `src/clean_global_voices.py` filters
 the v0.2.5 Parquet and sidecar, pinned by SHA-256. `src/test_global_voices_contract.py`
@@ -1510,7 +1510,7 @@ the byline.
 
 **Investigated:** 2026-10-06
 **Verdict:** keep 888 of 1,000 texts as `public-domain` (was 837); drop 18 in copyright
-and 94 unresolved. Tracks slayerlabs/polish-dynaword-v2#44.
+and 94 unresolved.
 
 ### Verified
 
@@ -1519,7 +1519,7 @@ and 94 unresolved. Tracks slayerlabs/polish-dynaword-v2#44.
   first rebuild this way. An unknown author now fails closed unless the work is listed
   as anonymous.
 - Translator death years now come from Biblioteka Narodowa / VIAF authority records,
-  cited in `polish-dynaword-v2/scripts/novels_rights_screen.py`. Józef Pracki d. 1882,
+  cited in the paper repository's `scripts/novels_rights_screen.py`. Józef Pracki d. 1882,
   Edmund Cięglewicz d. 1928, Stanisław Sierosławski d. 1941, Bolesław Londyński d. 1928
   (22 texts), Ludwik Jenike d. 1903, Adam Stögbauer d. 1916, Kazimierz Bronikowski
   d. 1909, Maria Cumft-Pieńkowska d. 1920, Izydor Berman d. 1942, Konrad Drzewiecki
@@ -1541,7 +1541,7 @@ and 94 unresolved. Tracks slayerlabs/polish-dynaword-v2#44.
 
 - "A. L." (`1000_novels_0734`, Schopenhauer 1901): BN records the cryptonym only. The
   row still rests on its Wolne Lektury footer (Inferred).
-- Edition rights (Art. 99¹–99²); dedup against `wolne_lektury` / `eltec_pol` (#45).
+- Edition rights (Art. 99¹–99²); dedup against `wolne_lektury` / `eltec_pol`.
 
 **Lesson:** a Wolne Lektury public-domain footer is a claim, not evidence. One of the
 12 it covered had a translator who died in 1960.
@@ -1558,8 +1558,8 @@ unresolved, so the release keeps 887 texts and drops 18 in copyright and 95 unre
 **Verdict** — kept, 63 documents dropped: 6,141 → 6,078 documents, 103,009,797 →
 90,654,820 tokens (input: the v0.2.5 Parquet, unchanged since v0.2.1). Each dropped
 document is a novel, or one tome of a novel, that `eltec_pol` also carries. 50 ELTeC
-novels are concerned, and `eltec_pol` stays whole (maintainer decision A,
-polish-dynaword-v2 #72). The filter is `src/clean_wolne_lektury.py` and the pair list is
+novels are concerned, and `eltec_pol` stays whole (maintainer decision).
+The filter is `src/clean_wolne_lektury.py` and the pair list is
 `data/wolne_lektury/wolne_lektury.eltec_pairs.tsv`.
 - Score: the share of a WL document's 5-word shingles found in the ELTeC novel that
   shares the most with it. Tokenisation follows `near_dedup`, and a trailing WL footer is
@@ -1585,10 +1585,9 @@ polish-dynaword-v2 #72). The filter is `src/clean_wolne_lektury.py` and the pair
 
 **Reason** — 51 of the 100 ELTeC novels duplicate WL texts. ELTeC's CC BY 4.0 licence
 is the clean one (`rights_audit_2026-10/reports/literary.md`: eltec_pol CLEAN). WL
-licence labels are not touched here; #57 relabels them per row.
+licence labels are not touched here; the v0.2.6 rights audit relabels them per row.
 
-References: polish-dynaword-v2 #72 (decision), #45 (method);
-`rights_audit_2026-10/literary/eltec_meta.tsv` (ELTeC file names and titles).
+References: `rights_audit_2026-10/literary/eltec_meta.tsv` (ELTeC file names and titles).
 
 **Lesson** — Jaccard near-dedup misses a novel that one source splits into tomes.
 Measure containment from the split side.
@@ -1599,7 +1598,7 @@ Measure containment from the split side.
 characters, 51,981,548 → 10,759,728 tokens (input: Hub PR 65; references: main at
 563eea8, after Hub PR 67). Each dropped text is already in `wolne_lektury` (629) or
 `eltec_pol` (37); none is in both, since Hub PR 67 dropped the WL copies of ELTeC novels
-(before it, 35 of the 37 were in both). Maintainer decision on polish-dynaword-v2 #45:
+(before it, 35 of the 37 were in both). Maintainer decision:
 drop the 1000_novels copy and keep both other sources whole. The filter is the dedup
 step of `src/clean_1000_novels.py`, run after the rights screen. Covers and matched ids
 for all 887 screened texts are in `data/1000_novels/1000_novels.dedup.tsv`.
@@ -1636,7 +1635,7 @@ for all 887 screened texts are in `data/1000_novels/1000_novels.dedup.tsv`.
   translation; WL has Boy's; 0.0388), Grabiński's *Na wzgórzu róż* (`0239`, 0.0043),
   Kraszewski's *Dziad i baba* (`0359`, a prose tale; WL has the poem; 0.0079) and
   Przybyszewski's *Nad morzem* (`0690`; WL has a 354-word piece; 0.0303).
-- Hub PR 67 (#72) merged first. Re-pinned to its wolne_lektury Parquet (SHA-256
+- Hub PR 67 merged first. Re-pinned to its wolne_lektury Parquet (SHA-256
   `f308d492…`, was `5833c25d…`) and rebuilt: the same 666 drops and the same Parquet. The
   split went from WL 664, ELTeC 37, both 35 to WL 629, ELTeC 37, both 0; 216 WL covers
   change. The 5 texts hand-checked between 0.3 and 0.7 and the 10 sampled drops stay WL
@@ -1645,8 +1644,6 @@ for all 887 screened texts are in `data/1000_novels/1000_novels.dedup.tsv`.
 **Reason** — 666 of the 887 public-domain texts, 41.2M tokens, are copies of documents
 that `wolne_lektury` or `eltec_pol` already release.
 
-References: polish-dynaword-v2 #45 (measurement and decision), #72 (wolne_lektury side).
-
 **Lesson** — score against the union of a source's documents. Wolne Lektury splits
 novels into tomes, and no single tome need hold half of the text.
 
@@ -1668,8 +1665,7 @@ the 221. A faithful copy of a public-domain text creates no new rights; the open
 question is what else the file holds. Our assessment is that this is too high a risk
 without the edition, and that none of the 221 can be confirmed from the metadata we have.
 
-References: polish-dynaword-v2 #104, #105; #43, #44 (rights screen), #45 (dedup);
-`polish-dynaword-v2/evidence/1000_novels_rights_screen.csv`.
+References: `evidence/1000_novels_rights_screen.csv` among the paper's ancillary files.
 
 **Lesson** — a public-domain author is not a public-domain file. Record the edition a
 text follows when it is admitted, or the text cannot be cleared later.
@@ -1677,8 +1673,7 @@ text follows when it is admitted, or the text cannot be cleared later.
 ## wikipedia — Polish Wikipedia (rebuild from the official dump)
 
 **Investigated** — 2026-10-05 (October 2026 rights audit, `artifacts/rights_audit_2026-10/reports/wikis_ugc.md`
-L44-46, `reports/citations.md` L17, `reports/citations/wikimedia.md` section 5;
-slayerlabs/polish-dynaword-v2#69).
+L44-46, `reports/citations.md` L17, `reports/citations/wikimedia.md` section 5).
 
 **Verdict** — `added`, rebuilt for v0.2.6 from the official Wikimedia dump; the SpeakLeash
 `plwiki` intermediary is removed. 1,468,656 documents / 3,140,841,847 characters /
@@ -1720,7 +1715,7 @@ decision. Per page in `data/wikipedia/wikipedia.copyvio-sample.jsonl`.
 compare literal ids against the old `data/wikipedia` (revision `fe44303`, 1,171,897 rows) and
 are stale after the id change.
 
-**Blockers** — none to merge. Open: the two pasted pages; the time of day of the 4.0 switch;
+**Blockers** — none to merge. Known limits: the two pasted pages; the time of day of the 4.0 switch;
 GFDL-era text; `{{ek}}` pages kept; appendix headings left in 56.4% of documents (2.96% of
 characters from the first such heading on) and whitespace-only lines in 11.9%; NERGAL pass
 before release.
@@ -1772,9 +1767,9 @@ the "speech-level provenance" line in the card was wrong. That is now a test in
   the live API (the audit asked for 20 rows to be spot-checked; not done here).
 - Whether the www.sejm.gov.pl terms reach API output; the API documents no licence.
 - (Fixed in a follow-up commit: the card's "Temporal novelty" paragraph named
-  `parlamint_pl` as registered; it left the registry in #54.)
+  `parlamint_pl` as registered; it left the registry.)
 
-References: polish-dynaword-v2 #57; `rights_audit_2026-10/REVIEW.html` `#sejm_api`;
+References: `rights_audit_2026-10/REVIEW.html` `#sejm_api`;
 `reports/citations/sejm_sis.md`.
 
 **Lesson** — a card line that says "speech-level" must be backed by a test on the
@@ -1808,8 +1803,8 @@ exist. Pin the bytes when the source is first read.
 
 ## nonsa — third-party quotations and `<poem>` pages dropped (2026-10-06)
 
-**Verdict** — kept as `release: None` (admission waits for the NERGAL pass,
-polish-dynaword-v2 #60), 612 documents dropped: 12,304 → 11,692 documents,
+**Verdict** — kept as `release: None` (admission waits for the NERGAL pass),
+612 documents dropped: 12,304 → 11,692 documents,
 16,883,883 → 15,439,683 tokens, 44,279,318 → 40,567,986 characters. Rebuilt from the
 pinned wikiteam3 dump (`nonsa.pl-20251009-history.xml.zst`, sha1 99ed6dc7…, verified
 by the script) with `src/build_nonsa_pl.py`; the retained rows equal the old shard
@@ -1845,7 +1840,7 @@ and wikisource, withdrawn 2026-10-05).
   deletions), and the rows are still in the shard.
 - No `ns` column in the parquet; the namespace is in `nonsa.attribution.jsonl`.
 
-References: polish-dynaword-v2 #60; `rights_audit_2026-10/REVIEW.html` (nonsa);
+References: `rights_audit_2026-10/REVIEW.html` (nonsa);
 `rights_audit_2026-10/wikis/nonsa_ns.json`; `reports/citations/samorzad_nonsa.md`.
 Licence read-only 2026-10-06: <https://nonsa.pl/wiki/Nonsensopedia:Prawa_autorskie>
 (revid 1850937, 2026-03-02) states Creative Commons Attribution-Share Alike 3.0, and
@@ -1900,14 +1895,14 @@ writing the filter, and say which drops are confirmed.
 **Verdict** — held (`release: None`). 118,902 → 32,460 documents, 135,072,403 →
 30,766,807 tokens, 361,275,710 → 81,385,172 characters; parquet sha256
 `8fad4b4eab069fb8a652f9dc6a2e7dbc8f49d1e1b6bfae0decd2b318d160a20a`. Admission waits for
-the NERGAL pass (#60). Registry, card and NOTICE now say "Sejm API data via contributor
+the NERGAL pass. Registry, card and NOTICE now say "Sejm API data via contributor
 snapshot" and that `source_url` is a day-level listing, not a per-speech URL.
 
 **Reason** — the source overlaps `parliamentary` (0.2.0) and `sejm_api` (0.2.5).
 `src/clean_sejm_api_2011_2022.py` drops a speech when one document of a pinned reference
 Parquet holds at least half of its 5-word shingles (containment >= 0.5; exact, no
 MinHash). 86,442 speeches (73%) are dropped, 86,272 of them dated 2011-2019. By
-reference set, with the maintainer's two #54 outcomes:
+reference set, with the maintainer's two outcomes for `parliamentary`:
 
 | Reference set | Dropped |
 |---|---:|
@@ -1916,8 +1911,8 @@ reference set, with the maintainer's two #54 outcomes:
 | `sejm_api` alone (outcome A, `parliamentary` withdrawn) | 216 |
 | `sejm_api` + `parliamentary` without its committee slice (outcome B) | 48,833 |
 
-Per-speech scores are in `sejm_api_2011_2022.decisions.jsonl`. If #54 changes
-`parliamentary`, re-pin `REFERENCES` and rerun the script on the restored inputs.
+Per-speech scores are in `sejm_api_2011_2022.decisions.jsonl`. If
+`parliamentary` changes, re-pin `REFERENCES` and rerun the script on the restored inputs.
 Hand check of 20 pairs at 0.3-0.7: 14 carry the speech's own wording in the reference
 (4 of 10 under 0.5, 10 of 10 from 0.5); the 6 that do not are dated 2020 or later.
 
@@ -1938,7 +1933,7 @@ Hand check of 20 pairs at 0.3-0.7: 14 carry the speech's own wording in the refe
 - `src/fetch_sejm_api_2011_2022.py` is unchanged because the ontology manifest pins its
   hash; rebuilding with it brings back the old wording and the 118,902 rows.
 
-References: polish-dynaword-v2 #57, #54, #60; `rights_audit_2026-10/REVIEW.html`
+References: `rights_audit_2026-10/REVIEW.html`
 `#sejm_api_2011_2022`.
 
 **Lesson** — "contained in a released source" needs a size floor and a date check: a
@@ -1946,8 +1941,8 @@ containment score of one stock sentence says nothing about reuse.
 
 ## ted_pl: two licence bases cited, text rebuilt from the TED API with fixed PII patterns (2026-10-06)
 
-Audit actions from slayerlabs/polish-dynaword-v2#57; stays `release: None` until the
-NERGAL pass (#60).
+Audit actions from the v0.2.6 rights audit; stays `release: None` until the
+NERGAL pass.
 
 ### Verified
 
@@ -1989,7 +1984,7 @@ download recipe runnable: a rescrub of already-redacted text cannot undo a false
 ## wikinews — Polish Wikinews (rebuild from the official dump, 2.5/4.0 labels)
 
 **Investigated** — 2026-10-06 (October 2026 rights audit, `artifacts/rights_audit_2026-10/reports/wikis_ugc.md`,
-`reports/citations.md` L17-26; slayerlabs/polish-dynaword-v2#57).
+`reports/citations.md` L17-26).
 
 **Verdict** — `kept`, relabelled; rebuilt from the official dump so the label follows the
 page's last revision. Documents, characters and tokens are unchanged from v0.2.5
@@ -2015,19 +2010,19 @@ Phabricator T381421, the 2.5 legal code (§4(a): no later-version clause), pl.wi
 Cutoff is 2024-12-16 00:00 UTC for the whole day: 6 documents were last edited that day, 2 of
 them before 08:32:24 UTC and so strictly under 2.5. Inferred: because CC BY 2.5 has no
 later-version clause, the `CC-BY-4.0` label for post-switch pages rests on the site's terms
-page alone. This is an open question, separate from E1 (wiki 3.0/4.0 labelling,
-v2#56), whose answer is still pending. The terms page also says articles created before
+page alone.
+The terms page also says articles created before
 2005-09-25 are public domain; 0 released documents were last edited before that date, so the
 clause does not change a label here. Evidence in `data/wikinews/wikinews.license-evidence.json`.
 
-**Blockers** — none to merge. Open: the 2.5 → 4.0 label and E1; the 2 boundary
+**Blockers** — none to merge. Known limits: the 2.5 → 4.0 label; the 2 boundary
 documents; copyright pastes in news text (the audit's 100-document press-copying sample was
 not run); the markup residue above.
 
 ## eurlex — relabelled to the EU reuse policy, 92 IASB documents dropped (2026-10-06)
 
-**Investigated** — 2026-10-06 (October 2026 rights audit, polish-dynaword-v2 #57; open
-question v2#56 B5; `rights_audit_2026-10/reports/citations.md`).
+**Investigated** — 2026-10-06 (October 2026 rights audit,
+`rights_audit_2026-10/reports/citations.md`).
 
 **Verdict** — kept and relabelled, the 92 IASB-notice documents dropped (maintainer
 decision 2026-10-06): 243,060 -> 242,968 documents, 5,976,949,249 -> 5,902,889,135
@@ -2077,7 +2072,7 @@ Ids are in `data/eurlex/eurlex.third-party-scan.json`.
 
 **Not established** — SpeakLeash's terms for `eurlex_corpus`; whether the ECB, EESC,
 Council, Parliament and the Court of Justice publish reuse terms of their own; the
-Publications Office's position on non-Commission documents (open question, v2#56 B5).
+Publications Office's position on non-Commission documents.
 
 **Lesson** — a reuse policy is not a Creative Commons licence: copy the clause, not the
 label, and read what the grant covers before using its name.
@@ -2085,7 +2080,7 @@ label, and read what the grant covers before using its name.
 ## wikibooks — Polish Wikibooks (rebuild from the official dump, 3.0/4.0 labels, GFDL review)
 
 **Investigated** — 2026-10-06 (October 2026 rights audit, `artifacts/rights_audit_2026-10/reports/wikis_ugc.md`,
-`reports/citations.md` L17-26; slayerlabs/polish-dynaword-v2#57).
+`reports/citations.md` L17-26).
 
 **Verdict** — `kept`, relabelled; rebuilt from the official dump so the label follows the
 page's last revision. Documents, characters and tokens are unchanged from v0.2.5
@@ -2107,7 +2102,7 @@ row (`wikibooks.attribution.jsonl`).
 
 **License evidence** — as for wikivoyage: Terms of Use oldid 554823 / 98524, the 3.0 and 4.0
 legal code, pl.wikibooks `rightsinfo` (BY-SA 4.0), read 2026-10-06. Cutoff 2023-06-07 00:00
-UTC; 2 documents were last edited that day. E1 (v2#56) is still pending.
+UTC; 2 documents were last edited that day.
 
 **GFDL** — 435 documents (259,065 tokens, 1.7%) were last edited before 2009-06-15 (the
 start of CC BY-SA on the Wikimedia sites, per the audit); they carry `CC-BY-SA-3.0`, counted
@@ -2120,9 +2115,9 @@ own text is under the GNU FDL (42,834 tokens), 10 name a GFDL-licensed upstream
 13 link the licence page only, 13 discuss licences, 2 match by accident. The largest is
 `wikibooks_30538` (35,619 tokens). No label and no drop follows from the review; the 17
 own-text and upstream documents (49,737 tokens) are the ones where a GFDL-only grant could
-matter, and keeping them is left to the maintainers; it stays an open question.
+matter, and keeping them is left to the maintainers.
 
-**Blockers** — none to merge. Open: E1; the 17 documents above. Two pages seen during the
+**Blockers** — none to merge. Known limits: the 17 documents above. Two pages seen during the
 review and not checked: `wikibooks_27716` (145 tokens) lists third-party course sources with
 their licences, among them CC-BY-NC 3.0 and "All Rights Reserved"; `wikibooks_27071` (1,287
 tokens) says it was moved in 2013 from the "Wolne podręczniki" portal with its creators'
@@ -2131,7 +2126,7 @@ consent and does not state the original licence. The markup residue above also s
 ## wikibooks — 46 pages withheld (update to the section above)
 
 **Investigated** — 2026-10-06 (copyright sample and licence trace of the follow-up, then the
-maintainer's decision; slayerlabs/polish-dynaword-v2#57).
+maintainer's decision).
 
 **Verdict** — `kept`, 46 pages withheld. Maintainer approval 2026-10-06, relayed by the
 coordinator. The numbers in the section above (9,112 documents, 38,892,589 characters,
@@ -2162,17 +2157,17 @@ boundary day and the 435 GFDL-era documents are unchanged. None of the 46 is amo
 documents of `wikibooks.gfdl-review.json`; only its `source_tokens` (15,313,229) and the share
 in its summary (0.80%) were regenerated.
 
-**Blockers** — none. Open: the copyright sample read 79 of the 9,112 pages and 90 more by one
+**Blockers** — none. Known limits: the copyright sample read 79 of the 9,112 pages and 90 more by one
 contributor; `wikibooks_27716` (145 tokens, lists CC-BY-NC and all-rights-reserved sources by
 name; kept) and `wikibooks_27071` (1,287 tokens, original licence not stated; kept);
 the 25 released pages that the "Od lamera do hakera" hub links in other books were not
-traced; E1 (v2#56); the 2 boundary documents; the 17 own-text and upstream GFDL documents;
+traced; the 2 boundary documents; the 17 own-text and upstream GFDL documents;
 the markup residue.
 
 ## wikivoyage — Polish Wikivoyage (rebuild from the official dump, 3.0/4.0 labels)
 
 **Investigated** — 2026-10-06 (October 2026 rights audit, `artifacts/rights_audit_2026-10/reports/wikis_ugc.md`,
-`reports/citations.md` L17-26; slayerlabs/polish-dynaword-v2#57).
+`reports/citations.md` L17-26).
 
 **Verdict** — `kept`, relabelled; rebuilt from the official dump so the label follows the
 page's last revision. Documents, characters and tokens are unchanged from v0.2.5
@@ -2198,17 +2193,17 @@ edited after the switch may carry earlier 3.0 text under 4.0 (3.0 §4(b)(ii)). T
 2023-06-07 00:00 UTC, not the audit's "06-08 if time of day unknown": 06-08 would put that
 day's 4.0 edits under 3.0, which 4.0 §3(b)(1) does not allow; 6 documents were last edited on
 2023-06-07 and may carry 3.0 text under a 4.0 label. The footer switch date was not checked.
-The answer to E1 (v2#56) is still pending. GFDL era (last edit before 2009-06-15): 0
+GFDL era (last edit before 2009-06-15): 0
 documents. Evidence in `data/wikivoyage/wikivoyage.license-evidence.json`.
 
-**Blockers** — none to merge. Open: E1; the 6 boundary documents; copyright pastes (travel
+**Blockers** — none to merge. Known limits: the 6 boundary documents; copyright pastes (travel
 text copied from guidebooks and tourist sites was not sampled for Wikivoyage); the markup
 residue above.
 
 ## wikivoyage — two pages withheld (update to the section above)
 
 **Investigated** — 2026-10-06 (copyright sample of the follow-up, then the maintainer's
-decision; slayerlabs/polish-dynaword-v2#57).
+decision).
 
 **Verdict** — `kept`, two pages withheld: `wikivoyage_1219` (Barcelona/Sagrada Família) and
 `wikivoyage_910` (Królewiec). Maintainer approval 2026-10-06, relayed by the coordinator. The
@@ -2232,16 +2227,16 @@ pages skipped → 13,652 written → 9 non-Polish dropped → 13,643. Diff again
 parquet: only the two ids removed, every other row identical; the 6 documents of the boundary
 day and the 0 GFDL-era documents are unchanged.
 
-**Blockers** — none. Open: the copyright sample read 80 of the 13,645 pages (40 per label),
+**Blockers** — none. Known limits: the copyright sample read 80 of the 13,645 pages (40 per label),
 so 13,565 were not checked; two further pages stay in the release with the source's direction
-unresolved (`wikivoyage_2477`, `wikivoyage_989`); E1 (v2#56); the 6 boundary documents; the
+unresolved (`wikivoyage_2477`, `wikivoyage_989`); the 6 boundary documents; the
 markup residue.
 
 ## wikivoyage — markup residue removed, licence labels by contribution date (update to the sections above)
 
-**Investigated** — 2026-10-17/18 (slayerlabs/polish-dynaword-v2#110, option B: label each record
-by when its text was contributed, not by the page's last revision; slayerlabs/polish-dynaword-v2#116,
-option A: remove category lines and file-link options first, then label the cleaned text).
+**Investigated** — 2026-10-17/18 (the labels by contribution: label each record
+by when its text was contributed, not by the page's last revision; the markup cleanup:
+remove category lines and file-link options first, then label the cleaned text).
 
 **Verdict** — `kept`, text cleaned and relabelled: 13,640 documents, 44,719,971 characters,
 16,959,180 tokens (was 13,643 / 45,176,480 / 17,125,240). Labels per record: `CC-BY-SA-3.0`
@@ -2251,7 +2246,7 @@ option A: remove category lines and file-link options first, then label the clea
 and `pre_cutoff_revid`; the build record holds the stub dump, the count per basis and the
 cleanup; the card explains each basis.
 
-**Residue cleanup (v2#116)** — `strip` is now v0.2.5's `strip_code` followed by `drop_residue`,
+**Residue cleanup** — `strip` is now v0.2.5's `strip_code` followed by `drop_residue`,
 which removes `Kategoria:...` lines, bare file titles and file-link options (`thumb|250px|...`)
 and keeps captions. `fetch_wikimedia.py retext` applied it to the committed text: 13,171 rows
 changed, 456,509 characters and 166,060 tokens removed. Three pages fell under 200 characters
@@ -2284,7 +2279,7 @@ rest. Rows labelled both still hold 58% of the tokens.
 **License evidence** — as in the sections above. Contributors' text carries the licence in force
 when it was contributed, so a page with text from both sides of the cutoff carries both.
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - Any added word makes a page 3.0 and 4.0; there is no threshold.
 - Text moved or merged from an older page into a page created after the cutoff is labelled 4.0
   alone.
@@ -2294,12 +2289,12 @@ when it was contributed, so a page with text from both sides of the cutoff carri
 - The comparison is on converted text, so words that `strip` drops (templates) are not compared.
 - The residue `drop_residue` leaves (see above) is compared like any other words.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## wikibooks — markup residue removed, licence labels by contribution date, one page withheld (update to the sections above)
 
-**Investigated** — 2026-10-17/18 (slayerlabs/polish-dynaword-v2#110 option B and
-slayerlabs/polish-dynaword-v2#116 option A, as for wikivoyage above).
+**Investigated** — 2026-10-17/18 (the labels by contribution and
+the markup cleanup, as for wikivoyage above).
 
 **Verdict** — `kept`, one more page withheld, text cleaned and relabelled: 9,048 documents,
 38,064,404 characters, 15,261,326 tokens (was 9,066 / 38,195,523 / 15,313,229). Labels per
@@ -2317,7 +2312,7 @@ in `DROPPED_PAGES["wikibooks"]` (47 pages now). `fetch_wikimedia.py retext` now
 listed after the build, as `prepare` would skip it, before it measures the cleanup; asked to
 withhold a page from a release that is already clean, it refuses.
 
-**Residue cleanup (v2#116)** — as for wikivoyage. Of 9,065 documents (38,194,654 characters,
+**Residue cleanup** — as for wikivoyage. Of 9,065 documents (38,194,654 characters,
 15,312,861 tokens) left after the withholding, 2,087 changed: 130,250 characters and 51,535
 tokens removed. 17 pages fell under 200 characters and were dropped (`wikibooks_1532`, `10162`,
 `14576`, `18215`, `21429`, `22859`, `23016`, `23110`, `23117`, `24149`, `24839`, `55101`,
@@ -2372,7 +2367,7 @@ rerun of retext, relabel and datasheet is byte-identical.
 69,072). Tokens and labels are updated and an `updated` note added; 39 of the 45 are 3.0 and 6
 both. `wikibooks_3083`, whose page states CC BY-SA 3.0, was 4.0 by its last edit and is now 3.0.
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - Any added word makes a page 3.0 and 4.0; there is no threshold.
 - Text moved or merged from an older page into a page created after the cutoff is labelled 4.0
   alone.
@@ -2384,13 +2379,13 @@ both. `wikibooks_3083`, whose page states CC BY-SA 3.0, was 4.0 by its last edit
 - The 5 pages deleted as spam or duplicates and `wikibooks_20267`, now in a user's sandbox, stay
   in the release.
 - As in the sections above: the copyright sample read 79 of the pages and 90 more by one
-  contributor; E1 (v2#56); the 17 own-text and upstream GFDL documents.
+  contributor; the 17 own-text and upstream GFDL documents.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## wikipedia — licence labels by contribution date, seven pages withheld (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#110 option B, as for wikivoyage and
+**Investigated** — 2026-10-18 (the labels by contribution, as for wikivoyage and
 wikibooks above; the deletion logs of released pages missing from the 2026-10-01 stub dump).
 
 **Verdict** — `kept`, seven more pages withheld, relabelled: 1,468,647 documents, 3,140,823,417
@@ -2415,11 +2410,11 @@ checked. They are in `COPIED_PAGES` (9 pages now) with the day and the log reaso
 `prepare` would, and adds it to the copyright sample's `rows_dropped_since`, so the sample's draw
 still reproduces; it refuses to run after `relabel`.
 
-**No residue cleanup (v2#116)** — `strip_article` already drops category and file links, so
+**No residue cleanup** — `strip_article` already drops category and file links, so
 `drop_residue` is not applied. Run on this release it would change 426 documents and remove
 477,849 characters, nearly all content: flattened medal tables whose line starts
 `Kategoria:Złoto:Srebro:Brąz:` and goes on with the results (`wikipedia_1515505`), and field
-labels such as `Kategoria: olej nieschnący` (`wikipedia_127177`). v2#116 read the 410 rows with a
+labels such as `Kategoria: olej nieschnący` (`wikipedia_127177`). The 410 rows with a
 `Kategoria:` line as medal-table headers; most are, but 14 documents hold a category link on a
 line of its own (16 lines, e.g. `wikipedia_103834` `Kategoria:Wielcy marszałkowie zakonu
 krzyżackiego`), left by malformed links. Also left: 13 documents with an unclosed file link
@@ -2458,7 +2453,7 @@ shipped revision). They are kept: their deletions are a quality call, not a righ
 deleted for copyright on 2017-05-23 and recreated by 2017-07-14; the released text was not
 compared with the deleted one. Counts per class are in the evidence's `deleted_since_dump`.
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - Any added word makes a page 3.0 and 4.0; there is no threshold.
 - Text moved or merged from an older page into a page created after the cutoff is labelled 4.0
   alone.
@@ -2471,12 +2466,12 @@ compared with the deleted one. Counts per class are in the evidence's `deleted_s
 - As in the sections above: pastes beyond the copyright sample, GFDL-era text, appendix headings,
   the NERGAL pass.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## wikinews — markup residue removed, licence labels by contribution date, switch window (update to the section above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#110 option B and
-slayerlabs/polish-dynaword-v2#116 option A, as for wikivoyage above; the hours of the licence
+**Investigated** — 2026-10-18 (the labels by contribution and
+the markup cleanup, as for wikivoyage above; the hours of the licence
 switch).
 
 **Verdict** — `kept`, text cleaned and relabelled: 23,573 documents, 29,400,544 characters,
@@ -2487,7 +2482,7 @@ documents (9,347,380 tokens, 85.2%), `CC-BY-2.5 AND CC-BY-4.0` 113 (81,128, 0.7%
 still released, 2,654 (1,120,223 tokens) are now 2.5, 113 both and 3,637 stay 4.0. Each sidecar
 line gains `license_basis` and `pre_cutoff_revid`, and `license_flags` where a flag below applies.
 
-**Residue cleanup (v2#116)** — as for wikivoyage. 24,204 of the 24,386 rows changed: 3,124,102
+**Residue cleanup** — as for wikivoyage. 24,204 of the 24,386 rows changed: 3,124,102
 characters and 1,171,145 tokens removed, 2,769,263 of the characters (89%) in category lines. 812
 pages fell under 200 characters and were dropped (26 of 54-99 characters, 284 of 100-149, 502 of
 150-199), short news items such as `wikinews_1836` "2005-04-05: Premier Wielkiej Brytanii ogłosił termin wyborów" (124). One more
@@ -2535,10 +2530,10 @@ date for the clause. The statement is not relied on: they are labelled by contri
 2.5, 13 both).
 
 **License evidence** — as in the section above, plus the notice revisions. `CC-BY-4.0` now covers
-only text saved after the edit notice named 4.0; 2.5 text keeps 2.5, so the open question above (a
+only text saved after the edit notice named 4.0; 2.5 text keeps 2.5, so the point above (a
 4.0 label on 2.5 text resting on the terms page alone) no longer arises.
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - The window rule (both for text saved between 08:07:33 and 23:19:21 UTC) is our assessment; if
   it changes, the 25 window rows are relabelled.
 - Any added word makes a page 2.5 and 4.0; there is no threshold.
@@ -2552,12 +2547,12 @@ only text saved after the edit notice named 4.0; 2.5 text keeps 2.5, so the open
 - The sidecar's attribution fields (permalink, history link, licence URLs, a note on the
   conversion) are a separate change.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## wiktionary_examples — licence labels by contribution date (update to the section above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#110, option B: label each example by
-when it was contributed, not by the page's last revision). v2#116 does not touch this source: rows
+**Investigated** — 2026-10-18 (the labels by contribution: label each example by
+when it was contributed, not by the page's last revision). The markup cleanup does not touch this source: rows
 are the `{{przykłady}}` lines, and the shipped text has no category lines or file-link options.
 
 **Verdict** — `kept`, relabelled; text, ids and counts unchanged (1,818 documents, 534,521
@@ -2599,14 +2594,14 @@ cited lines. A copy without a marker cannot be detected, like any unmarked quota
 **License evidence** — as in the section above. The 3.0 → 4.0 relabel of pre-2023 text, read from
 3.0 §4(b), is gone: each example keeps the licence it was contributed under.
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - The cutoff is midnight UTC; an example added on 2023-06-07 counts as 4.0. No shipped page was
   created that day; the latest pre-cutoff revision compared is from 2023-06-06 21:55 UTC.
 - An example moved from another page after the cutoff is labelled 4.0.
 - Words are split on whitespace, so a comma removed from a word counts as a changed word (both),
   e.g. `wiktionary_examples_211962`.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## `studia_bas`: licence version resolved to CC BY 3.0 PL
 
@@ -2644,7 +2639,7 @@ GET returned it on 2026-10-06. Retry that before routing a check to a human.
 
 ## biblioteka_nauki — licence versions restored from the API (2026-10-06)
 
-**Verdict** — `added`, kept for v0.2.6 minus 83 rows (polish-dynaword-v2#57; maintainer approval
+**Verdict** — `added`, kept for v0.2.6 minus 83 rows (maintainer approval
 2026-10-06). 42,068 → 41,985 documents, 673,112,676 → 669,909,430 tokens (−3,203,246),
 1,747,387,478 → 1,739,004,913 characters. The `license` column now carries the version the API
 gives per record: CC-BY-4.0 17,768, CC-BY-3.0-PL 8,098, CC-BY-3.0 2,413, CC-BY-2.0 1, CC-BY-SA-4.0
@@ -2684,7 +2679,7 @@ was assembled is not re-verified.
 
 ## sejm_api — "Przebieg posiedzenia" trailers cut, appended statements split out (2026-10-06)
 
-**Finding** (slayerlabs/polish-dynaword-v2#57) — the PiotrSty snapshot appends to 34,980
+**Finding** (v0.2.6 rights audit) — the PiotrSty snapshot appends to 34,980
 of 38,812 speeches the header of a later statement of the same sitting day
 ("Przebieg posiedzenia {name}: {term}. kadencja, {n}. posiedzenie, {d}. dzień - {label}
 … ({dd-mm-yyyy})"), and after 2,466 of those headers that statement's text. The trailers
@@ -2723,7 +2718,7 @@ in 90% of rows hid in plain sight because each row still began with the right sp
 
 ## sejm_api_2011_2022 — trailers cut before the cross-source dedup (2026-10-06)
 
-**Finding** (slayerlabs/polish-dynaword-v2#57) — the same "Przebieg posiedzenia" trailers
+**Finding** (v0.2.6 rights audit) — the same "Przebieg posiedzenia" trailers
 as in `sejm_api` (section above) are in 105,920 of the 118,902 pre-dedup speeches:
 97,723,744 characters. Hub PR 76 only stripped them from the shingled text when it scored
 containment; the published rows kept them.
@@ -2747,7 +2742,7 @@ itself dropped. Kept 2011-2019 rows go from 1,697 to 1,281. Parquet sha256
 
 ## sejm_interpellations — licence evidence and SIS caveat (2026-10-06)
 
-**Finding** (slayerlabs/polish-dynaword-v2#57) — the source had no license-evidence.json. The
+**Finding** (v0.2.6 rights audit) — the source had no license-evidence.json. The
 Regulamin Sejmu places interpellations, replies and written questions in the System Informacyjny
 Sejmu (arts. 192 ust. 7, 193 ust. 2, 195 ust. 2), whose terms page asks for a source note and bars
 SIS material from commercial offers; nothing on api.sejm.gov.pl says its output is part of SIS.
@@ -2760,10 +2755,10 @@ hashes, the SIS wording (Wayback 2026-06-03), the placement articles, the readin
 commercial clause is not an Open Data Act art. 15 ust. 1 condition, and the third-party sample.
 The datasheet states the caveats and gives the requested source note; the registry `added` is
 2026-09-05 and the entry has a `legal_note`. The Parquet rows still carry the global `ADDED`
-(2026-07-19) until `build_dynaword.py` stamps the registry date (v2#80).
+(2026-07-19) until `build_dynaword.py` stamps the registry date (the v0.2.6 card and registry edits).
 
 **Not established** — the reading of the SIS terms (an inference, not confirmed); the SIS page on the fetch date (bot
-challenge); the Kancelaria Sejmu BIP reuse page; personal names (NERGAL, v2#60).
+challenge); the Kancelaria Sejmu BIP reuse page; personal names (NERGAL).
 
 ## sejm_committee_transcripts — only statements by official speakers (2026-10-07)
 
@@ -2788,7 +2783,7 @@ and by `src/clean_sejm_committee_transcripts.py`, which re-filters the pinned pr
 stored turn is cut at each label line (piece 0 keeps the id and author, later pieces get the label
 as author and the id `<id>_<k>`), page blocks (page number + stenographer initials, `Pełny Zapis
 Przebiegu Posiedzenia:` title, `Komisji ... (Nr N)`) are removed as a unit with a word hyphenated
-across one rejoined (v2#95), and every piece is classified. 653,411 turns → 740,018 pieces →
+across one rejoined, and every piece is classified. 653,411 turns → 740,018 pieces →
 613,722 records (727,570 raw + 86,607 split out; joint-sitting duplicates 74,117, exact duplicates
 77, under 15 characters 16,080, not official 110,181), 443,928,915 → 298,555,500 characters,
 166,310,754 → 112,003,155 tokens; the 110,181 dropped pieces are 127,167,730 characters (28.6%
@@ -2849,7 +2844,7 @@ role-anchored labels. Cut at everything that looks like a label, then classify.
 
 ## sejm_api_2011_2022 — dedup re-pinned to the PPC TEI `parliamentary` (2026-10-08)
 
-**Finding** (slayerlabs/polish-dynaword-v2#54, option 5A) — the cross-source dedup scored
+**Finding** — the cross-source dedup scored
 against the v0.2.5 `parliamentary`, which Hub PR 108 replaces with a rebuild from the PPC TEI
 dump under a speaker allowlist. The rebuild keeps Sejm plenary text up to term 8 but drops
 speakers the allowlist does not match, so a dedup against it keeps rows the old shard dropped.
@@ -2863,11 +2858,11 @@ heuristic. The duplicate outcome key `sejm_api_only` (same count as `only_sejm_a
 speeches 0.50-0.71 contained in a 2019-2023 Senate document) and 1,514 rows (7,126,113
 characters) are kept again: 573 (4,908,128) whose speaker the allowlist drops (citizens'
 initiative representatives, the President's Chancellery, the Prosecutor General's office, agency
-heads; slayerlabs/polish-dynaword-v2#99), 470 (1,084,769) not in the dump (435 of them
+heads), 470 (1,084,769) not in the dump (435 of them
 "(tekst niewygłoszony)"), 357 (911,196) mostly in kept text but under 0.5 (315 lose mainly
 floor interjections), 114 (222,020) to other rebuild drops. The 23 speeches of MP Piotr van der
 Coghen stay dropped: the rebuild's name check keeps a member's name with a particle (decision
-1A of 2026-10-08). Parquet sha256 `f6ec51dfd87b47b93905adcdd8f8dfb3f8449d6319df4153526c900bf258062f`.
+of 2026-10-08). Parquet sha256 `f6ec51dfd87b47b93905adcdd8f8dfb3f8449d6319df4153526c900bf258062f`.
 
 **Re-pin (2026-10-08)** — Hub PR 111 repaired Windows-1250 misdecoding in `parliamentary` and
 dropped 18,612 exact copies, so its parquet is `07903953…`. `REFERENCES` is re-pinned to it and
@@ -2877,10 +2872,9 @@ containment of a speech in one document, and the encoding repair changed no scor
 decisions (every speech scoring 0.3 or more, byte-identical).
 
 **Not established** — the hand checks in the card still date from Hub PR 76. The 573 rows whose
-speaker no allowlist rule matches stay here until the allowlist is applied to this source
-(slayerlabs/polish-dynaword-v2#99).
+speaker no allowlist rule matches stay here until the allowlist is applied to this source.
 
-**Update (2026-10-19)** — the allowlist is applied (slayerlabs/polish-dynaword-v2#99, section "sejm_api,
+**Update (2026-10-19)** — the allowlist is applied (section "sejm_api,
 sejm_api_2011_2022 — speaker allowlist" below): the dedup against the re-pinned `sejm_api` leaves 34,601 rows (the 34,598
 above and 3 more), 671 are dropped by speaker and 33,930 are published. The counts above are those of the dedup alone,
 against the earlier pin.
@@ -2893,9 +2887,9 @@ published. The 33,930 rows published before are unchanged, in order and in every
 the 11 keys (all kept as `mp_no_party`, 7 on member and 4 on day evidence) and is re-pinned to `06723c31…`; the
 `sejm_api` table is byte-identical.
 
-## parliamentary — filter effectiveness (v2#114)
+## parliamentary — filter effectiveness
 
-**Investigated** — 2026-10-18 and 19 (slayerlabs/polish-dynaword-v2#114; the allowlist is #54).
+**Investigated** — 2026-10-18 and 19.
 
 **Verdict** — the allowlist is `kept` as built; no rule changed in this entry. Of the official text in the sittings
 it filters, it keeps 95.8% of the characters (95% interval 93.6-97.3%). In a hand-labelled sample it let one guest
@@ -2986,7 +2980,7 @@ git (the sample holds speech text with names).
   C and E are eyeballed on 40, 20 and 20, not counted; B is counted from 70 labels of 2,378.
 - Recall treats every kept paragraph as official text; the kept sample has 1 guest in 240.
 - Sittings only. The 286,135 interpellation, question and reply records (kept whole; the allowlist does not apply to
-  them) are not in this sample; the check against the official texts (v2#114) is where they come in.
+  them) are not in this sample; the check against the official texts is where they come in.
 
 **Release steps** — none for this entry. If the allowlist is widened to the forms above, re-run
 `src/fetch_parliamentary.py`, the stats and the card totals, and re-pin `sejm_api_2011_2022` (its `REFERENCES` name
@@ -2996,9 +2990,9 @@ the shard's parquet hash).
 the Jeffreys interval says up to 3.5%, and only the scan for the one mechanism that the single guest showed gave a
 number to plan with (0.3M characters).
 
-## sejm_api, sejm_api_2011_2022 — speaker allowlist (v2#99)
+## sejm_api, sejm_api_2011_2022 — speaker allowlist
 
-**Investigated** — 2026-10-19 (slayerlabs/polish-dynaword-v2#99; the allowlist is #54, its check #114). Both sources
+**Investigated** — 2026-10-19. Both sources
 kept every speaker of the plenary sittings; `parliamentary` and `sejm_committee_transcripts` keep the allowlist only.
 
 **Verdict** — the allowlist (`classify_speaker`, `MEMBER_RULES`), unchanged, is applied to both sources by a pinned
@@ -3076,10 +3070,10 @@ prosecutors, the Rada Mediów Narodowych); X external guest.
 **Found on the way**
 - The allowlist refuses the President's Chancellery (secretaries of state, deputy heads), the Prosecutor General's
   office, IPN, agency heads (UOKiK, GUS, UODO, UdSC, GIP, GIS, the Civil Service, the NBP, the RCL) and foreign
-  parliaments; #114 lists them as left out by design. Refused roles are 65 of the 123 dropped `sejm_api` rows (53%) and
+  parliaments. Refused roles are 65 of the 123 dropped `sejm_api` rows (53%) and
   650 of the 671 `sejm_api_2011_2022` rows (97%); the rest have no role evidence. Widening the allowlist would change
   all three sources.
-- The rule table lacks the forms above; #114 lists government plenipotentiaries among the forms `parliamentary` misses.
+- The rule table lacks the forms above; the forms `parliamentary` misses include government plenipotentiaries.
   The rules sit in `fetch_sejm_committee_transcripts.py` and `fetch_parliamentary.py`, so a change reaches the
   committee transcripts and `parliamentary` too.
 - `sejm_api` is a pinned reference of the `sejm_api_2011_2022` dedup; Hub PR 158 re-pinned it to the filtered release
@@ -3110,7 +3104,7 @@ ministers. Label from the evidence of the same day, and read every drop when the
 
 ## wikinews — lone option lines removed (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the main leftover of the section
+**Investigated** — 2026-10-18 (the markup cleanup: the main leftover of the section
 above, a line holding only an option word, removed in v0.2.6 after all).
 
 **Verdict** — `kept`, text cleaned again and relabelled: 23,573 documents, 29,399,003 characters,
@@ -3156,7 +3150,7 @@ rule (their own PRs, next), their contract tests fail on `drop_residue(text) ==
 
 ## wikinews — option words before captions removed (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the second rule change of the
+**Investigated** — 2026-10-18 (the markup cleanup: the second rule change of the
 plwiki_talk section below, Hub PR 135, applied to this release; Hub PRs 137–139, one per source).
 
 **Verdict** — `kept`, text cleaned again: 23,573 documents, 29,397,981 characters, 10,968,912
@@ -3174,7 +3168,7 @@ a rerun of retext, relabel and datasheet changes nothing.
 
 ## wikivoyage — lone option lines removed, four table rows rebuilt (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the rule of the wikinews section
+**Investigated** — 2026-10-18 (the markup cleanup: the rule of the wikinews section
 above, applied to this release).
 
 **Verdict** — `kept`, text cleaned again and relabelled: 13,640 documents, 44,720,085 characters,
@@ -3201,7 +3195,7 @@ next), its contract test fails on `drop_residue(text) == text`.
 
 ## wikivoyage — option words before captions removed (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the second rule change of the
+**Investigated** — 2026-10-18 (the markup cleanup: the second rule change of the
 plwiki_talk section below, Hub PR 135, applied to this release; Hub PRs 137–139, one per source).
 
 **Verdict** — `kept`, text cleaned again: 13,640 documents, 44,719,617 characters, 16,958,893
@@ -3221,7 +3215,7 @@ relabel and datasheet changes nothing.
 
 ## wikibooks — lone option lines removed, relabelled (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the rule of the wikinews section
+**Investigated** — 2026-10-18 (the markup cleanup: the rule of the wikinews section
 above, applied to this release; the last of the three).
 
 **Verdict** — `kept`, text cleaned again and relabelled: 9,048 documents, 38,063,266 characters,
@@ -3252,7 +3246,7 @@ of the source), the closest audit pattern's 25 documents 68,917 and `own_text` 4
 
 ## wikibooks — option words before captions removed (update to the sections above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116: the second rule change of the
+**Investigated** — 2026-10-18 (the markup cleanup: the second rule change of the
 plwiki_talk section below, Hub PR 135, applied to this release; Hub PRs 137–139, one per source).
 
 **Verdict** — `kept`, text cleaned again: 9,048 documents, 38,062,836 characters, 15,260,848 tokens
@@ -3278,8 +3272,8 @@ documents. Matched tokens are now 121,964 (0.80% of the source), the closest pat
 
 ## plwiki_talk — file-link residue removed, licence labels by contribution date (update to the section above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116 option A, with category lines
-kept; slayerlabs/polish-dynaword-v2#110 option B, as for wikivoyage above).
+**Investigated** — 2026-10-18 (the markup cleanup, with category lines
+kept; the labels by contribution, as for wikivoyage above).
 
 **Verdict** — `kept`, text cleaned and relabelled: 43,054 documents, 144,590,122 characters,
 55,564,235 tokens (was 43,054 / 144,604,697 / 55,568,733). Labels per record: `CC-BY-SA-3.0`
@@ -3290,7 +3284,7 @@ are now 3.0 (17,902,746 tokens), 8,528 both and 2,437 stay 4.0. Each sidecar lin
 `license_basis` and `pre_cutoff_revid`; `qa.json` holds the stub dump, the count per basis
 (`licence_labels`) and the cleanup (`residue_cleanup`); the card explains each basis.
 
-**Residue cleanup (v2#116)** — both talk builders now end with `drop_residue(text,
+**Residue cleanup** — both talk builders now end with `drop_residue(text,
 categories=False)` after the PII redaction: file-link options and bare file titles go, `Kategoria:`
 lines stay, since on talk pages they are discussion about categories. A record left under 500
 characters is dropped as `too_little_text_after_residue_cleanup`. `src/clean_plwiki_talk.py retext`
@@ -3340,7 +3334,7 @@ word (both); for 3 the API returns no pre-cutoff text (`pre_text_hidden`, both).
 in force when it was contributed, so a page with text from both sides of the cutoff carries both.
 1,986 records were last edited before 2009-06-15, the GFDL era (see the card).
 
-**Blockers** — none to merge. Open:
+**Blockers** — none to merge. Known limits:
 - Any added word makes a page 3.0 and 4.0; there is no threshold.
 - Text moved or merged from an older page into a page created after the cutoff is labelled 4.0
   alone.
@@ -3350,11 +3344,11 @@ in force when it was contributed, so a page with text from both sides of the cut
   compared, and the residue `drop_residue` leaves is compared like any other words.
 - A rebuild from the dump labels by last revision; `relabel` has to follow it.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## plwiki_user_talk — file-link residue removed, licence labels by contribution date (update to the section above)
 
-**Investigated** — 2026-10-18 (slayerlabs/polish-dynaword-v2#116 and #110, as for plwiki_talk
+**Investigated** — 2026-10-18 (the markup cleanup and the labels by contribution, as for plwiki_talk
 above; same code, same stub dump).
 
 **Verdict** — `kept`, text cleaned and relabelled: 380,295 documents, 1,019,726,109 characters,
@@ -3365,7 +3359,7 @@ main were 3.0 243,773 (160,360,307 tokens) and 4.0 136,754 (221,084,804). The 24
 released stay 3.0; of the 4.0 rows, 32,804 are now 3.0 (39,073,338 tokens), 22,558 both and 81,196
 stay 4.0.
 
-**Residue cleanup (v2#116)** — `clean_plwiki_talk.py retext`: 27,221 rows changed; with the dropped
+**Residue cleanup** — `clean_plwiki_talk.py retext`: 27,221 rows changed; with the dropped
 rows, 1,002,350 characters and 429,775 tokens went. 232 records fell under 500 characters and were
 dropped (`too_little_text_after_residue_cleanup`; 36 labelled 3.0 on main, 196 4.0), all of them
 500–557 characters before; 171 hold little more than the pending-changes notice, which lost its
@@ -3389,16 +3383,16 @@ records were last edited before 2009-06-15.
 **License evidence** — as for plwiki_talk above.
 
 **Blockers** — none to merge after the plwiki_talk PR, which carries the code: until both are
-merged, this release fails its contract tests (labels and `drop_residue(text) == text`). Open: as for
+merged, this release fails its contract tests (labels and `drop_residue(text) == text`). Known limits: as for
 plwiki_talk above.
 
-README, CHANGELOG and totals are left to v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## uzp_orzeczenia_pl — G1 personal-data pass
 
 **Investigated** — 2026-10-18 and 19, re-scored for NERGAL 2.0.5 and 2.0.6 on 2026-10-21; firm rules extended and
-the pass applied to the shard on 2026-10-21 (slayerlabs/polish-dynaword-v2#112; the legitimate-interest note is
-shared with #55, NERGAL is #60).
+the pass applied to the shard on 2026-10-21 (the legitimate-interest note is
+shared with `saos`).
 
 **Verdict** — `kept`, no ruling dropped. `src/build_uzp_orzeczenia_pl.py g1` has run on the shard in this PR:
 21,491 of 34,715 rulings change; characters 1,110,310,083 → 1,108,021,951, tokens 422,051,565 → 420,753,012. Against
@@ -3572,7 +3566,7 @@ that count was 11 in S6 and S6 special recall read 11/16; after it, 1 and 15/16,
 
 ## uzp_orzeczenia_pl — section numbers restored from the IP filter (update to the section above)
 
-**Investigated** — 2026-10-19 (slayerlabs/polish-dynaword-v2#112).
+**Investigated** — 2026-10-19.
 
 **Verdict** — `kept`, the same 34,715 rulings with rebuilt text: `[REDACTED:IP]` tags 21,957 → 50, 2,793 documents
 changed, 1,110,310,083 characters and 422,051,565 tokens (was 1,110,428,704 / 422,043,881). This is the
@@ -3611,11 +3605,11 @@ one; an address written with no network word, port or CIDR suffix near it would
 
 **Release steps** — `g1` has run on this shard, as in the section above, and added the slice version for the
 G1 parquet, with its stats, that the contract test requires.
-README, CHANGELOG and totals are left to v2#74 and v2#80.
+README, CHANGELOG and totals are left to the v0.2.6 card and registry edits.
 
 ## uzp_orzeczenia_pl — G1 rules, round 2, not applied (update to the sections above)
 
-**Investigated** — 2026-10-21 (slayerlabs/polish-dynaword-v2#112).
+**Investigated** — 2026-10-21.
 
 **Verdict** — `kept`, the shard unchanged. A second round of firm and address rules is in
 `src/build_uzp_orzeczenia_pl.py` but has not run on the shard: on the held-out sample it misses the address bar
@@ -3692,10 +3686,10 @@ set before it is scored.
 
 ## eurlex — row review: third-party credits, UN ECE regulations and Parliament stubs dropped (update to the section above)
 
-**Investigated** — 2026-10-19 (slayerlabs/polish-dynaword-v2#108).
+**Investigated** — 2026-10-19.
 
-**Verdict** — kept; 480 more documents dropped by a pinned row review (maintainer decisions 1A, 2A and 4A on
-v2#108, 2026-10-19): 242,968 → 242,488 documents, 5,902,889,135 → 5,859,513,831 characters, 2,350,709,056 →
+**Verdict** — kept; 480 more documents dropped by a pinned row review (maintainer decisions of
+2026-10-19): 242,968 → 242,488 documents, 5,902,889,135 → 5,859,513,831 characters, 2,350,709,056 →
 2,333,038,468 tokens. Every kept row is otherwise identical to v0.2.5.
 
 **Method** — `src/clean_eurlex.py` `CANDIDATE_RULES` (the (c) sign, photo headings, credits and permissions,
@@ -3746,9 +3740,9 @@ is still the v0.2.5 scan report.
 
 ## kultura_bez_barier_pl — row review: two joint-project titles dropped, licence scope open (update to the section above)
 
-**Investigated** — 2026-10-20 (slayerlabs/polish-dynaword-v2#111).
+**Investigated** — 2026-10-20.
 
-**Verdict** — kept; 2 of the 9 documents dropped by a pinned row review (maintainer decision 5A on v2#111,
+**Verdict** — kept; 2 of the 9 documents dropped by a pinned row review (maintainer decision of
 2026-10-20): 9 → 7 documents, 608,794 → 359,517 characters, 217,888 → 126,743 tokens, 316 → 214 pages (91,145
 tokens, 41.8% of the shard). Every kept row is identical to the shard before the review; the licence basis fields are
 new in the sidecar.
@@ -3772,7 +3766,7 @@ organisations named as authors or co-producers.
   Institut für Bildung und Kultur e.V. (kubia), and lists experts of all four. `paszport_do_sztuki`: "a joint project of
   the foundation and cultural institutions"; Łazienki Królewskie, Zachęta, the Wilanów Palace Museum and the National
   Museum in Warsaw took part and the foundation led it.
-- **Licence scope (open)** — the catalogue's body sentence says the foundation releases "je" under CC BY-SA 3.0, right
+- **Licence scope** — the catalogue's body sentence says the foundation releases "je" under CC BY-SA 3.0, right
   after naming the pictograms; the page's meta description ties the licence to the pictograms alone. The earlier reading
   above (the sentence covers the publications) is not established. The maintainer has asked the foundation.
 - **Licence basis** — `publisher` for 6 records (the PDF is linked from the catalogue page, read 2026-10-19), `file_note`
@@ -3784,7 +3778,7 @@ still marks the 9 as selected; the 2 review drops are in the review table.
 **Not established** — the scope of the catalogue's licence statement; whether the individual authors of the 7 kept
 titles are staff or commissioned (the foundation's rights from them are not published); the NERGAL pass (unchanged).
 
-## wolne_lektury — author and title filled, attribution sidecar, edition checks (v2#109)
+## wolne_lektury — author and title filled, attribution sidecar, edition checks
 
 **Investigated** — whether every `wolne_lektury` row carries the author, translator and title Wolne Lektury (WL)
 gives, whether the full licence URL, the credits and the "Książka pochodzi z serwisu Wolne Lektury" line are
@@ -3910,7 +3904,7 @@ details; by the maintainer's decisions of the same day:
 - Kept as WL wrote them, since none changes whether a row is free: 25 rows whose WL rights line gives a death year BN
   gives otherwise (for example "Wacław Rolicz-Lieder zm. 1940", BN 1912, 8 rows) and 6 rows whose WL `date.pd` is
   earlier than the translator's or author's death allows.
-- Open: 74 cannot-tell row-flags on 69 kept rows, left as WL gives them and listed in v2#109.
+- Left as WL gives them: 74 cannot-tell row-flags on 69 kept rows.
 
 **Not established** — the editors of the print editions and their death years; the first-edition form of any author
 name; what SpeakLeash changed and who removed the footer; whether
@@ -3918,10 +3912,10 @@ name; what SpeakLeash changed and who removed the footer; whether
 
 ## biblioteka_nauki — row review: 4,142 rows dropped, per-record licence basis added (update to the sections above)
 
-**Investigated** — 2026-10-20 (slayerlabs/polish-dynaword-v2#111).
+**Investigated** — 2026-10-20.
 
-**Verdict** — kept; 4,142 of 41,985 documents dropped by a pinned row review (maintainer decisions 1A, 2A and 3 on
-v2#111, 2026-10-20): 41,985 → 37,843 documents (-9.9%), 1,739,004,913 → 1,534,529,869 characters, 669,909,430 →
+**Verdict** — kept; 4,142 of 41,985 documents dropped by a pinned row review (maintainer decisions of
+2026-10-20): 41,985 → 37,843 documents (-9.9%), 1,739,004,913 → 1,534,529,869 characters, 669,909,430 →
 590,924,796 tokens (-78,984,634, -11.8%). Every kept row is identical to the shard before the review; the licence basis
 fields are new in the sidecar.
 
@@ -3972,12 +3966,12 @@ The rules' measurements and the hand reads are in `data/biblioteka_nauki/bibliot
   version. The year rule is a lower bound: the years are those of the generic licence texts (the Polish port may be
   later, which would add rows), and the platform shows the licence an item has today.
 - **Printed NC/ND (4), another work (16), prior publication, permission and translation (63 rows), API flags (5)** —
-  decision 1A. 16 files are a different work from the one described: neither the title nor any author appears where the text
+  maintainer decision of 2026-10-20. 16 files are a different work from the one described: neither the title nor any author appears where the text
   starts, ends or is attributed (all 51 candidates read; 30 each of two looser sets were sampled and none was another
   work). Prior publication and translation rows were read in context for the tight families (reprint words in the front
   matter, permission, earlier publication, title markers, translator credit, translation statement); the broad
   translator-word family (1,326 rows, mostly footnotes) was not.
-- **Restriction notice, no CC note (97)** — decision 2A: the front or back matter prints a restriction on copying or
+- **Restriction notice, no CC note (97)** — maintainer decision of 2026-10-20: the front or back matter prints a restriction on copying or
   reproduction and no CC licence of its own. All 120 hits were read; 6 of the 110 hits outside `file_note` rows are not
   a notice on the work. This is the weakest evidence of the drops: it is a reader's judgement that the printed
   restriction contradicts the platform's CC label; no journal page was read to see which of the two the journal means.
@@ -3998,7 +3992,7 @@ now for the rows the year rule does not catch; the false-negative rate of the `f
 words or a form the parser does not read were not counted); whether the 97 restriction-notice drops would stand if the
 journals' own licence pages were read; recall of the reprint and translation rules beyond the tight families.
 
-## eltec_pol — author and title filled, attribution sidecar, edition checks (v2#109)
+## eltec_pol — author and title filled, attribution sidecar, edition checks
 
 **Investigated** — whether every `eltec_pol` row carries the author and title ELTeC gives, whether the ELTeC record,
 the CC BY 4.0 URL and the Wolne Lektury (WL) overlap are recorded per document, and what the text and the edition
@@ -4114,10 +4108,10 @@ texts digitised from Wolne Lektury.
 
 ## open_icm_pl — row review: 74 of 85 records dropped, per-record licence basis added (update to the section above)
 
-**Investigated** — 2026-10-05 (repository check), 2026-10-20 (publisher pages, row review; slayerlabs/polish-dynaword-v2#111).
+**Investigated** — 2026-10-05 (repository check), 2026-10-20 (publisher pages, row review).
 
-**Verdict** — kept as a pilot slice of 11 records; 74 of the 85 dropped by a pinned row review (maintainer decisions 4A and
-1A on v2#111, 2026-10-20): 85 → 11 documents, 6,219,019 → 1,203,239 characters, 2,358,174 → 441,475 tokens (1,916,699 tokens,
+**Verdict** — kept as a pilot slice of 11 records; 74 of the 85 dropped by a pinned row review (maintainer decisions of
+2026-10-20): 85 → 11 documents, 6,219,019 → 1,203,239 characters, 2,358,174 → 441,475 tokens (1,916,699 tokens,
 81.3% of the shard, dropped). Three kept rows change in the `license` column only; the other 8 kept rows are identical to
 the shard before the review (the digest of every input row is in `open_icm_pl.licence-basis.json`, recomputed by a test).
 The licence basis fields are new in the sidecar.
@@ -4166,20 +4160,20 @@ the first year of the licence version of the label it would carry after the rela
 
 **Kept as decided** — the 11 records above. `qa.json`, `decisions.jsonl` and `overlap_audit.json` describe the
 100-record pilot and still list the 85 (and the 95 and 100); the 74 review drops are in the review table. The README row
-and CHANGELOG for this source are older than v2#111 and were not changed.
+and CHANGELOG for this source are older than the per-record licence basis and were not changed.
 
 **Not established** — the publisher statements are as read on 2026-10-20: for older articles a publisher may have applied
 a statement to its back issues later. The year rule compares years only and its table is a lower bound (a Polish port or
 a journal may have adopted a version later): `6fe4e956` was created in 2013 and its publisher page is dated 2013-06-24,
 before CC 4.0 was released (November 2013), yet it is kept with CC-BY-4.0; whether that licence applied when the text was
-published is open. Three statements are journal-level, not article-level. `8c8e1344` has no version.
+published is not established. Three statements are journal-level, not article-level. `8c8e1344` has no version.
 The 3 PDFs behind the relabelled records and all 77 `license.txt` bitstreams were not downloaded (the endpoint serves a
 CAPTCHA page to anonymous clients). The 64 depositor-only drops may include correctly labelled records: no statement
 found or readable is not a statement that the label is wrong. The NERGAL pass (unchanged).
 
 ## uzp_orzeczenia_pl — held off main until v0.2.8 (update to the sections above)
 
-**Investigated** — 2026-10-21 (slayerlabs/polish-dynaword-v2#112).
+**Investigated** — 2026-10-21.
 
 **Verdict** — `removed` from main until v0.2.8 (planned for v0.2.7 until 2026-10-22); the source was never released. The G1 pass leaves part of the sole
 traders' firm names and addresses unmasked, and round 2 (the section above) missed its address bar on the held-out
@@ -4210,8 +4204,8 @@ while the folder is missing.
 **Investigated** — 2026-10-22.
 
 **Verdict** — `removed` from main until the scope of the catalogue's licence statement is answered; the source was
-never released. The catalogue's CC BY-SA 3.0 sentence may cover only the accessibility pictograms (see "Licence scope
-(open)" above), and main is as public as a release tag: the default config (`data/*/*.parquet`) served the shard to
+never released. The catalogue's CC BY-SA 3.0 sentence may cover only the accessibility pictograms (see "Licence scope"
+above), and main is as public as a release tag: the default config (`data/*/*.parquet`) served the shard to
 `load_dataset` and the viewer while it was on main. Removed: the whole `data/kultura_bez_barier_pl/` folder (shard,
 sample, attribution, decisions, review table, licence-basis evidence, QA, stats, datasheet and NOTICE), which comes
 back with the answer. Kept: the registry entry (`release: None`), `src/build_kultura_bez_barier_pl.py`,

artifacts/source_license_review_v0_3.md

@@ -46,7 +46,7 @@ Operational result: keep `WiktorS/polish-news` out of the clean open corpus. Rev
 3. `PleIAs/Polish-PD`: sample OCR quality and filter aggressively.
 4. `pelcra/PLLuMIC`: accept gated terms and import as `dialogue_sft` / final-phase source.
 5. `ipipan/polqa`, `clarin-pl/poquad`, `clarin-pl/PUGG`, `clarin-pl/ComplexQA`: import as QA/eval/final-phase buckets after dedup.
-6. `oscar-corpus/mOSCAR`: only if the above is not enough for modern web diversity and legal review is acceptable.
+6. `oscar-corpus/mOSCAR`: only if the above is not enough for modern web diversity and a rights review accepts it.
 
 ## Required gates before publish
 

artifacts/uzp_orzeczenia_pl_legitimate_interest.md

@@ -1,7 +1,7 @@
 # Legitimate-interest note: personal data in `uzp_orzeczenia_pl`
 
 Our assessment of 2026-10-18, not legal advice. Written for this source; the reasoning is meant to be reused for
-`saos` (slayerlabs/polish-dynaword-v2#55), whose filters differ.
+`saos`, whose filters differ.
 
 ## What the rulings contain
 

artifacts/uzp_orzeczenia_pl_ontology_manifest.json

@@ -73,7 +73,7 @@
     }
   ],
   "pending": [
-    "formal legal review of art. 4(2) application",
+    "formal rights review of art. 4(2) application",
     "cross-source text deduplication",
     "benchmark contamination check",
     "review of quoted third-party text",

configs/source_candidates_v0_3.json

@@ -95,7 +95,7 @@
       "target_bucket": "web_fallback",
       "priority": "medium",
       "decision": "maybe_after_legal_review",
-      "notes": "CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and legal review."
+      "notes": "CC-BY tag but web-crawl copyright/provenance risk remains. Only use after filtering and a rights review."
     },
     {
       "repo_id": "oscar-corpus/OSCAR-2301",

data/biblioteka_nauki/biblioteka_nauki.licence-basis.json

@@ -148,7 +148,6 @@
     "rows_before_review_content_sha256": "cc4ad9ce16aaef3c7bc5197743ad58182e3d5011f78cbf53896083e7a8541954",
     "v0.2.5_parquet_sha256": "cc93cdd5ca3e6576a1da0aa294df1308e879b8516ac30911289e430c69a07b26"
   },
-  "issue": "slayerlabs/polish-dynaword-v2#111",
   "journal_adoption": {
     "consequence": "an issue from before its journal adopted CC may remain wherever the publication year is not earlier than the licence version",
     "obtainable": false,
@@ -513,7 +512,7 @@
       "rows": 37843,
       "rows_sha256": "49a6272befa665daa9c36abd616c951efd0459e5d35a56adea1dc04440772744"
     },
-    "note": "drops decided by the maintainer on 2026-10-20 (v2#111); the table lists every row a rule nominated, `drop` or `keep`, and nothing else",
+    "note": "drops decided by the maintainer on 2026-10-20; the table lists every row a rule nominated, `drop` or `keep`, and nothing else",
     "reasons": {
       "drop": {
         "api_flagged_reprint_translation": "the platform flags the item as a reprint or a translation (`reprintPublication` or `translationPublication`)",

data/biblioteka_nauki/biblioteka_nauki.license-evidence.json

@@ -5,7 +5,7 @@
  "license_spdx": "MIXED (per row: CC-BY-2.0, CC-BY-3.0, CC-BY-3.0-PL, CC-BY-4.0, CC-BY-SA-3.0, CC-BY-SA-3.0-PL, CC-BY-SA-4.0)",
  "review_scope": "Licence name, URL and version of every one of the 42,068 rows of the previous release (v0.2.5 after the Public Domain Mark drop, Hub PR 62), read from the platform API. Not a review of the texts, of the authors' standing to license, or of third-party material inside the items.",
  "finding": "The API gives CC BY 2.0 / 3.0 / 3.0 PL / 4.0 and CC BY-SA 3.0 / 3.0 PL / 4.0 per record; v0.2.5 and the first v0.2.6 build collapsed all of them to 4.0 by keeping only license.iconId. The `license` column now holds the API's version. 12 rows the API no longer serves (HTTP 404) are withheld; so are, by the maintainer's decision of 2026-10-06, 56 rows whose text prints a NonCommercial or NoDerivatives licence and 15 rows whose text says it is reprinted or translated by permission of a publisher. articles_2011995 (API: CC BY-SA 4.0; earlier recovery: CC BY) is kept, labelled CC-BY-SA-4.0. No row's text changed.",
- "after_the_review_of_2026_10_20": "The counts in this file (`shipped`, `withheld`, `flagged_action`, `verified`) describe the state of Hub PR 81, before the review of 2026-10-20 (polish-dynaword-v2#111): 4,142 of those 41,985 shipped rows are dropped, among them the 5 rows flagged as a reprint or translation. Current counts are in biblioteka_nauki.review.csv (the 4,205 reviewed rows, each with its decision and reason), biblioteka_nauki.licence-basis.json (the rules and measurements) and stats.json. The API licence per row is unchanged and still in biblioteka_nauki.license-api.json.",
+ "after_the_review_of_2026_10_20": "The counts in this file (`shipped`, `withheld`, `flagged_action`, `verified`) describe the state of Hub PR 81, before the review of 2026-10-20: 4,142 of those 41,985 shipped rows are dropped, among them the 5 rows flagged as a reprint or translation. Current counts are in biblioteka_nauki.review.csv (the 4,205 reviewed rows, each with its decision and reason), biblioteka_nauki.licence-basis.json (the rules and measurements) and stats.json. The API licence per row is unchanged and still in biblioteka_nauki.license-api.json.",
  "upstream_api": "https://bibliotekanauki.pl/api/{articles|books|chapters}/{id}",
  "retrieved": "2026-10-05..2026-10-06",
  "retrieval_script": "data/biblioteka_nauki/recover_license_versions.py (one global request clock, 3 workers, 0.15 s between request starts, back-off on 429/5xx, stops on 403 or a non-JSON 200; no CAPTCHA or terms page was met)",
@@ -75,7 +75,7 @@
   "text prints a NonCommercial or NoDerivatives licence (maintainer decision 2026-10-06)": 56,
   "text says it is reprinted or translated by permission of a publisher (maintainer decision 2026-10-06)": 15
  },
- "maintainer_decision": "2026-10-06 (polish-dynaword-v2#57, Hub PR 81): keep the 12 HTTP 404 rows withheld; relabel articles_2011995 to CC-BY-SA-4.0 and keep it; withhold 56 rows whose text prints an NC or ND licence and 15 rows whose text says it is reprinted or translated by permission of a publisher; keep rows whose wording is a citation, a quoted work's notice, prose about licence types, a permission for a figure or table, or a translation permitted by the author alone. The ids and one reason per id are in src/clean_biblioteka_nauki.py (TEXT_NC_ND, TEXT_REPRINT) and in the stats `withheld`.",
+ "maintainer_decision": "2026-10-06 (Hub PR 81): keep the 12 HTTP 404 rows withheld; relabel articles_2011995 to CC-BY-SA-4.0 and keep it; withhold 56 rows whose text prints an NC or ND licence and 15 rows whose text says it is reprinted or translated by permission of a publisher; keep rows whose wording is a citation, a quoted work's notice, prose about licence types, a permission for a figure or table, or a translation permitted by the author alone. The ids and one reason per id are in src/clean_biblioteka_nauki.py (TEXT_NC_ND, TEXT_REPRINT) and in the stats `withheld`.",
  "flagged": {
   "translationPublication": 3,
   "reprintPublication": 2

data/biblioteka_nauki/biblioteka_nauki.md

@@ -17,7 +17,7 @@ name, URL, icon) on 2026-10-05/06, one request per row of the previous release
 (`data/biblioteka_nauki/recover_license_versions.py`). The API responses are committed as
 `biblioteka_nauki.license-api.json`. The earlier recovery stored only `license.iconId` (`cc-by` /
 `cc-by-sa`), which drops the version, so v0.2.5 labelled every row 4.0 although the API gives 2.0, 3.0,
-3.0 PL and 4.0 (polish-dynaword-v2#57).
+3.0 PL and 4.0.
 
 | `license` label | documents | API licence (`nameEn`) |
 |---|---:|---|
@@ -38,8 +38,7 @@ recovery (CC BY); it is kept and labelled CC-BY-SA-4.0 (maintainer decision 2026
 37,843 rows shipped, only the `license` column differs from the previous release; ids, order and text
 are unchanged.
 
-Three items labelled "Domena publiczna" were dropped for v0.2.6 (`src/clean_biblioteka_nauki.py`,
-polish-dynaword-v2#52): `books_2142716`, `books_2147851`, `articles_2097363` (376,155 tokens).
+Three items labelled "Domena publiczna" were dropped for v0.2.6 (`src/clean_biblioteka_nauki.py`): `books_2142716`, `books_2147851`, `articles_2097363` (376,155 tokens).
 Biblioteka Nauki gives them API licence id 22, the CC Public Domain Mark 1.0, which is a label, not a
 waiver or a licence. They are original works first published 2019-2022, so the Polish life+70 term has
 not run out, and `books_2147851` itself prints "Copyright by Jerzy Zając 2019".
@@ -51,7 +50,7 @@ licence check does not look at those flags; the 5 rows are dropped in the review
 83 of the 42,068 rows of the previous release are not shipped. The ids, with one reason per id, are in
 `biblioteka_nauki.stats.json` (`withheld`) and, for the two text-based classes, in
 `src/clean_biblioteka_nauki.py`, whose build fails if a listed id is not reached. The text-based
-classes were decided by the maintainer on 2026-10-06 (polish-dynaword-v2#57, Hub PR 81).
+classes were decided by the maintainer on 2026-10-06 (Hub PR 81).
 
 | reason class | rows | what it is |
 |---|---:|---|
@@ -64,8 +63,7 @@ types, a permission for a figure or table, or a translation permitted by the aut
 readings are one reader's judgement of about 300 characters of context around each match.
 
 ## Per-record licence basis
-Each line of `biblioteka_nauki.attribution.jsonl` also says what the row's licence rests on
-(polish-dynaword-v2#111): `licence_basis`, `licence_basis_url` and `licence_basis_date`.
+Each line of `biblioteka_nauki.attribution.jsonl` also says what the row's licence rests on: `licence_basis`, `licence_basis_url` and `licence_basis_date`.
 
 | `licence_basis` | rows | what it is | URL | date |
 |---|---:|---|---|---|
@@ -79,7 +77,7 @@ genuine notes of the work's own licence; rows that print a licence in a form the
 were not counted.
 
 ## Rows dropped after review
-4,142 of the 41,985 rows left by the steps above are dropped (polish-dynaword-v2#111, maintainer
+4,142 of the 41,985 rows left by the steps above are dropped (maintainer
 decisions of 2026-10-20). `biblioteka_nauki.review.csv` (`id,decision,reason`, pinned by SHA-256 in
 `src/clean_biblioteka_nauki.py`) lists the 4,205 rows a rule nominated, each with `drop` or `keep`, and
 nothing else; the build refuses to run when the rules and the table disagree.
@@ -147,7 +145,7 @@ Minimal, per Dynaword guidelines (heavy filtering left to downstream use):
 - licence not confirmed by the API as an allowlisted CC BY / BY-SA version, HTTP 404 (`src/clean_biblioteka_nauki.py`): **12**
 - text prints a NonCommercial or NoDerivatives licence (`src/clean_biblioteka_nauki.py`): **56**
 - text says it is reprinted or translated by permission of a publisher (`src/clean_biblioteka_nauki.py`): **15**
-- reviewed drops (`src/clean_biblioteka_nauki.py`, polish-dynaword-v2#111): **4,142**
+- reviewed drops (`src/clean_biblioteka_nauki.py`): **4,142**
 - read 43,576 → kept 37,843
 
 Token counts are a tiktoken (cl100k) proxy; no Llama-3 count is published.

data/biblioteka_nauki/recover_attribution.py

@@ -16,8 +16,8 @@ HERE = os.path.dirname(os.path.abspath(__file__))
 LIC = os.path.join(HERE, "licenses.jsonl")
 OUT = os.path.join(HERE, "attribution.jsonl")
 PROG = os.path.join(HERE, "attribution_progress.txt")
-# The v0.2.4 run also kept "public-domain": the CC Public Domain Mark, a label rather than a grant
-# (polish-dynaword-v2#52). src/clean_biblioteka_nauki.py drops it from the shipped Parquet; its KEEP
+# The v0.2.4 run also kept "public-domain": the CC Public Domain Mark, a label rather than a grant.
+# src/clean_biblioteka_nauki.py drops it from the shipped Parquet; its KEEP
 # set must equal this one.
 CLEAN = {"cc-by", "cc-by-sa", "cc0"}
 WORKERS = 8

data/biblioteka_nauki/recover_license_versions.py

@@ -1,10 +1,10 @@
 #!/usr/bin/env python3
 """Re-fetch the licence name, URL and version of every shipped Biblioteka Nauki row.
 
-recover_licenses.py kept only `license.iconId` (cc-by / cc-by-sa), which drops the licence version
-(polish-dynaword-v2#57). This asks `GET bibliotekanauki.pl/api/{type}/{id}` once per row of the
-shipped Parquet and stores the API's own `license` object. It judges nothing: the allowlist and the
-fail-closed handling live in src/clean_biblioteka_nauki.py.
+recover_licenses.py kept only `license.iconId` (cc-by / cc-by-sa), which drops the licence version.
+This asks `GET bibliotekanauki.pl/api/{type}/{id}` once per row of the shipped Parquet and stores the
+API's own `license` object. It judges nothing: the allowlist and the fail-closed handling live in
+src/clean_biblioteka_nauki.py.
 
 Polite by construction: one global request clock (`--interval`), a few workers, back-off on
 429/5xx and connection errors (Retry-After honoured), and a circuit breaker. The run stops, and

data/dziennik_ustaw/dziennik_ustaw.md

@@ -86,7 +86,7 @@ works. This is our assessment, not legal advice.
 
 **Credit:** Dziennik Ustaw; Kancelaria Sejmu RP for the ELI interface. As a precaution we also
 give the System Informacyjny Sejmu source note, although the terms page does not mention the
-ELI interface and whether the note applies is open:
+ELI interface:
 
 > Źródło: System Informacyjny Sejmu (Kancelaria Sejmu RP)
 

data/eltec_pol/eltec_pol.md

@@ -30,7 +30,7 @@ Fetched directly from the COST-ELTeC/ELTeC-pol GitHub repository (level1 TEI) by
 Author metadata present for **100** documents. Empty values mean the upstream record did not expose a machine-readable author field.
 
 ## Legal scope note
-Flagged, nothing excluded (polish-dynaword-v2#109). 3 documents (eltec_pol_2455254, eltec_pol_2455279, eltec_pol_2455281) carry 150 numbered note markers (1, 61 and 88) and no note text, and their TEI files hold none either (no <note> element; <front> is title-page lines, <back> is empty), so the notes are missing from ELTeC's files and the files do not say whose notes they would be. A fourth document (eltec_pol_2455282) has 33 markers, 17 of them in 17 note paragraphs of the text (569 characters). No TEI header names an editor of a print edition. One document (eltec_pol_2455267) keeps period spelling by our heuristic count, so the modernisation is not uniform. 10 documents contain a "Przedmowa" or "Wstęp" heading (4 of the "Wstęp" ones are chapter titles, restored with the other chapter titles); we read these as the authors' own and kept them. The appendix 'Dodatek' of eltec_pol_2455321 is the author's draft passages for a chapter, as in Wolne Lektury's text, whose base is S. Pigoń's edition (Glob, 1985), not the 1908 print the TEI header names; its short labels ('Brulionowe rzuty rozdziału...', '(A)', '[Rękopis nieczytelny]') are the editor's and are kept with the drafts. 4 documents contain 3,649 soft hyphens (U+00AD), invisible line-break hints that were digitised with the text.
+Flagged, nothing excluded. 3 documents (eltec_pol_2455254, eltec_pol_2455279, eltec_pol_2455281) carry 150 numbered note markers (1, 61 and 88) and no note text, and their TEI files hold none either (no <note> element; <front> is title-page lines, <back> is empty), so the notes are missing from ELTeC's files and the files do not say whose notes they would be. A fourth document (eltec_pol_2455282) has 33 markers, 17 of them in 17 note paragraphs of the text (569 characters). No TEI header names an editor of a print edition. One document (eltec_pol_2455267) keeps period spelling by our heuristic count, so the modernisation is not uniform. 10 documents contain a "Przedmowa" or "Wstęp" heading (4 of the "Wstęp" ones are chapter titles, restored with the other chapter titles); we read these as the authors' own and kept them. The appendix 'Dodatek' of eltec_pol_2455321 is the author's draft passages for a chapter, as in Wolne Lektury's text, whose base is S. Pigoń's edition (Glob, 1985), not the 1908 print the TEI header names; its short labels ('Brulionowe rzuty rozdziału...', '(A)', '[Rękopis nieczytelny]') are the editor's and are kept with the drafts. 4 documents contain 3,649 soft hyphens (U+00AD), invisible line-break hints that were digitised with the text.
 
 Statistics were recomputed directly from the released parquet file.
 

data/eurlex/eurlex.license-evidence.json

@@ -4,7 +4,6 @@
   "reviewed_at": "2026-10-19",
   "label": "LicenseRef-EU-Reuse-Policy",
   "previous_label": "CC-BY-4.0",
-  "issue": "slayerlabs/polish-dynaword-v2#57, #108; open question: #56 B5",
   "reuse_basis": {
     "documents": [
       {
@@ -58,7 +57,7 @@
     "rows": 242488,
     "chars": 5859513831,
     "tokens": 2333038468,
-    "change": "per-row license column CC-BY-4.0 -> LicenseRef-EU-Reuse-Policy, the 92 IASB-notice documents dropped (v2#57) and the 480 documents the row review drops (v2#108), by src/clean_eurlex.py; every kept row is otherwise identical to v0.2.5"
+    "change": "per-row license column CC-BY-4.0 -> LicenseRef-EU-Reuse-Policy, the 92 IASB-notice documents dropped and the 480 documents the row review drops, by src/clean_eurlex.py; every kept row is otherwise identical to v0.2.5"
   },
   "document_kinds_heuristic": {
     "method": "first 200 characters of each document",
@@ -108,7 +107,7 @@
         "keep:legal_argument": 3,
         "keep:not_a_credit": 31
       },
-      "review_note": "Superseded by the v2#108 row review: each of the 97 documents is a review row; the split is by decision:reason. The 2026-10-06 first-occurrence labels (33 third-party, 25 EU, 7 binary, 32 noise) counted eurlex_1888527 and eurlex_1900544 as credits; both quote a (c) line inside a legal argument (two Advocate General's opinions) and are kept. eurlex_1825861, a Commission staff working document whose photographs need the rights holder's direct permission, is dropped."
+      "review_note": "Superseded by the row review: each of the 97 documents is a review row; the split is by decision:reason. The 2026-10-06 first-occurrence labels (33 third-party, 25 EU, 7 binary, 32 noise) counted eurlex_1888527 and eurlex_1900544 as credits; both quote a (c) line inside a legal argument (two Advocate General's opinions) and are kept. eurlex_1825861, a Commission staff working document whose photographs need the rights holder's direct permission, is dropped."
     },
     "author_header_categories": {
       "method": "regex on the first 3000 characters; counts in the scan file",
@@ -116,12 +115,11 @@
       "eesc": 5154,
       "state_aid_sa": 2941,
       "un_ece_regulation": 253,
-      "note": "Not third-party under the notice by themselves: they are texts of EU bodies or Commission notices. They matter only if those bodies' reuse terms differ (not checked). The audit's rough shares (ECB ~4%, state aid ~2.5%) do not reproduce with these patterns. UN ECE regulations are reviewed and dropped in v2#108 by a wider rule (307 documents; see review); this header count is the v0.2.5 scan's."
+      "note": "Not third-party under the notice by themselves: they are texts of EU bodies or Commission notices. They matter only if those bodies' reuse terms differ (not checked). The audit's rough shares (ECB ~4%, state aid ~2.5%) do not reproduce with these patterns. UN ECE regulations are reviewed and dropped by a wider rule (307 documents; see review); this header count is the v0.2.5 scan's."
     },
     "not_found": "No large-scale reproduction of non-EU copyrighted works was found by these patterns. Recall for third-party text they do not target is unknown (Inferred).",
-    "decision": "2026-10-06, maintainer: drop the 92 IASB-notice documents (Hub PR 71). 2026-10-19, maintainer (v2#108): drop the 480 documents the review marks drop (35 third-party credits, 307 UN ECE regulations, 138 European Parliament stubs). The other (c)-sign and author-header documents are kept.",
+    "decision": "2026-10-06, maintainer: drop the 92 IASB-notice documents (Hub PR 71). 2026-10-19, maintainer: drop the 480 documents the review marks drop (35 third-party credits, 307 UN ECE regulations, 138 European Parliament stubs). The other (c)-sign and author-header documents are kept.",
     "review": {
-      "issue": "slayerlabs/polish-dynaword-v2#108",
       "file": "data/eurlex/eurlex.review.csv",
       "sha256": "082feea1a6d99c2b3c2680189cef4577257a454008fd3e751b45300df51fe5a2",
       "decided": "2026-10-19, maintainer: drop third-party credit documents, UN ECE regulations republished in the Official Journal (widened rule, hand-checked), and the European Parliament divider and discharge stubs; keep the rest",
@@ -775,6 +773,6 @@
     "whether the ECB, EESC, Council, Parliament and the Court of Justice publish reuse terms of their own, and whether they differ from the notice",
     "the 'Unless otherwise specified' clause per document: only the IASB notice, the (c) sign, photo-credit lines and UN ECE regulation headers were searched for",
     "Official Journal documents quoting national, standards bodies' or other private texts in annexes, other than UN ECE regulations",
-    "the Publications Office's position on non-Commission documents (open question, v2#56 B5)"
+    "the Publications Office's position on non-Commission documents"
   ]
 }

data/govpl/govpl.md

@@ -11,7 +11,7 @@ gov.pl — Polish government press releases
 - **Added:** 2026-07-19; PAP reprints removed 2026-10-06; other third-party texts removed 2026-10-19; prosecutor-office releases removed 2026-10-22
 
 ## Licensing — traceable basis
-gov.pl offers its text under CC BY-SA 4.0. The site footer and https://www.gov.pl/web/gov/prawa-autorskie say: "Treści tekstowe publikowane w serwisie (z wyłączeniem treści audiowizualnych), są udostępniane na licencji typu Creative Commons: uznanie autorstwa - na tych samych warunkach 4.0 (CC BY-SA 4.0)." All rows rely on this offer as it stood when they were fetched (2026-07-19). Wayback captures of the page from 2026-07-18 (https://web.archive.org/web/20260718065324/https://www.gov.pl/web/gov/prawa-autorskie) and 2026-10-15 (https://web.archive.org/web/20261015193442/https://www.gov.pl/web/gov/prawa-autorskie) show the same text. Earlier footers were different: CC BY 3.0 PL from the first capture (2019-08-05) to 2022-08-17, CC BY-NC-ND 3.0 PL from 2022-08-19 to 2025-12-12, and today's text from 2025-12-18. By the date listed on gov.pl, 35,207 rows date from the CC BY-NC-ND 3.0 PL period, 24,201 from the CC BY 3.0 PL period and 7,381 from before 2019-08-05. That today's offer also covers texts first published under an earlier footer is our reading; it is not certain. The same page excludes works whose copyright belongs to third parties; using those needs the owners' consent. Articles crediting PAP and other third-party texts are therefore dropped (legal note).
+gov.pl offers its text under CC BY-SA 4.0. The site footer and https://www.gov.pl/web/gov/prawa-autorskie say: "Treści tekstowe publikowane w serwisie (z wyłączeniem treści audiowizualnych), są udostępniane na licencji typu Creative Commons: uznanie autorstwa - na tych samych warunkach 4.0 (CC BY-SA 4.0)." All rows rely on this offer as it stood when they were fetched (2026-07-19). Wayback captures of the page from 2026-07-18 (https://web.archive.org/web/20260718065324/https://www.gov.pl/web/gov/prawa-autorskie) and 2026-10-15 (https://web.archive.org/web/20261015193442/https://www.gov.pl/web/gov/prawa-autorskie) show the same text. Earlier footers were different: CC BY 3.0 PL from the first capture (2019-08-05) to 2022-08-17, CC BY-NC-ND 3.0 PL from 2022-08-19 to 2025-12-12, and today's text from 2025-12-18. By the date listed on gov.pl, 35,207 rows date from the CC BY-NC-ND 3.0 PL period, 24,201 from the CC BY 3.0 PL period and 7,381 from before 2019-08-05; 5 rows whose mistyped year (such as 1024) falls before that date are not counted. That today's offer also covers texts first published under an earlier footer is our reading; it is not certain. The same page excludes works whose copyright belongs to third parties; using those needs the owners' consent. Articles crediting PAP and other third-party texts are therefore dropped (legal note).
 
 ## Provenance
 Ministry/agency press releases fetched directly from gov.pl (editor-content per article, HTML stripped) by src/fetch_govpl.py. Subsites + their non-uniform news-section paths are enumerated by src/discover_govpl.py into the committed manifest src/govpl_subsites.json (gov.pl has no sitemap/API). v0.2.6 removes PAP reprints, the third-party texts listed in govpl.third-party-review.csv and the prosecutor offices' releases from the v0.2.5 Parquet with src/clean_govpl.py, which also writes govpl.attribution.jsonl: each row's gov.pl URL, subsite, title and date.

data/open_icm_pl/open_icm_pl.checksums.json

@@ -2,11 +2,11 @@
   "NOTICE.md": "e4f645cd0c26e4408910ee78d3a0653519a48f76166c7ece644a6992168e8af7",
   "open_icm_pl.attribution.jsonl": "4b490d4505f226317435dd11069c0b071a5e58d497e42e151818439a62c4dfa3",
   "open_icm_pl.decisions.jsonl": "8f1d90bb052398bcd18eae457218f8b5b9d68ce8dd114168b1fff305570a4eda",
-  "open_icm_pl.licence-basis.json": "1f3df27ad104feac062fc4264c5393dbcd5f6abc8502ddacd0690fff2fbca6cb",
-  "open_icm_pl.md": "fe46629782610df7705c3360cb647a904c24f4a881d392426cbfe719dfb9a272",
+  "open_icm_pl.licence-basis.json": "4a3f97a164ddfc1eabc26f7312a268dee4e0a47db8f4eddc114bb0d40e70eeff",
+  "open_icm_pl.md": "8df2f53e731ee50d900a97182eacbe0a814c7c96acc3311a5efe7669d7aeb8f6",
   "open_icm_pl.overlap_audit.json": "95d29efa6a0038913bcb91a4bdc6ba25aa85d081eff5f85fedc9c63e2f50e967",
   "open_icm_pl.parquet": "8a5e2cdbd3bb5575907d2d14e740316b1df010fd24d8c230bf779b5a6f9aa399",
-  "open_icm_pl.publisher-check.json": "df988ed8ee92db30c8d39babebf513faa795eed3b902a876e60d65bdf7394039",
+  "open_icm_pl.publisher-check.json": "a127bace44823cb49ff7bbef57ef2cf2127d9737dc64cff0373dd08cdcb74353",
   "open_icm_pl.qa.json": "0f9bf4531e33d0ec141b50ad08756ac90c663eac0786691ac80bb0f6f265dcce",
   "open_icm_pl.review.csv": "f98a7dc73181b746e91a6033291a944599fc88cb69ce51a2c235c35cf647873a",
   "open_icm_pl.sample.jsonl": "8f79168ad93142aefeb159f0d527a89ad55e1c21971a0320222c51bdbf839ce0",

data/open_icm_pl/open_icm_pl.licence-basis.json

@@ -20,7 +20,6 @@
     "result": "8 rows had a hit: 2 file_note, 3 conflicts, 1 photo credit (a third-party note, not the work's: not a basis), 2 unrelated words; no other printed licence found",
     "rows_read": 85
   },
-  "issue": "slayerlabs/polish-dynaword-v2#111",
   "item_pages": "not requested: open.icm.edu.pl/robots.txt disallows /search, /browse/*, /info/* and admin paths for all agents and all paths for ClaudeBot",
   "licence_files": {
     "bytes": 2894,
@@ -79,7 +78,7 @@
     ],
     "read": "2026-10-20",
     "rows": 80,
-    "sha256": "df988ed8ee92db30c8d39babebf513faa795eed3b902a876e60d65bdf7394039",
+    "sha256": "a127bace44823cb49ff7bbef57ef2cf2127d9737dc64cff0373dd08cdcb74353",
     "what": "for each of the 80 records whose licence label only the repository states, whether the publisher's own page states a licence, as read on 2026-10-20"
   },
   "repository_check": {

data/open_icm_pl/open_icm_pl.md

@@ -4,7 +4,7 @@ This target-specific slice is derived from the deterministic 100-record OPEN ICM
 
 - Acquired: 100; retained after source QA: 95
 - Exact title/DOI overlaps with `biblioteka_nauki` excluded: 10
-- Dropped in the licence review (v2#111; no licence basis beyond the repository's own field, or the year rule): 74
+- Dropped in the licence review (no licence basis beyond the repository's own field, or the year rule): 74
 - Proposed records: 11
 - Proposed tokens: 441,475 (`cl100k_base` proxy)
 - Characters: 1,203,239; author coverage: 100.0%
@@ -44,7 +44,7 @@ the status of every quoted third-party passage.
   (32), CC-BY-SA-3.0-PL (38), CC-BY-SA-3.0 (1), CC-BY-4.0 (7) and
   CC-BY-SA-4.0 (7). `Dozwolony użytek` and NC variants are rejected.
 
-## Licence basis (v2#111)
+## Licence basis
 
 The repository label (`dc.rights`) is set by the depositor, so it is not a
 basis by itself. A record is kept only if a second source states the licence;
@@ -111,7 +111,8 @@ Not established:
 ## Scrubber and residual-PII spot check
 
 Scrubber that ran: **pattern redaction in `src/fetch_open_icm_pl.py`** (e-mail ->
-`[REDACTED:EMAIL]`, labelled phone -> `[REDACTED:PHONE]`, ID/IBAN patterns;
+`[REDACTED:EMAIL]`, labelled phone -> `[REDACTED:PHONE]`; it has no ID or IBAN rule,
+and the 11 kept records hold no IBAN, account-number, 11-digit or ID-card pattern;
 pipeline counts over the 85-record shard before the licence review: 112 e-mail
 and 4 labelled-phone redactions; the 11 kept records hold 15 `[REDACTED:EMAIL]`
 and 2 `[REDACTED:PHONE]` placeholders).

data/open_icm_pl/open_icm_pl.publisher-check.json

@@ -11,7 +11,6 @@
     "publisher_cc": "a publisher page states a Creative Commons licence (quote and scope given; `scope` is `article` when the article's own page or metadata carries it, `journal` when only a journal-level page does)",
     "publisher_other": "a publisher page was read and states no CC licence, or a NC/ND one, or none for this article"
   },
-  "issue": "slayerlabs/polish-dynaword-v2#111",
   "read": "2026-10-20",
   "rows": [
     {

data/parliamentary/parliamentary.md

@@ -29,24 +29,22 @@ The Sejm's interpellations, questions and the government's replies are held unti
   Parliamentary Corpora. PPC's CC BY licence covers its annotations; none of the linguistic
   annotation layers is shipped.
 - Credit for the Sejm material, as a precaution: "Źródło: System Informacyjny Sejmu (Kancelaria Sejmu RP)".
-  The dump came from IPI PAN's server, not from the Sejm's system, and whether the Sejm's
-  source note applies to it is open. The root `NOTICE.md` has the details.
+  The dump came from IPI PAN's server, not from the Sejm's system. The root `NOTICE.md` has
+  the details.
 - **License:** official parliamentary materials are excluded from copyright by
   [art. 4(2) of the Polish Copyright Act](https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf);
   reuse is governed by the
   [Polish Open Data and Reuse of PSI Act](https://eli.gov.pl/api/acts/DU/2023/1524/text.html).
   IPI PAN's PPC page says "The parliamentary data is public domain." (and that the corpus
-  annotations are CC BY). We read the page on 2026-10-18; its footer says it was last edited on
-  2026-10-02, and a copy is archived in polish-dynaword-v2 (`evidence/ppc_page_2026-10-18.html`,
-  pull request 120). We record the sentence as provenance, not as our legal basis: the basis is art. 4(2), our
+  annotations are CC BY). We read the page on 2026-10-18 and keep a copy of the page as read; its footer says it was last edited on
+  2026-10-02. We record the sentence as provenance, not as our legal basis: the basis is art. 4(2), our
   assessment, and it does not rest on that page. The page does not date the sentence's first
   publication. Asked whether the sentence also covers IPI PAN's database right (art. 6 of the Act
   of 27 July 2001 on the protection of databases), IPI PAN confirmed by e-mail on 2026-10-19
   that it covers its right as producer of the database, and agreed to the resource's inclusion in
   this collection under the collection's CC BY-SA 4.0 tag. That is IPI PAN's consent about its
   own rights and about the collection-level tag; it does not give the texts a licence and does not
-  speak for the Sejm or the Senate. The reply is archived in polish-dynaword-v2
-  (`evidence/ipipan_reply_2026-10-19.md`, pull request 122). Asked for its basis, IPI
+  speak for the Sejm or the Senate. The reply is kept on file and is not published. Asked for its basis, IPI
   PAN answered (e-mail, 2026-10-07) that it is public information made available by statute (the
   Open Data Act and the Access to Public Information Act), with no agreement with the
   Chancelleries. Asked whether the documents it received from the Sejm Library (Biblioteka
@@ -57,11 +55,9 @@ The Sejm's interpellations, questions and the government's replies are held unti
   the turns the allowlist keeps (MPs and senators, members of the government, heads of constitutional
   organs and Sejm and Senate staff). We do not treat
   statements of guests at sittings (people from outside Parliament and the administration) as official
-  documents, so they are left out (polish-dynaword-v2#54). The statement does not extend to third-party
+  documents, so they are left out. The statement does not extend to third-party
   texts read out or quoted inside a kept turn (for example citizens' letters), which are kept unfiltered
   as part of the record, and personal data is a separate matter (see "PII" under "Collection and normalization").
-  <!-- polish-dynaword-v2#107 owns the art. 4 wording and the licence label of this card;
-  polish-dynaword-v2#114 changed neither. -->
 
 ## What the texts are
 
@@ -149,7 +145,7 @@ Every step lives in `src/fetch_parliamentary.py`; the shard is reproducible from
   | `sejm_committee_secretary` | "Sekretarz Komisji ... NAME" (base rule) | 543,439 | 8,392 |
   | `mp` | MP labels with a party tag (base rule) | 3,804 | 13 |
 
-  Evidence for the PPC rules (also in the code comments and on polish-dynaword-v2#54): "P. X" and
+  Evidence for the PPC rules (also in the code comments): "P. X" and
   "S. X" are announced as MPs and senators in the same texts ("Głos ma poseł Nowicki" before "P.
   Nowicki", "Głos ma s. Karnicka"); "P." before an office in 1919-39 is "Pan" and is cut before the
   label is classified (12 persons), and "P." in the Senate (17k characters) stays out. A marshal
@@ -263,7 +259,7 @@ against `dziennik_ustaw` and a benchmark-overlap check remain targets). PII trea
 pattern-based (see above). The legal status above is our assessment, not legal advice. Training
 benefit is an untested hypothesis.
 
-Filter check (polish-dynaword-v2#114, 2026-10-19; labels by one labeller, Claude): 480 paragraphs of speaker-attributed text, 40 in each of 12 strata (kept or dropped, times
+Filter check (2026-10-19; labels by one labeller, Claude): 480 paragraphs of speaker-attributed text, 40 in each of 12 strata (kept or dropped, times
 pre-1989 plenary, pre-1989 committee, Sejm plenary and committee since 1989, Senate plenary, Senate
 committee), drawn at random in proportion to length. All shares are shares of characters, not of speeches.
 

data/plwiki_talk/plwiki_talk.checksums.json

@@ -2,10 +2,10 @@
   "NOTICE.md": "a5423296de5fa97853db553ad035bd6182c69c6d887e36f291f9d6b0ab8dcce6",
   "plwiki_talk.attribution.jsonl": "686c4873dcb9c2dba40ed2954e3d15faa28691d28b45e4b6368855f42995ee9e",
   "plwiki_talk.decisions.jsonl": "b8e064a147db4606efb724ab3cbcf73bb86eec64203e15de1e42e64d35863867",
-  "plwiki_talk.md": "8fcd1a0e11889fcb71b662cb566763a70fc5ce8de56f8283684900ddf252d2a1",
+  "plwiki_talk.md": "286eba45716a0ced5fe21af005038ea09f488f36c0692b9dbe8dcf524b2629a8",
   "plwiki_talk.overlap_audit.json": "5ef4592279700c06e3f74b99a46b1ddd80a0c732ba44c17caba130b18f9d39c0",
   "plwiki_talk.parquet": "340e7f90609893ae6cfd033112322ea077c33faf8fb7c6cb7d8c2e270b0079ca",
-  "plwiki_talk.qa.json": "4ee1b629145be3f54692e33a554bebb696462e283f27ca55e71fc2c41a5bb0b9",
+  "plwiki_talk.qa.json": "cc6d37b06b348456b07bc550a9b37158748d5a8b3939cb09158bb63f47903227",
   "plwiki_talk.sample.jsonl": "0a9cc8fdd13a48956efb6b589cd07541c6b7755348b9c089024efc360106705e",
   "plwiki_talk.stats.json": "4bab29a20f64b95a89a03bf1d8c70052956a002a5c4aa1d5c41637178debc050"
 }

data/plwiki_talk/plwiki_talk.md

@@ -28,7 +28,7 @@ its quality was not measured.
 
 File-link residue the stripping leaves (image options such as `thumb|200px`, bare file
 titles) is removed; category lines stay, since the talk pages discuss categories. v0.2.6
-removed it from the committed text with `src/clean_plwiki_talk.py retext` (v2#116):
+removed it from the committed text with `src/clean_plwiki_talk.py retext`:
 536 records changed and none fell under 500 characters; near-duplicate
 decisions were not re-traced (`plwiki_talk.qa.json`, `residue_cleanup`). The review measurements
 below (spot check, overlap audit) were taken on the 43,054 records before it.
@@ -101,7 +101,7 @@ the shipped `last_revision_at` and wikitext length (and the `permalink` revision
 data in them is redacted as in the text, by the same patterns and with the same false positives (two ISBNs in a
 row can read as an account number, a DOI or a `::` as an IP address); `qa.json` (`citations`) counts the
 redactions.
-v0.2.5 labelled every row `CC-BY-SA-4.0`; before this relabel, main labelled each row by its page's
+The contributed build (Hub PR 36) labelled every row `CC-BY-SA-4.0`; before this relabel, main labelled each row by its page's
 last revision alone (`CC-BY-SA-4.0` from 2023-06-07).
 
 Text quoted from other works is not under these licences. It is used under the quotation right

data/plwiki_talk/plwiki_talk.qa.json

@@ -2,7 +2,6 @@
   "benchmark_overlap": "pending",
   "citations": {
     "citations": 5215,
-    "issue": "slayerlabs/polish-dynaword-v2#110",
     "redacted": {
       "account_candidate": 3,
       "email": 2,
@@ -35,7 +34,6 @@
       "shipped_before_cutoff": 17990
     },
     "created_on_cutoff_day": 3,
-    "issue": "slayerlabs/polish-dynaword-v2#110",
     "stub": {
       "checked_at": "2026-10-18",
       "name": "plwiki-20261001-stub-meta-history.xml.gz",
@@ -67,7 +65,6 @@
     },
     "dropped_duplicate": [],
     "dropped_short": [],
-    "issue": "slayerlabs/polish-dynaword-v2#116",
     "not_retraced": {
       "note": "near-duplicate decisions and rejected rows' characters are the build's",
       "rejected_as_copies_of_dropped_rows": 0

data/plwiki_user_talk/plwiki_user_talk.checksums.json

@@ -2,10 +2,10 @@
   "NOTICE.md": "3f0c91b0658a4831f6d6063164532e2f531ac42f90044e5bf71fa649f687ae48",
   "plwiki_user_talk.attribution.jsonl": "6d8502cce055525c00afb267df38a8dc22573f7a6a1710fe44457b719a21972b",
   "plwiki_user_talk.decisions.jsonl": "20758b5be755965f3c2e1f45abdbffa49cdbae2aea1cc1ab715d9fba38def0c1",
-  "plwiki_user_talk.md": "8668881ddb8f29961770b6c6e40d6e0bd4b58607f324ffdc63407f594e434a71",
+  "plwiki_user_talk.md": "03e37e1bee783dfde7f5f5283e3f538693455489812dce60face246e900f4e8c",
   "plwiki_user_talk.overlap_audit.json": "52bbe088455f25ed082470c8ffd93ddcebfef633f514cf3c6dc565afc35afc6e",
   "plwiki_user_talk.parquet": "13ccc93cf72528a2cda5045a16d6f942af5f1a39727d9c3781fa903ea4a794f9",
-  "plwiki_user_talk.qa.json": "f6796a8e3015a8bcaf0b2ba9c7b34c9b91eb7bbf0940c88d242e725981d054a8",
+  "plwiki_user_talk.qa.json": "2f1a227eae8df37df66ea1f77b8ac567212b3cc0f9a6c11e83842aabe42d21d7",
   "plwiki_user_talk.sample.jsonl": "00864a3a50bf20cf434fa2379e27c5b00a12f7a210a2a55449e49ee7880d5e5a",
   "plwiki_user_talk.stats.json": "37733b981591ca4ad7e64f172da7af8f953b4a3020b7e45cf887a291eb639983"
 }

data/plwiki_user_talk/plwiki_user_talk.md

@@ -29,7 +29,7 @@ plwiki_talk slice.
 
 File-link residue the stripping leaves (image options such as `thumb|200px`, bare file
 titles) is removed; category lines stay, since the talk pages discuss categories. v0.2.6
-removed it from the committed text with `src/clean_plwiki_talk.py retext` (v2#116):
+removed it from the committed text with `src/clean_plwiki_talk.py retext`:
 27,221 records changed and 232 fell under 500 characters and were dropped
 (`too_little_text_after_residue_cleanup` in the decisions); near-duplicate
 decisions were not re-traced (`plwiki_user_talk.qa.json`, `residue_cleanup`). The review measurements
@@ -103,7 +103,7 @@ the shipped `last_revision_at` and wikitext length (and the `permalink` revision
 data in them is redacted as in the text, by the same patterns and with the same false positives (two ISBNs in a
 row can read as an account number, a DOI or a `::` as an IP address); `qa.json` (`citations`) counts the
 redactions.
-v0.2.5 labelled every row `CC-BY-SA-4.0`; before this relabel, main labelled each row by its page's
+The contributed build (Hub PR 40) labelled every row `CC-BY-SA-4.0`; before this relabel, main labelled each row by its page's
 last revision alone (`CC-BY-SA-4.0` from 2023-06-07).
 
 Text quoted from other works is not under these licences. It is used under the quotation right

data/plwiki_user_talk/plwiki_user_talk.qa.json

@@ -2,7 +2,6 @@
   "benchmark_overlap": "pending",
   "citations": {
     "citations": 3200,
-    "issue": "slayerlabs/polish-dynaword-v2#110",
     "redacted": {
       "email": 1,
       "ipv4": 4,
@@ -33,7 +32,6 @@
       "shipped_before_cutoff": 243737
     },
     "created_on_cutoff_day": 57,
-    "issue": "slayerlabs/polish-dynaword-v2#110",
     "stub": {
       "checked_at": "2026-10-18",
       "name": "plwiki-20261001-stub-meta-history.xml.gz",
@@ -994,7 +992,6 @@
         "id": "plwiki_user_talk_5373770"
       }
     ],
-    "issue": "slayerlabs/polish-dynaword-v2#116",
     "not_retraced": {
       "note": "near-duplicate decisions and rejected rows' characters are the build's",
       "rejected_as_copies_of_dropped_rows": 0

data/sejm_api/sejm_api.md

@@ -41,8 +41,7 @@ The same page says that SIS material, in particular live streams and archive mat
 produced by iTV Sejm, "nie mogą być elementem komercyjnej oferty żadnej agencji, stacji
 telewizyjnej, portalu ani innego podmiotu". We report that clause as information only. It is
 not a licence term of this dataset, and we have not verified whether it reaches these texts
-or binds a re-user. A dated copy of the page (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`)
-is kept in polish-dynaword-v2 (`evidence/sis_copyright_2026-10-19.*`, pull request 121);
+or binds a re-user. We keep a copy of the page as read on 2026-10-19 (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`);
 the root `NOTICE.md` has the details.
 
 ## Provenance and reproducibility
@@ -97,7 +96,7 @@ is `id, text, source, added, created, token_count, license, author`.
 
 ## Trailer cut
 
-The rights audit (slayerlabs/polish-dynaword-v2#57) found that the snapshot
+The v0.2.6 rights audit found that the snapshot
 appends to 34,980 of the 38,812 fetched speeches a "Przebieg posiedzenia" header
 of a later statement of the same sitting day, and sometimes that statement's
 text. `src/clean_sejm_api.py` cuts every trailer (23,485,670 characters). An
@@ -118,7 +117,7 @@ texts that became exact duplicates are dropped.
 The plenary stenograms record everyone who takes the floor, and the snapshot keeps them
 all: members, ministers, and also guests such as representatives of citizens'
 legislative initiatives, heads of agencies, prosecutors and a foreign parliament's
-speaker. Since slayerlabs/polish-dynaword-v2#99 this source applies the allowlist of
+speaker. Since v0.2.6 this source applies the allowlist of
 `parliamentary` and `sejm_committee_transcripts`: MPs and senators, government members,
 organs acting in their official capacity, and chancellery staff. A speech whose speaker
 held none of these roles on that sitting day is dropped. A text that was not delivered

data/sejm_api_2011_2022/sejm_api_2011_2022.md

@@ -10,7 +10,7 @@ from the already merged `sejm_api` 2023-onward artifact.
 - Source documentation: https://api.sejm.gov.pl/sejm.html
 - Domain: political/spoken
 - Language: Polish (`pl`)
-- License: `public-domain (official documents)` (a descriptive term, not a licence; see "Licensing and privacy")
+- **License:** `public-domain (official documents)` (a descriptive term, not a licence; see "Licensing and privacy")
 - Created: 2011-11-08 through 2022-12-14
 - Added: 2026-08-31
 - Contributor: Piotr Styła (`PiotrSty` on Hugging Face; `PiotrStyla` on GitHub)
@@ -49,7 +49,7 @@ from the already merged `sejm_api` 2023-onward artifact.
 
 ## Trailer cut
 
-The rights audit (slayerlabs/polish-dynaword-v2#57) found that the snapshot
+The v0.2.6 rights audit found that the snapshot
 appends to 105,920 of the 118,902 speeches a "Przebieg posiedzenia" header of a
 later statement of the same sitting day, and sometimes that statement's text.
 `src/clean_sejm_api_2011_2022.py` cuts every trailer (97,723,744 characters) with
@@ -68,17 +68,17 @@ published, 34 of them cut from a speech that is itself dropped as contained.
 
 ## Existing-source overlap
 
-The rights audit (slayerlabs/polish-dynaword-v2#57) found that this source
+The v0.2.6 rights audit found that this source
 re-publishes text already in `parliamentary`. A speech is dropped when one document
 of a pinned reference set holds at least half of its 5-word shingles (containment
 >= 0.5 over NFKC-casefolded `\w+` tokens, the tokenisation of the repo's `near_dedup`,
 in `src/shingle_containment.py`; the trailers are cut first, see "Trailer cut"). The
 reference set is every Parquet whose registry `release` is not `None`:
-`parliamentary` (the PPC TEI rebuild of slayerlabs/polish-dynaword-v2#54, Hub PR 108, with
+`parliamentary` (the PPC TEI rebuild, Hub PR 108, with
 Hub PR 111's encoding repair and exact-copy drop) and `sejm_api` (after its own trailer cut),
 each pinned by SHA-256 in `REFERENCES`. Re-pinning from Hub PR 108's `parliamentary` to Hub
 PR 111's left every output of this script byte-identical except the pin in the stats.
-Re-pinning `sejm_api` to its allowlist-filtered release (slayerlabs/polish-dynaword-v2#99)
+Re-pinning `sejm_api` to its allowlist-filtered release
 publishes 3 rows more (Wiewiórowski 2014-08-29, Bielak-Jomaa 2015-04-22, Nowak 2019-05-16),
 all contained in one `sejm_api` document that the filter removed, and changes no other row.
 The comparison with the v0.2.5 build below was measured before that change.
@@ -119,7 +119,7 @@ Against the v0.2.5 `parliamentary` this step dropped 93,861 rows. Against the re
   speaker because no allowlist rule matches: representatives of citizens' legislative
   initiative committees, the President's Chancellery, the Prosecutor General's office
   and agency heads. The allowlist did not apply to this source then; it does since
-  slayerlabs/polish-dynaword-v2#99 (see "Speaker allowlist"), after this comparison. The
+  v0.2.6 (see "Speaker allowlist"), after this comparison. The
   figures here are those of the containment step and do not include that drop.
 - 470 rows (1,084,769 characters) are not in the dump: under 0.3 of their shingles
   occur in its Sejm sittings of the same day. 435 of them are statements entered into
@@ -133,7 +133,7 @@ Against the v0.2.5 `parliamentary` this step dropped 93,861 rows. Against the re
   speaker lines inside a turn (8) and a committee chair who is not a member (1).
 
 The 23 speeches of MP Piotr van der Coghen stay dropped: the rebuild keeps a name with a
-particle when it is a member's (slayerlabs/polish-dynaword-v2#54, decision 1A).
+particle when it is a member's.
 
 Known limits: containment is measured against single reference documents, not a union,
 and a threshold of 0.5 with no minimum size also drops short formulaic speeches. The
@@ -148,7 +148,7 @@ the speech's own wording in the reference (4 of 10 between 0.3 and 0.5, 10 of 10
 
 The snapshot keeps everyone who takes the floor in the plenary sittings: members, ministers,
 and also guests such as representatives of citizens' legislative initiatives, heads of
-agencies, prosecutors and the President's Chancellery. Since slayerlabs/polish-dynaword-v2#99
+agencies, prosecutors and the President's Chancellery. Since v0.2.6
 this source applies the allowlist of `parliamentary` and `sejm_committee_transcripts`: MPs
 and senators, government members, organs acting in their official capacity, and chancellery
 staff. A speech whose speaker held none of these roles on that sitting day is dropped. A text
@@ -253,8 +253,7 @@ The same page says that SIS material, in particular live streams and archive mat
 produced by iTV Sejm, "nie mogą być elementem komercyjnej oferty żadnej agencji, stacji
 telewizyjnej, portalu ani innego podmiotu". We report that clause as information only. It is
 not a licence term of this dataset, and we have not verified whether it reaches these texts
-or binds a re-user. A dated copy of the page (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`)
-is kept in polish-dynaword-v2 (`evidence/sis_copyright_2026-10-19.*`, pull request 121);
+or binds a re-user. We keep a copy of the page as read on 2026-10-19 (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`);
 the root `NOTICE.md` has the details.
 
 The source release heuristically redacts email addresses, PESEL

data/sejm_committee_transcripts/sejm_committee_transcripts.md

@@ -13,11 +13,12 @@ heads of constitutional organs, Sejm and Senate staff; see "Speaker allowlist"):
   `GET /sejm/term{n}/committees/{code}/sittings/{num}/pdf`. Each record pins its
   exact source PDF endpoint in `sejm_committee_transcripts.attribution.jsonl`
   (`source_url`); the per-sitting `content_sha1` fingerprints the transcript body.
-- License: no licence is attached to the texts. Our assessment is that the kept
-  transcripts, which record statements by official speakers, are excluded from copyright by
-  [art. 4 pkt 2 of the Polish Copyright Act](https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf);
-  the per-record `license` value `public-domain (official documents)` is a descriptive term
-  for that status, not a licence (see "Legal status" below). The
+- **License:** `public-domain (official documents)` (the `license` value of every record;
+  a descriptive term, not a licence). No licence is attached to the texts. Our assessment
+  is that the kept transcripts, which record statements by official speakers, are excluded
+  from copyright by
+  [art. 4 pkt 2 of the Polish Copyright Act](https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf)
+  (see "Legal status" below). The
   [Polish Open Data and Reuse of PSI Act, arts. 2(12), 5, 6, 14, 15, 17](https://eli.gov.pl/api/acts/DU/2023/1524/text.html)
   is cited as the reuse framework. No upstream Creative Commons grant is claimed. Evidence snapshot:
   `sejm_committee_transcripts.license-evidence.json` (verified 2026-09-18;
@@ -65,8 +66,7 @@ The same page says that SIS material, in particular live streams and archive mat
 produced by iTV Sejm, "nie mogą być elementem komercyjnej oferty żadnej agencji, stacji
 telewizyjnej, portalu ani innego podmiotu". We report that clause as information only. It is
 not a licence term of this dataset, and we have not verified whether it reaches these texts
-or binds a re-user. A dated copy of the page (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`)
-is kept in polish-dynaword-v2 (`evidence/sis_copyright_2026-10-19.*`, pull request 121);
+or binds a re-user. We keep a copy of the page as read on 2026-10-19 (SHA-256 `cb4d21a44f5b09d50cdf95518aad77632b2750b5df6535c1d8aa894152b80469`);
 the root `NOTICE.md` has the details.
 
 ## What the texts are

data/studia_bas/studia_bas.license-evidence.json

@@ -297,7 +297,7 @@
     },
     "pdf_notices": {
       "retrieved_at": "2026-10-06T17:08Z/17:12Z",
-      "note": "A maintainer downloaded each issue PDF in a browser from orka.sejm.gov.pl; scripted GETs to the issue agent got an HTTP 403 from the host's bot protection (slayerlabs/polish-dynaword-v2#79). url is the browser's recorded download source. Every file's sha256 equals its eligible_pdfs entry, so these are the bytes the build read. Text was extracted with PyMuPDF; a notice counts when, after whitespace normalisation, it is pdf_issue_editorial_notice or pdf_article_license_notice word for word. No other licence wording occurs: the remaining matches for licencj-/licens- are article prose (licencjat, patent licences, music licences).",
+      "note": "A maintainer downloaded each issue PDF in a browser from orka.sejm.gov.pl; scripted GETs to the issue agent got an HTTP 403 from the host's bot protection. url is the browser's recorded download source. Every file's sha256 equals its eligible_pdfs entry, so these are the bytes the build read. Text was extracted with PyMuPDF; a notice counts when, after whitespace normalisation, it is pdf_issue_editorial_notice or pdf_article_license_notice word for word. No other licence wording occurs: the remaining matches for licencj-/licens- are article prose (licencjat, patent licences, music licences).",
       "issue_editorial_notices": 17,
       "article_notices": 115,
       "article_notices_vs_rows": "115 articles carry the notice; the build read 115 and kept 112 (3 dropped by the language gate, in issues 67, 69 and 82).",

data/ted_pl/ted_pl.license-evidence.json

@@ -4,8 +4,8 @@
   "second_read_at": "2026-10-06T00:19:00Z",
   "license": "TED notice status, no licence granted",
   "license_spdx": "LicenseRef-TED-Notice-Status",
-  "status": "no-licence-status-recorded-held-for-nergal",
-  "review_scope": "Rights status and PII pattern scan. 15,544 shipped notices re-fetched; 1,320 removed by TED buyer legal type in the v0.2.6 rebuild (buyer_type_rebuild_2026_10) and 14 as third-party text (third_party_sample_review_2026_10, third_party_screen_2026_10); 14,210 retained. Source stays release: None until the NERGAL pass (slayerlabs/polish-dynaword-v2#60). Method: .agents/skills/source-rights-review. V = read on the page on the retrieval date; I = inferred, not verified, not confirmed.",
+  "status": "no-licence-status-recorded",
+  "review_scope": "Rights status and PII pattern scan. 15,544 shipped notices re-fetched; 1,320 removed by TED buyer legal type in the v0.2.6 rebuild (buyer_type_rebuild_2026_10) and 14 as third-party text (third_party_sample_review_2026_10, third_party_screen_2026_10); 14,210 retained. The source ships in v0.2.6 without a NERGAL pass, which runs before v0.2.7. Method: .agents/skills/source-rights-review. V = read on the page on the retrieval date; I = inferred, not verified, not confirmed.",
   "finding": "No licence is granted for the notice text and none is claimed; LicenseRef-TED-Notice-Status names a status, not a licence, and the label is provisional. (1) EU: the TED legal notice states that procurement notices in the Supplement to the OJEU can be freely reused for commercial or non-commercial purposes unless otherwise noted (V); we follow it as terms of reuse, not as a grant, because the Decision it names covers Commission documents (I). (2) PL, our assessment: notices of Polish public-sector contracting authorities may be official documents excluded from copyright (Copyright Act art. 4 pkt 2, a contested reading), many short or formulaic passages are not protected works, and the Open Data Act supports re-use where it applies (I). Only notices whose TED buyer legal type is a public authority, a body governed by public law or a public undertaking are kept. Third-party works and personal data are not covered; status outside Poland has not been assessed. Open points are listed in not_checked.",
   "basis_eu": {
     "mark": "V",
@@ -143,8 +143,8 @@
     "whether each company contracting authority is a 'podmiot zobowiazany' under art. 3",
     "whether the 'urzedowe dokumenty' carve-out of Copyright Act art. 4 pkt 2 covers a tender notice",
     "equivalence between TED notice content and the same content on the national BZP platform",
-    "the confirmation of the Polish status by a ruling or an authority's statement (open question)",
-    "personal names (NERGAL pass, v2#60)",
+    "the confirmation of the Polish status by a ruling or an authority's statement",
+    "personal names (NERGAL pass)",
     "the raw TED downloads are kept only in the working directory of the rebuild, not in the repository",
     "the buyer legal type declared to TED is taken as given; pre-eForms notices use the Search API's mapping of the older authority-type code"
   ],

data/ted_pl/ted_pl.md

@@ -39,7 +39,7 @@ notices of private contracting entities are excluded (see "Filtering"). We grant
 no licence for these texts, and none follows from the TED legal notice, whose
 terms we follow. `LicenseRef-TED-Notice-Status` names this status; it is not a
 licence, and the label is provisional. Clause text, URLs, retrieval dates and
-page hashes are in `ted_pl.license-evidence.json` (slayerlabs/polish-dynaword-v2#57).
+page hashes are in `ted_pl.license-evidence.json`.
 
 **EU: TED legal notice** (<https://ted.europa.eu/en/legal-notice>, read
 2026-10-05, sha256 of the page as read `c010943c01af22cf9cb30a32ca4044a3a1ee4e36a090e2dd2d8ee3d15a498997`).

data/wikibooks/wikibooks.build.json

@@ -449,7 +449,6 @@
         "title": "Linux/SteamOS"
       }
     ],
-    "issue": "slayerlabs/polish-dynaword-v2#116",
     "rows_changed": 2141,
     "step": "fetch_wikimedia.drop_residue"
   },

data/wikibooks/wikibooks.gfdl-review.json

@@ -549,6 +549,6 @@
  ],
  "updated": {
   "at": "2026-10-19",
-  "note": "Tokens and labels follow the release after the markup cleanup (slayerlabs/polish-dynaword-v2#116) and the labels by contribution date (v2#110). The pattern and the closest audit pattern match the same 45 and 25 documents in the cleaned text, so the classes and notes stand. Tokens before: matched 122,192, source 15,313,229, closest 69,072. The lone file-link option lines dropped later the same day (v2#116) took wikibooks_1712 from 5,309 to 5,307 tokens: matched 121,973 to 121,971, source 15,261,326 to 15,260,992, closest 68,919 to 68,917; same 45 and 25 documents. Re-cleaning with the file-link rules of Hub PR 135 (v2#116), also on 2026-10-18, took wikibooks_24460 from 4,442 to 4,435 tokens: matched 121,971 to 121,964, source 15,260,992 to 15,260,848, closest 68,917 to 68,910; same 45 and 25 documents. On 2026-10-19 the 4 documents created after 2008-11-01 were withheld, and 3 pages carrying {{GFDL-tekst}} (no pattern match here: the template is dropped with the markup): matched 45 to 41 (121,964 to 85,331 tokens), source 15,260,848 to 15,220,800, closest 25 to 22 (68,910 to 67,896)."
+  "note": "Tokens and labels follow the release after the markup cleanup and the labels by contribution date. The pattern and the closest audit pattern match the same 45 and 25 documents in the cleaned text, so the classes and notes stand. Tokens before: matched 122,192, source 15,313,229, closest 69,072. The lone file-link option lines dropped later the same day took wikibooks_1712 from 5,309 to 5,307 tokens: matched 121,973 to 121,971, source 15,261,326 to 15,260,992, closest 68,919 to 68,917; same 45 and 25 documents. Re-cleaning with the file-link rules of Hub PR 135, also on 2026-10-18, took wikibooks_24460 from 4,442 to 4,435 tokens: matched 121,971 to 121,964, source 15,260,992 to 15,260,848, closest 68,917 to 68,910; same 45 and 25 documents. On 2026-10-19 the 4 documents created after 2008-11-01 were withheld, and 3 pages carrying {{GFDL-tekst}} (no pattern match here: the template is dropped with the markup): matched 45 to 41 (121,964 to 85,331 tokens), source 15,260,848 to 15,220,800, closest 25 to 22 (68,910 to 67,896)."
  }
 }