Datasets:

SlayerLab
/

polish-dynaword

Running log of sources that were investigated, whether or not they were added. Append-only, version-independent — unlike the *_v0_3.md artifacts, which are point-in-time snapshots of one release's candidate review.

Purpose: stop the same source being re-investigated from scratch every few months. A negative finding is as valuable as a positive one, as long as it records what was checked and what would change the answer.

Related, don't duplicate:

source_scouting_v0_3.md — v0.3 candidate shortlist (HuggingFace-centric)
source_candidate_audit_v0_3.md — machine-generated audit of those HF cards
source_license_review_v0_3.md — v0.3 license decision table

How to add an entry

One ## section per source. Keep these fields:

Investigated — date, so a stale finding is visible as stale.
Verdict — added / blocked / hold / rejected, plus the one-line reason.
Enumeration — how you would actually get the data, or why you can't.
License evidence — what you checked, with URLs. Separate what you verified yourself from what you read second-hand; label the weak ones.
Blockers — what must be resolved before this can ship.
Lesson — anything generalizable to other sources. Optional.

Write down the API endpoints and quirks you had to dig for. That digging is the expensive part, and it is invisible in a verdict line.

zpe.gov.pl — Zintegrowana Platforma Edukacyjna

Investigated: 2026-08-27 Verdict: hold — technically very tractable, blocked on per-material license evidence. Do not build a fetcher until the license question below is answered.

Ministry of Education platform (successor to epodreczniki.pl), e-materials across all school subjects. Attractive because the corpus has no modern instructional/pedagogic register at all, and this is contemporary, clean, editorially reviewed Polish prose.

Enumeration — solved

There is no sitemap (/sitemap.xml → 404) and site search is client-side, but https://zpe.gov.pl/config.js exposes an undocumented JSON API:

window.CP_CONFIGURATION = {... "apiUrl":"https://zpe.gov.pl/api/v1/" ...}

GET https://zpe.gov.pl/api/v1/search?query=&limit=20&offset=N
→ {"count": 23843, "offset": 0, "limit": 20,
   "data": [{"id": "P18l4AKOi", "title": "Fotosynteza",
             "subject": [...], "stage": [...], "tags": [], ...}]}

Verified directly (2026-08-27):

count = 23,843 materials for the empty query.
Deep paging works — offset=5000 returns rows, no cursor needed.
robots.txt is User-agent: * / Allow: /.
Article pages /a/<slug>/<CODE> are fully server-rendered; text lives in data-editor-content blocks. No JS execution required to extract prose.
Book pages /b/<slug>/<CODE> render their material list client-side (0 server-side /a/ links), so /b/ is useless for crawling. Use the API.

Other endpoints probed and 404: api/v1/materials, api/v1/document(s)/<CODE>, api/v1/footer, api/v1/menu, api/v1/search/materials. Only api/v1/search?query= responded.

License evidence — the blocker

Verified directly:

No document-level license marker exists in the served HTML. The only machine-readable license strings are per-figure, inside <div class="figure_caption_bibliography">, e.g. Źródło: GroMar, licencja: CC BY 3.0. and Źródło: Sławomir Szaruga, licencja: CC BY 3.0. These attach to illustrations, not to the prose.
The site footer is <div id="footer_app"></div> — JS-rendered. This matters: the accepted legal basis for both govpl and samorzad_gov_pl in src/sources.py is a server-side footer string declaring CC BY-SA 4.0. ZPE has no such string. That precedent does not transfer to ZPE.
The platform regulamin states: "Korzystanie z Materiału odbywać się może tylko i wyłącznie w zakresie dozwolonym przepisami ustawy o prawie autorskim i prawach pokrewnych" — statutory permitted-use only, which is not an open license. Reading the surrounding clauses, this section appears scoped to material uploaded by users to the platform rather than to ministry e-materials, but the wording is not self-limiting and a reviewer will land on it.

Second-hand, not verified — confirm before relying on any of it:

Ministry-produced e-materials are said to be CC BY 3.0, funded under POKL 2007–2013 / POWER 2014–2020.
User-contributed materials are reported to carry CC BY-NC-SA 4.0. NC is already on the reject list (source_scouting_v0_3.md), so those must be excluded, not merely flagged.
Literary works embedded in e-materials are reported to sit under a non-free "licencja edukacyjna" where authors or heirs did not agree to open release. This is the highest-risk content class on the platform.

Consequences for design

ZPE cannot be a single-license shard. It has to be per-record license-gated, license_spdx: MIXED, same shape as biblioteka_nauki and europeana.
Available heuristic, not license evidence: ministry materials come back with subject and stage populated; user uploads return "subject": [], "stage": [] (the first empty-query result is an all-caps user upload with both empty). Usable as a pre-filter, useless as provenance.
Rough size if 10–15k ministry materials survive the gate: **30–100M tokens**. Estimate only — nothing was counted. Good register, modest scale.

Extraction hazards (seen in a single sampled page)

Interactive-exercise chrome: Pokaż podpowiedź, Pokaż odpowiedź buttons.
JSON blobs embedded in data-editor-params attributes (gameTitle, feedbackPositive, …) — must not leak into text.
Per-figure attribution lines repeated throughout the body.
Glossary / Zapamiętaj! boilerplate repeated across many materials — needs frequency-based boilerplate detection, not a hand-written blocklist.

What would unblock this

Any one of: a resolvable per-material license field (API or page markup); an authoritative ORE/MEN page granting open reuse of e-material text; or written permission. Absent that, ZPE stays out — the datasheet could not honestly answer "where does the license statement live", which is a required field for any PR.

Lesson

Check for a server-rendered license statement early. A JS-rendered footer is not usable license evidence, and "it's a gov.pl domain" is not a license — the govpl and samorzad_gov_pl entries hold up because a specific footer string was fetched and quoted, not because of the domain.

ParlaMint-PL — parliamentary debates 2015–2022

Investigated: 2026-08-27 Verdict: ready to build, low yield — license and distribution verified, the fetcher is the only missing piece. But it is a date-hole patch worth ~36M net-new tokens, not a diversity win. Size it before prioritising it.

Already registered in src/sources.py as parlamint_pl, but there is no shard in data/.

Size — measured, not guessed

From the official stats at github.com/clarin-eric/ParlaMint/tree/v5.0/Build/Metadata (ParlaMint-overview-stats.tsv, ParlaMint-speeches-stats.tsv):

ID  Lang  Bodies   Ts  From     To       Yrs  Mw/Yr  Mw
PL  pl    low+upp  4   2015-11  2022-06  6.7  5.35   36.06

36.06M words, 228,326 speeches, 4 terms, both chambers.
Token conversion measured on our own parliamentary shard (same register, same cl100k encoder; 6,000 docs / 9.7M words): 2.654 tokens/word, 7.24 chars/word.
→ ~96M tokens gross, ≈1% of the current 9.68B corpus.
Download: ParlaMint-PL.tgz, 185.3 MB.

Overlap with the existing `parliamentary` shard — the real issue

ParlaMint-PL is derived from PPC, which is exactly what the parliamentary shard is (1991–2019). Time overlap runs 2015-11 → 2019-12, about 4.2 of the 6.7 years, ≈22M words likely already present. Net new ≈ 14M words ≈ 36M tokens, covering 2020-01 → 2022-06.

That estimate is soft in both directions:

parliamentary has no per-document dates — created is the single literal '1991-01-01, 2019-12-31' on all 324,622 rows. The overlap cannot be measured by filtering, only by comparing content.
PPC includes committee sittings and interpellations; ParlaMint-PL is plenary only. So ParlaMint's 2015–2019 portion is plausibly a subset of what already ships, making the real overlap higher than the time-share implies.

Exact dedup will not catch this. build_dynaword.py dedups on sha1 of the whole document. ParlaMint ships one document per session; PPC's segmentation differs, so byte-identical documents are unlikely and the duplicates pass straight through the gate. This needs near-dedup, for which the repo has no committed tooling — the HPLT expansion's near-dedup was run outside src/. Budget for that, or the shard silently duplicates existing data.

Version note

ParlaMint 5.0 does not extend Polish — PL still ends 2022-06, same as the 3.0 deposit already cited in sources.py (handle 11356/1486; 5.0 is 11356/2004). Upgrading buys corrections, not coverage. Neither version closes the gap to sejm_api (2023 onward): mid-2022 through end-2022 stays empty.

Verified

CLARIN.SI handle 11356/1486 resolves (HTTP 200). The upstream URL recorded in sources.py is correct and live.
License: CC BY 4.0, stated on the repository record.
ParlaMint 3.0 overall: 26 comparable corpora, 7,468,674 utterances, 1,147,800,265 words; debates mostly 2015 → mid-2022.
The distribution ships a plain-text (.txt) edition alongside TEI/XML and linguistically-annotated .ana.xml. Filenames are per-session, e.g. ParlaMint-PL_2017-07-20-sejm-46-3.txt. Use the .txt edition — no TEI parsing needed.
Polish covers both chambers: -sejm- and -senat- session files.

Why it is still worth doing

Fills 2020 → mid-2022, between parliamentary (ends 2019) and sejm_api (2023 onward). Same CC-BY-4.0 family as existing sources, so there is no new license question to answer, and the plain-text edition means no TEI parsing. Cheap. Just do not expect it to move the totals — and do not build it before a source that adds a register the corpus lacks.

Drift found while checking

src/sources.py (the parlamint_pl provenance field) states the corpus was "Fetched directly from the CLARIN.SI plain-text edition (ParlaMint-PL.tgz) by src/fetch_parlamint.py". That script does not exist in src/. The provenance describes a fetch that no committed code performs. Either write the script or correct the field before this ships.

Resolved 2026-10-05: the parlamint_pl entry was removed from src/sources.py (no fetcher, no data, never released). To add ParlaMint-PL later, write the fetcher and register it again as a new source.

Sejm API — interpelacje and zapytania poselskie

Investigated: 2026-08-27 Verdict: hold — recheck. Not a license problem; api.sejm.gov.pl was unreachable from the investigating network on the day, so the endpoints could not be confirmed. Retry before writing anything off.

What happened

api.sejm.gov.pl resolves (194.41.12.231) but every request hangs and then fails — plain curl times out at 45s, a proxied fetch returns ECONNRESET.
https://www.sejm.gov.pl/ responds normally (302), and other external hosts were fine the same session, so this is specific to the API host: either an outage, or that host filtering the source network.
src/fetch_sejm_api.py (committed, working) uses this same host, so the API is normally reachable. Treat as transient until proven otherwise.

Unverified — what to check on retry

Whether sejm/term{N}/interpellations and sejm/term{N}/writtenQuestions exist, and their paging parameters.
Whether the ministry's reply body is retrievable as text, or only as a linked PDF. This decides whether the source is worth building: the replies are the substantial half.
Document counts per term.

Why it is on the list

Reuses the existing fetch_sejm_api.py legal basis unchanged — official parliamentary materials outside copyright under art. 4(2) pr. aut., reusable under the Open Data Act. Question/answer prose is argumentative register, which the corpus is short on. No new license work needed.

CLARIN-PL DSpace (PELCRA Spokes, and the `1000_novels` upstream)

Investigated: 2026-08-27 Verdict: blocked — temporary. The repository is down for reorganization. Nothing to decide until it returns.

Verified

https://clarin-pl.eu/dspace/ serves a maintenance notice: "Trwają prace nad repozytorium. Prowadzimy prace serwisowe oraz reorganizację repozytorium DSpace… Repozytorium będzie ponownie dostępne wkrótce."
Every item handle under it 404s: 11321/312, 11321/544, 11321/545, 11321/547 — nginx 404, not a DSpace "item not found".
The global resolver is broken for this prefix too: hdl.handle.net/11321/* returns HTTP 500.

So no PELCRA record (Spokes, EMO, PARL, EMI) can have its license checked right now. Conversational Polish remains the corpus's biggest register gap, so this is worth re-checking periodically.

Side effect on shipped data

src/sources.py records https://clarin-pl.eu/dspace/handle/11321/312 as the upstream for 1000_novels, which is a released shard. That URL currently does not resolve. Presumed temporary, but the citation in a published datasheet is dead as of this date.

Expected concern when it returns

PELCRA spoken corpora are transcribed private conversation. Expect NC-flavoured or restricted licensing, and expect a personal-data question — these recordings involve identifiable private individuals, not public officials acting officially. Both must be resolved before use.

NIK — Najwyższa Izba Kontroli audit reports

Investigated: 2026-08-27 Verdict: hold — permissive but non-standard license, and enumeration is unsolved. Viable with work; not a quick win.

License evidence — verified

https://www.nik.gov.pl/zasady/prawa-autorskie.html states:

"Korzystanie z materiałów publikowanych w serwisach internetowych Najwyższej Izby Kontroli nie wymaga uzyskania zgody NIK. Nie dotyczy to elementów odrębnie zastrzeżonych (do których prawa autorskie przysługują osobom trzecim), w szczególności oznaczonych materiałów zdjęciowych oraz tekstów publikowanych pierwotnie w innych źródłach."

Attribution to "Najwyższa Izba Kontroli (nik.gov.pl)" is requested, not required.

Assessment: a custom permissive notice, not a standard open license — no CC, no CC0, no SPDX identifier. It carries a third-party carve-out that matters here, because audit reports quote external sources. A separate and probably stronger argument is that audit reports are materiały urzędowe under art. 4 pkt 2 pr. aut., the same basis as dziennik_ustaw and sejm_api. If this source is built, lead with the statutory argument and cite the site notice as corroboration.

Enumeration — unsolved

robots.txt allows crawling, but disallows /szukaj/, /p/, /g2/, /cms/, /do_druku/. /g2/ looks like the file store, which likely means report PDFs are off-limits to crawlers.
/kontrole/wyniki-kontroli-nik/ returns 200 (~72 kB) but the report list is JS-rendered: zero report links in the server HTML, only pagination scaffolding (fmax,js,strona,1.html).
The /prosta/ ("simple") variant is also JS-rendered — 0 report links.
No /sitemap.xml (404).

Same shape as ZPE: a backend API almost certainly exists and would need to be found (check the page's JS config for an apiUrl, as worked for ZPE).

Verdict rationale

Register value is real — long-form analytical prose, unlike statute text. But between a bespoke license, a third-party-content carve-out, PDF paths blocked by robots.txt, and unsolved enumeration, this is several days of work. Lower priority than ParlaMint-PL or Sejm interpelacje.

Billion-scale survey — what can actually move the totals

Investigated: 2026-08-27

Context: most candidates land at 30–200M tokens, which is noise against a 9.68B corpus. This entry records the only options measured at 500M–5B, i.e. comparable to european_hplt_v3_pl (2.56B), eurlex (2.38B), parliamentary (1.65B). All numbers below were measured on the day, not taken from cards.

Token conversion throughout: 2.654 tokens/word, measured on the repo's own parliamentary shard (6,000 docs / 9.7M words, cl100k). Use it instead of guessing — Polish runs far above the ~1.3 rule of thumb for English.

candidate	measured	est. tokens	license	blocker
SAOS judgments	540,338 docs, mean 3,067 w	~4.4B (floor ~2.8B)	art. 4 pkt 2 pr. aut.	register: pure legal
FineWeb-2 `pol_Latn`	152,003,298 docs, 225.9 GB	~83B raw, 4–8B filtered	ODC-By	dedup vs existing HPLT
PleIAs/Polish-PD	68,176 docs, 0.75B words	~2.0B nominal	public domain	OCR corruption
HPLT leftovers	shards `8_5`, `8_6`, 31.6 GB	~230M	CC0	none — just small

SAOS — biggest clean win, already half-built

https://www.saos.org.pl/api/search/judgments?pageSize=1&pageNumber=0 reports totalResults: 540338. Sampled 100 judgments from api/dump/judgments?pageSize=100&pageNumber=0: mean 3,067 words, median 1,920, p10 816, p90 7,197, court types COMMON and CONSTITUTIONAL_TRIBUNAL.

540,338 × 3,067 = 1.66B words → ~4.4B tokens. Using the median instead, to guard against long-tail skew, still gives ~2.75B. Either way this is larger than any shard currently in the corpus.

src/fetch_saos.py already exists (uncommitted as of this date) and documents the 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 verified on sampled text, not assumed — this is the one candidate here where personal data is a live risk rather than a formality.

Cost of shipping it: legal/official goes from 48% to roughly 65% of the corpus. That is the trade, and it should be a deliberate decision rather than a side effect of it being the easy build.

FineWeb-2 Polish — only billion-scale contemporary prose

datasets-server size endpoint, config pol_Latn: 152,003,298 rows / 225.9 GB. At the measured 0.367 tokens/char that is ~83B tokens raw.

License odc-by. Same argument shape as the accepted HPLT basis: the packaging is licensed, the underlying web text is not owned by the publisher. Note that ODC-By is attribution-required, unlike HPLT's CC0 — a different obligation shape, and it has to be stated in the datasheet.

Dedup is not the bottleneck — this was checked, not assumed

The obvious worry is overlap with the existing HPLT shard, since both derive from Common Crawl. It is the smallest of the problems:

The existing HPLT shard is 1.6M docs against FineWeb-2's 152M. Even total overlap removes ~~1% of candidates (~~12% if working from the HQ subset below).
Internal dedup is already done upstream. The schema carries minhash_cluster_size, i.e. FineWeb-2 ran MinHash per language before release.

Cross-dedup against HPLT is still worth doing for hygiene, and the repo still has no committed near-dedup tooling — but it is not what makes this source expensive.

The actual blocker: the HPLT gate cannot be ported

FineWeb-2 pol_Latn columns are:

text, id, dump, url, date, file_path, language,
language_score, language_script, minhash_cluster_size, top_langs

src/clean_hplt_v3.py filters on WDS quality score, register class (ID/OP/HI/IN/NA), and MT-probability. None of those exist in FineWeb-2. Every signal the existing web pipeline depends on is absent, so "filter it like we filtered HPLT" is not an option — it would mean building quality classification over 152M documents from scratch.

`epfml/FineWeb2-HQ` — the better target

A published quality-filtered derivative, config pol_Latn:

13,384,885 rows, odc-by, 180.4 GB parquet (size dominated by an embeddings column, not text).
Adds a quality_score column, retained per document, so the selection threshold stays a tunable knob instead of a baked-in decision — and the classifier is someone else's published work, which is easier to cite than a home-grown one.
Sampled 880 docs across 22 evenly-spread offsets: mean 2,663 chars/doc → ~13.1B tokens for the whole HQ subset. Larger than the entire current corpus.

quality_score → token yield

Distribution from the statistics endpoint (computed on a partial 407,184-row sample, not all 13.4M): min 0.013, max 0.9999, mean 0.145, median 0.0496 — heavily right-skewed, so most of HQ still scores low.

Doc fractions from that histogram, mean lengths from the 880-doc sample:

`quality_score` ≥	% docs	docs	mean chars	est. tokens
0.00 (all)	100%	13,384,885	2,663	13.1B
0.11	31.0%	4,149,000	2,375	3.6B
0.21	20.0%	2,677,000	2,074	2.0B
0.31	14.6%	1,954,000	1,929	1.4B
0.41	11.1%	1,486,000	1,912	1.0B
0.51	8.5%	1,138,000	1,798	0.75B
0.70	4.8%	642,000	1,750	0.41B
0.90	1.8%	241,000	914	0.08B

A second pass computing fractions directly from the 880-doc sample agreed with these within ~15%. Treat them as ±20%, good enough to pick a threshold.

Watch out: quality_score anti-correlates with document length. Mean length falls from 2,663 chars at no threshold to 914 at ≥0.90. The classifier is not rewarding long-form prose, so raising the threshold costs more tokens than the doc-count alone implies, and it may actively select against the extended prose this corpus is short on. Sample the text at whatever threshold is chosen before committing to it.

Recommended operating point

Threshold ≈ 0.20 → ~2.7M docs, ~2.0–2.3B tokens. Comparable to the largest existing shards (european_hplt_v3_pl 2.56B, eurlex 2.38B) and it improves the balance rather than wrecking it: web goes 26% → ~41%, legal drops 48% → ~39%.

Taking all 13.1B instead would put the corpus at ~22.7B with web at ~69% — trading "two-thirds legalese" for "two-thirds crawl". Same failure, different axis.

Work order if this is built

Pick HQ vs full, and a quality_score threshold (see above).
Build the PII scrub. sources.py already claims one for the HPLT shard that has no implementing code in src/; shipping a second web source with the same gap is indefensible.
Near-dedup against the existing HPLT shard — needed, but ~1–12% effect.
Compute. Tokenizing billions of tokens is the genuinely expensive step.

PleIAs/Polish-PD — good on paper, rough in practice

68,176 rows. Sampled word_count at four offsets (n=200): mean 10,971 words, median 8,468 → 0.75B words → **2.0B tokens**. Columns are identifier, file, title, creator, language, word_count, character_count, complete_text. Public domain, pre-1884 monographs and periodicals, largely sourced from Polish digital libraries (jbc.bj.uj.edu.pl in the sample).

Register-wise this is the most attractive of the three — books and periodicals, against a corpus that is only 1.9% literature.

But the OCR is bad. First sampled record (an 1817 newspaper) opens:

U
RUSKI
czwaRTEŁ SONAER
5 PC Nę R
WIADOMOŚC

with pre-1884 orthography throughout (Nayiaśnieyszy, roskazać, klassy).

The existing OCR gate will not catch this. build_dynaword.py applies MIN_ALPHA_RATIO = 0.70 when is_ocr: True, which tests the ratio of letters to symbols. Garbled words made of ordinary letters pass it cleanly. Filtering this source needs a dictionary- or perplexity-based check, which does not exist here. Treat the 2.0B as nominal; real usable yield is materially lower.

HPLT is exhausted — correction to an earlier assumption

https://data.hplt-project.org/three/sorted/pol_Latn/ lists exactly 19 shards: 10_1, 9_1–9_4, 8_1–8_6, 7_1–7_4, 6_1–6_2, 5_1–5_2. There are no WDS bins below 5 for Polish.

sources.py records 17 of those as ingested. The only unused shards are 8_5 (24.24 GB) and 8_6 (7.37 GB) — roughly 230M tokens at the post-filter rate implied by the existing shard.

So "add more HPLT bins" is not a path to billions. It is one small top-up, and then the source is finished.

Conclusion

Every clean billion-scale Polish source is legal, official, or archaic — which follows directly from what falls outside copyright. The only billion-scale contemporary general prose is web crawl, and that is a re-derivation of Common Crawl the corpus already draws on.

If the goal is raw scale: SAOS. If the goal is a corpus that is not two-thirds legalese, there is no billion-scale option, and the honest plan is a number of 50–200M-token sources across different registers.

common_voice_pl — Common Voice Polish validated sentences (v26.0)

Investigated — 2026-09-05 (review round 2026-09-29).

Verdict — added with a unit caveat (proposed with release: None, merge blocked on a maintainer unit decision). 45,043 unique validated speech prompts / 995k cl100k_base tokens, CC0-1.0, from Mozilla Common Voice Scripted Speech 26.0 (Polish).

Enumeration — official dataset record on Mozilla Data Collective (https://mozilladatacollective.com/datasets/cmqinmu2a00winq07gyrtri0q, id cmqinmu2a00winq07gyrtri0q, release 2026-06-12), read via a pinned row mirror with DuckDB column projection (the 4.48 GB audio is never pulled). 138,798 validated clip rows -> 45,043 unique normalized sentences (91,198 repeated clips collapsed; 2,538 too-short and 19 too-long rows rejected).

License evidence — verified AT THE ORIGIN: the MDC dataset record's license field is "Creative Commons Zero v1.0 Universal (CC0-1.0)", and the same license/license_url ride along per source row (build rejects non-CC0 rows). Terms covenant: MDC Terms of Use (Data Consumer License, clause viii) prohibit re-identifying individual contributors/sources — carried to downstream users in the datasheet and registry.

Blockers — (1) unit decision: every doc is a single speech prompt (min 12 / median 56 / max 153 chars; 0 docs reach the 200-char corpus gate, no concatenation exists) — accept sentence units with a per-source gate override + training cap, or withdraw as unit-incompatible; (2) before release: NERGAL pass (full-shard residual scan was clean — 0 contacts across 45,043 docs — but the prompts are user-contributed speech material).

Lesson — speech corpora's natural unit (the prompt sentence) is smaller than a pretraining corpus's document gate; measure the length distribution BEFORE promising the source, and keep the unit question in the PR description up front — it is the first thing a reviewer will ask.

sejm_interpellations — 10th-term written Q&A (interpellations + replies)

Investigated — 2026-09-05 (review round 2026-09-28).

Verdict — added (proposed with release: None). 25,950 documents / 28.2M cl100k_base tokens of written MP questions and ministry replies — argumentative administrative prose, a register the floor-speech shards do not carry.

Enumeration — GET /sejm/term10/interpellations and .../writtenQuestions plus /reply/{key}/body on api.sejm.gov.pl (the endpoints the 2026-08-27 "hold — recheck" entry above could not confirm; they exist and page fine). PDF-only replies (onlyAttachment) and keyless stubs skipped. Term 9 returned X-Total-Count: 0; terms 7–8 skipped (PPC covers through 2019).

License evidence — same basis as sejm_api: Copyright Act art. 4(2) (https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf) + Open Data Act arts. 2(12), 5, 14, 17 (https://eli.gov.pl/api/acts/DU/2023/1524/text.html). No CC grant claimed.

Blockers — none to merge with release: None. Before release: NERGAL pass — interpellations name private constituents (constituent letters and citizen cases are quoted in questions), which is exactly the case the old scrub_pii (#20) could not cover and #38 moved out of this repo. Caveats for the card: single term (10th), private citizens named.

Lesson — the header chrome (nr / recipient / title / signatories / date) belongs in meta, not text: stripping it makes the body comparable across question and reply docs, and the before/after is the fastest reviewer evidence for any Sejm HTML source.

Edukacja Medialna — media-literacy lessons (`edukacja_medialna_pl`)

Investigated — 2026-09-06 (review round 2026-09-28).

Verdict — added. 252 records / 728,987 cl100k_base tokens of CC BY-SA 3.0 lesson materials — small, but a school-teaching register nothing else in the corpus covers.

Enumeration — https://edukacjamedialna.edu.pl/lekcje/ lists every lesson (255 at fetch). Each lesson page carries its source XML: a utwor document with Dublin Core fields and the authored powiesc body. src/build_edukacja_medialna_pl.py fetches page + XML, checksums both (page_sha256, xml_sha256) and builds from the pinned artifacts/raw_xml.jsonl.gz; nothing else is needed.

License evidence — verified at the origin (the mirror is not evidence for itself): every lesson page ends with Licencja: Creative Commons Uznanie autorstwa - Na tych samych warunkach 3.0 linking https://creativecommons.org/licenses/by-sa/3.0/ (e.g. https://edukacjamedialna.edu.pl/lekcje/abc-bezpieczenstwa-w-sieci/), the site footer repeats the grant (http://creativecommons.org/licenses/by-sa/3.0/deed.pl), and the source XML carries dc:rights + dc:rights.license. The build gate requires the statement on page AND XML. Per-record rights/rights_urls and hashes are pinned in attribution.jsonl.

Blockers — none to merge with release: None. Before a release: NERGAL pass (the in-build redaction is e-mail/labelled-phone patterns only — 1 match), cross-source exact/near dedup, benchmark-overlap check, and maintainer review of embedded third-party excerpts and of the corpus-level BY-SA 3.0 treatment (proposal in the datasheet: per-record attribution + share-alike note, adaptation released under CC BY-SA 4.0).

Lesson — cite the license where the publisher states it (per-lesson Licencja: line + XML rights fields), never a mirror. Pin the statement with page_sha256/rights_urls so it survives site redesigns.

ted_pl — TED Polish public-procurement notices

Investigated — 2026-09-08 (review round 2026-09-28).

Verdict — added (proposed with release: None). 15,544 notices / 42.5M cl100k_base tokens of Polish administrative/technical procurement prose (2023-02-15 – 2024-05-15).

Enumeration — 20,588 official TED Search API records for Polish-labelled notices (1,255 via official XML fallback); the PleIAs mirror is a pinned identifier index only (its Polish preview carries U+FFFD damage). Narrative fields rebuilt from API, structured contact/address/person blocks excluded by construction.

License evidence — TED legal notice (https://ted.europa.eu/en/legal-notice): procurement notices freely reusable commercially/non-commercially unless otherwise noted; metadata CC0 (separate grant). Legal instrument behind it: Commission Decision 2011/833/EU on the reuse of Commission documents (https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32011D0833). Text recorded as LicenseRef-TED-Notice-Reuse, not relabelled CC0.

Blockers — none to merge with release: None. Before release: NERGAL pass and per-notice legal-exception review — the 60-record spot check found one real residual contact (ted_pl_188212-2024: "A.Decem - 815384157" — name + unlabelled phone the patterns missed) and free-text names are un-scoped beyond the sample. Text overlap with uzp_orzeczenia_pl is negligible (0.085% chars, 0 paragraphs — shared statutory citations), so ordinary cross-source dedup suffices.

Lesson — unlabelled phone numbers beside personal names defeat label-anchored patterns (same lesson as open_icm_pl's glyph noise): a zero pattern count is not evidence in contact-bearing registers — run the spot check and the model scrubber.

kultura_bez_barier_pl — cultural-accessibility guides (FKzB)

Investigated — 2026-09-07 (review round 2026-09-28).

Verdict — added (proposed with release: None). 9 guides / 218k cl100k_base tokens (316 pages) of prepared didactic Polish for cultural-sector accessibility work - tiny, but a clean CC BY-SA 3.0 source with preserved raw PDFs and per-record evidence.

Enumeration — the foundation's catalogue https://kulturabezbarier.org/publikacje/: 12 linked publications discovered, 9 retained. 3 rejected conservatively: 2 NIMOZ co-published volumes (PDF copyright notices name the third-party institute) and 1 for legacy-font extraction damage.

License evidence — the catalogue's own statement (quoted in the datasheet) offers the foundation-authored publications under CC BY-SA 3.0 with attribution and share-alike terms; reading is catalogue-level ("plik pierwotny"), pictograms explicitly included, no per-PDF embedded assertion claimed. Per-record corroboration: PDF metadata + credit lines; attribution.jsonl pins the evidence URL and scope note.

Blockers — none to merge with release: None. Before release: NERGAL pass — the all-9 spot check found real residual contacts in podsumowanie_wizyt_studyjnych (coordinator names + phones under Tel./SMS label variants the pattern misses, e-mails beside them redacted) — and maintainer sign-off on the catalogue-level scope reading.

Lesson — catalogue-level license statements need the reading written down (what "je" covers), and the safest per-record confirmation is negative: reject every item with third-party authorship signals, as this PR does with the NIMOZ volumes. Also: Tel./SMS-style label variants defeat label-anchored phone patterns — third independent pattern-evasion case in this review round.

plwiki_talk — Polish Wikipedia talk pages (ns 1+5)

Investigated — 2026-09-11 (review round 2026-09-28).

Verdict — added (proposed with release: None). 43,054 pages / 55.6M cl100k_base tokens of user-generated editorial discussion — article talk (ns 1, 41,828 pages) and project talk (ns 5, 1,226 pages). Exact namespace complement to wikipedia (ns 0) and plwiki_user_talk (ns 3).

Enumeration — the same pinned plwiki-20260901-pages-meta-current dump as the ns-3 slice (7 sha1-verified shards); 544,591 non-redirect talk pages enumerated, 43,054 retained after QA/dedup (500,652 dropped as <500 chars).

License evidence — Wikimedia Terms of Use section 7 (CC BY-SA 4.0): https://foundation.wikimedia.org/w/index.php?title=Policy:Terms_of_Use/pl&oldid=584706. Per-record canonical URL + page history is the attribution trail.

Blockers — none to merge with release: None. Before release: NERGAL pass (user-generated talk; a 60-record residual spot check found only a DOI false positive and public signatures, but free text can hide more), and text-level cross-shard dedup per the measured overlap with plwiki_user_talk / wikipedia (review thread).

Lesson — namespace partitioning prevents page-level duplication but not text-level repetition: talk pages quote article text and share templated notices across namespaces. Measure overlap at text level before claiming novelty — the ID-only audit said "0 collisions" and proved nothing.

plwiki_user_talk — Polish Wikipedia user talk, ns 3

Investigated — 2026-09-11 (review round 2026-09-28).

Verdict — added (proposed with release: None). 380,527 records / 381.4M cl100k_base tokens of user-generated editor-to-editor conversation — a register the corpus otherwise does not draw from.

Enumeration — official dump plwiki-20260901-pages-meta-current, 7 sha1-verified shards against dumpstatus.json; ns=3 non-redirect pages with text = 450,867 candidates. Rebuilt deterministically by src/build_plwiki_usertalk_pl.py from the pinned shards.

License evidence — Wikimedia Terms of Use section 7 (community content under CC BY-SA 4.0): https://foundation.wikimedia.org/w/index.php?title=Policy:Terms_of_Use/pl&oldid=584706 (verified, HTTP 200). Per-record canonical page URL + page history is the attribution trail, pinned in attribution.jsonl (380,527 lines, LFS).

Blockers — none to merge with release: None. Before release: a NERGAL pass (user-generated conversation is the policy's core case; the in-build redaction is patterns only and a 60-record residual spot check found 0 hits but covers 0.016%), and cross-shard cleanup against plwiki_talk for 47 shared long notice lines (99.3% of the measured ns3↔ns1/5 line overlap; genuine thread copies are ~0.13% of user-talk chars).

Lesson — text-level overlap audits beat ID audits: the ID comparison reported "0 collisions", the text pass found 28.7% char containment against plwiki_talk — 99.3% of it just 47 template/notice lines shared across namespaces. Also: line-level boilerplate rules need their length cap revisited (the <160-char rule lets long templated notices survive in both corpora).

open_agh_chemistry_pl — Open AGH chemistry textbooks

Investigated — 2026-09-05 (review round 2026-09-28).

Verdict — added (proposed with release: None). 370 modules / 564k cl100k_base tokens from four Polish AGH chemistry textbooks (general, inorganic, polymer, corrosion) — didactic LaTeX-bearing prose.

Enumeration — the AGH e-textbook platform (https://epodreczniki.open.agh.edu.pl/handbook/{29,1394,37,1893}); 374 modules discovered, 370 retained (each with a selection decision). Pinned snapshot PiotrSty/open-agh-chemistry-pl@07d38bfd with raw module HTML.

License evidence — verified AT THE ORIGIN: each book's official EPUB rights page (served by AGH at /rpc/preview/handbooks/{id}/epub/light, preserved verbatim with rights_sha256 in books.json) states "Publikacja udostępniona jest na licencji Creative Commons Uznanie Autorstwa - Na tych samych warunkach 4.0…" with attribution terms and a "Wersja oryginalna" marking requirement. Per-book, not catalogue-level — the strong pattern.

Blockers — none to merge with release: None. Before release: NERGAL pass (two-pattern redaction only; the 60-record spot check was clean — DOIs only), cross-source dedup and benchmark checks.

Lesson — repo hygiene findings worth keeping: (1) sources.py line endings — main's own blob is CRLF today, so the AGENTS.md "commit LF" rule as written would CREATE whole-file rewrites; the minimal-diff rule is "match main", and the repo-wide normalization is maintainer work; (2) the entry shipped without added/release — the registry contract check belongs in the contract test.

openstax_pl — OpenStax Poland textbooks (8 volumes)

Investigated — 2026-09-05 (review round 2026-09-28).

Verdict — added (proposed with release: None). 1,422 pages / 4.63M cl100k_base tokens of Polish academic textbook prose (physics ×3, psychology, micro/macro, marketing, nutrition) — human translations of the English OpenStax originals by named academics.

Enumeration — openstax.org book containers (Rex/Connexions HTML) via openstax.pl catalog; 1,612 pages discovered, 1,422 retained (each with a decision; MathML-conversion failures rejected). Pinned snapshot PiotrSty/openstax-pl-textbooks@5b31f74e.

License evidence — verified AT THE ORIGIN per book: each Polish edition's foreword page (openstax.org/books//pages/przedmowa) states the CC BY 4.0 grant with the attribution condition "uznania autorstwa OpenStax i osób podpisanych pod oryginałem"; links per book in the datasheet/PR, excerpts + sha256 preserved in books.json.

Blockers — none to merge with release: None. Before release: NERGAL pass (two-pattern redaction only; spot check of 60 was fully clean), cross-source dedup and benchmark checks (exercises/tables may reference omitted figures).

Lesson — for translated editions, "by whom" is a licensable-quality question: the foreword credits (named translators per chapter) are the evidence — pin them per record (edition_contributors) instead of asserting "translated" in the abstract. Same for MT: state its absence explicitly when the credits are human.

wiadomosci_statystyczne_pl — Wiadomości Statystyczne / The Polish Statistician (GUS)

Investigated — 2026-09-10 (review round 2026-09-28).

Verdict — added (proposed with release: None). 169 articles / 1.5M cl100k_base tokens of peer-reviewed statistics-journal content (research, statistics history, conference reports, obituaries) — small, but a formal academic register with a hard per-article license gate.

Enumeration — deterministic crawl of the server-rendered https://ws.stat.gov.pl/ /Archives index: 460 articles (2022+), 315 with the explicit per-article CC BY-SA 4.0 line, 194 acquired, 182 after QA/dedup, minus 13 exact title/DOI matches against biblioteka_nauki -> 169.

License evidence — GUS journals under CC BY-SA 4.0 since 2022-01-01 (https://nauka.stat.gov.pl/News/Info/60) AND a verbatim per-article "udostępniony na licencji CC BY-SA 4.0" line on each landing page (preserved HTML snapshots in the source repo; the line also appears in the PDF text). Pre-2022 articles show only the site-wide footer and are excluded — that archive's license status is unresolved.

Blockers — none to merge with release: None. Before release: NERGAL pass (pattern redaction covers e-mail/labelled-phone only; a 60-record spot check found just DOIs/ORCIDs, but case-study text can hide more) and cross-source text dedup (the 13 exclusions were metadata-level title/DOI matches only).

Lesson — the per-article license line kept verbatim in the record text makes the grant self-evidencing in the corpus itself — worth copying for other publishers that stamp each article.

sejm_committee_transcripts — Sejm committee full transcripts (terms 9–10)

Investigated — 2026-09-18 (review round 2026-09-28).

Verdict — added (proposed with release: None). 653,411 speaker turns / 166.3M cl100k_base tokens of spoken multi-party committee discussion — a working register the plenary-only political sources do not cover.

Enumeration — api.sejm.gov.pl: GET /sejm/term{n}/committees and .../sittings, then .../sittings/{num}/pdf for the "pelny zapis" transcripts. 9,175 term-9/10 sittings known; 8,658 have a zapis PDF upstream, 455 have none, 62 failed after retries; 7,746 sittings survive joint-sitting dedup.

License evidence — official parliamentary materials: Copyright Act art. 4(2) (https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf) + Open Data/PSI Act arts. 2(12), 5, 6, 14, 15, 17 (https://eli.gov.pl/api/acts/DU/2023/1524/text.html). No CC grant claimed; evidence snapshot in sejm_committee_transcripts.license-evidence.json.

Blockers — none to merge with release: None. Before release: NERGAL pass (454 pattern redactions cover only e-mail/PESEL/NIP/REGON/labelled phone/IBAN; quoted citizens' statements remain), near-dedup vs dziennik_ustaw/eurlex (embedded statute/print quotations), benchmark-overlap check.

Lesson — pdftotext -layout + role-anchored speaker labels reconstruct turns well (98.9% of the ASW pilot's audited corrected characters), but it is the pilot comparison that makes the reconstruction auditable — keep that step for future spoken sources.

Studia BAS — Kancelaria Sejmu research & public-policy journal (2021–2025)

Investigated: 2026-08-29 Verdict: ready — official open-access transition under CC BY 3.0 PL starting with issue 2(66)/2021; verified per-article license statements and clean born-digital PDF text layer.

Quarterly peer-reviewed journal published by the Biuro Analiz Sejmowych (BAS) of the Chancellery of the Sejm (Kancelaria Sejmu). Focuses on public policy, socio-economic research, constitutional and EU law, public finance, and legislative commentary. Provides contemporary, formal academic/policy analysis prose — a critical register bridging official administrative/legal texts and general expository literature without being raw statutory legalese or web crawl.

Archive enumeration & fetch route — verified

Archive URL: https://studiabas.sejm.gov.pl/journal.nsf/pubBAS.xsp?view=12&seria=4&lang=PL
Enumeration: The Lotus Domino / XPages archive portal uses dynamic AJAX updates. Querying issues by year (2021–2025) requires parsing the initial xspForm state ($$viewid, $$ajaxid, cbYear) and submitting multipart POST requests.
PDF Resolution: Issue records point to download endpoints on orka.sejm.gov.pl (either direct issue PDF paths or the download-counting agent LiczOpen?OpenAgent&<UNID>).
WAF / Network: studiabas.sejm.gov.pl is fronted by Imperva WAF, requiring browser-like TLS handshakes (curl_cffi impersonating Chrome) for reliable programmatic querying and retrieval.

License evidence — verified cutoff from issue 2(66)/2021

Historical cutoff: Issues 1–65 (prior to 2/2021) were published under traditional all-rights-reserved copyright.
Open-access grant: Starting with issue 2(66)/2021 (issue 66), the journal officially transitioned to open access under Creative Commons Attribution 3.0 Poland (CC BY 3.0 PL / CC-BY-3.0-PL).
Dual-gate verification: Ingestion enforces a hard issue boundary (issue >= 66) and additionally validates the presence of the explicit CC BY 3.0 PL statement inside every article PDF ("Artykuł został udostępniony na licencji Creative Commons – Uznanie Autorstwa 3.0 Polska (CC BY 3.0 PL)"). Issues or articles lacking this statement fail closed.

PDF characteristics, text extraction & non-OCR status

Born-digital PDFs (is_ocr: False): Whole-issue PDFs feature high-fidelity digital text layers, not scanned or OCR-derived text.
Article segmentation: Multi-article issue PDFs are sliced at article title pages identified via title font size (max_size >= 12), registered DOI (10.31268/StudiaBAS.YYYY.NN), and the in-PDF CC BY license block.
Layout normalization:
- Running headers (page numbering, journal title, ISSN 2082-0658, studiabas.sejm.gov.pl) and author affiliation footnote blocks are stripped from the main body.
- PDF line-break hyphenation is reconstructed for Polish compounds (e.g. przedsię-\nbiorców, rolno-\n-spożywczy).
- Paragraph breaks are preserved and reconstructed using vertical spacing, indentation, and heading font shifts.
Polish scope: Ingestion extracts the Polish title, Polish abstract (Streszczenie), Polish keywords (Słowa kluczowe), and full Polish article body. Incidental English titles/abstracts (Abstract, Keywords) in dual-language front matter are excluded from the main text.

Observed yield (2021–2025)

Extracted articles: 115 articles across issues 66 to 81 (2021–2025).
Retained articles: 112 articles after dropping 3 entirely English-language articles.
Token volume: ~112 long-form scholarly articles yielding ~0.8–1.2M tokens of dense, high-register Polish analytical prose.

Author, affiliation & PII handling

Scholarly attribution: Lead and contributing author names and DOIs are preserved in document metadata (author, doi, source_url) as official academic attribution.
PII scrubbing: Author email addresses, ORCID identifiers (orcid.org/0000-...), phone numbers, and national identifiers (PESEL, NIP, REGON) appearing in contact blocks or bios are systematically scrubbed and replaced with stable [PII] tokens, preserving sentence syntax.

Inclusion rationale & quality/bias caveats

Inclusion rationale: High-quality, editorially reviewed policy research and academic discourse with unambiguous CC BY 3.0 PL licensing, permanent DOI provenance, and zero OCR noise. Fills a major gap in formal analytical Polish.
Quality & bias caveats:
- Institutional focus: Content is oriented toward state policy, parliamentary advisory topics, economics, and legal regulation.
- Complex formatting: Rare multi-column tables, statistical charts, or boxed exhibits in PDFs may yield minor layout artifacts or omitted figure captions.

Cross-source exact dedup — verified 2026-08-31

All 112 released studia_bas texts are internally unique.
Full byte-for-byte text comparison against 4,366,651 documents in all 20 other local Parquets found 0 exact-text matches.
No studia_bas record required removal. A future rebuild of biblioteka_nauki still needs cross-source dedup because its current upstream catalogue exposes 105 of these articles even though the released local shard does not contain them.

open_icm_pl — OPEN ICM open-access publications (pilot)

Investigated — 2026-09-10 (review round 2026-09-28).

Verdict — added (proposed with release: None). 85 publications / 2.36M cl100k_base tokens — a deterministic 100-record pilot of the OPEN ICM repository (ICM UW). Small on purpose: yields must not be extrapolated to the 6,183 eligible items.

Enumeration — official DSpace API (/server/api/core/items), fail-closed license/language gate; repository-generated TEXT bitstreams (MD5-verified) fetched after the origin's interactive CAPTCHA. 100 acquired -> 95 after QA -> 85 after removing 10 exact title/DOI matches against biblioteka_nauki.

License evidence — verified AT THE ORIGIN: per-item dc.rights + dc.rights.uri in the portal's item metadata (example: "Uznanie autorstwa 3.0 Polska" -> creativecommons.org/licenses/by/3.0/pl/ on https://open.icm.edu.pl/items/0078ca93-45a5-4bef-912d-297c069ee6ab, also machine-readable at /server/api/core/items/0078ca93-...); 77 of 85 items additionally ship a per-item license.txt bitstream (checksum-pinned). Allowlist: CC BY / CC BY-SA only (3.0-PL / 3.0 / 4.0); NC and "dozwolony użytek" rejected.

Blockers — none to merge with release: None. Before release: NERGAL pass - the residual spot check FOUND one glyph-garbled contact the patterns missed (JKolbusz @wsiz.rzeszow.pl in open_icm_pl_6ec903ca-...); extraction noise defeats pattern matching, so the model-based scrubber is the real gate. Also cross-source text dedup (the 10 exclusions were metadata-level) and review of quoted third-party passages.

Lesson — (1) pattern scrubbers fail on PDF-extraction glyph noise - always spot-check noisy sources, a "0 residual" pattern count is not evidence; (2) CAPTCHA-gated origins are reproducible only via pinned checksums, not re-fetch - say so in the datasheet instead of pretending a before/after can be regenerated.

`1000_novels` rebuilt as its public-domain subset

Investigated: 2026-10-05 Verdict: keep 837 of 1,000 texts as public-domain; drop 17 in copyright and 146 unresolved.

Verified

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): 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 status (0221, 0292, 0294, 0450). 1000_novels_0981 is Seneca's letter to Lucilius filed under Żuławski, headed "PRZEKŁAD" with no translator named, so it now fails closed.
src/clean_1000_novels.py writes per-row author (with translator), title, death year and basis to 1000_novels.attribution.jsonl. Titles: 790 match the CLARIN file name, 32 are read by position (Inferred), 15 are read by hand.

Not established

Edition rights (Art. 99¹–99²).
The 12 Wolne Lektury rows rest on WL's statement (Inferred).
No dedup yet against wolne_lektury / eltec_pol.

Lesson: a file name is metadata, not evidence. Two texts did not match the work their file name gives, and 4 translator credits were wrong.

wiktionary_examples — Polish Wiktionary usage examples (rebuild)

Investigated — 2026-10-05 (October 2026 rights audit, artifacts/rights_audit_2026-10/reports/wikis_ugc.md).

Verdict — added, rebuilt for v0.2.6. Quotations from third-party works dropped; 1,818 documents / 534,521 characters (v0.2.5: 8,639 / 2,889,761). Licence relabelled CC-BY-SA-3.0 → CC-BY-SA-4.0.

Enumeration — official dump plwiktionary-20261001-pages-articles.xml.bz2 (sha1 a097f6e2…), parsed by src/fetch_wiktionary_examples.py. 116,157 Polish entries; 78,245 example lines, of which 32,664 carry {{źródło}}/<ref> and 319 name a source by hand ("(z Internetu)", "(B. Prus: Lalka)", "(Wikipedia)", Bible references, "(…)"). The audit's sample estimated 23% cited; the full count is 42%.

License evidence — verified: Wikimedia Terms of Use (contributors license their text CC BY-SA 4.0; page sha256 913bdefc…) and the pl.wiktionary footer linking BY-SA 4.0, both read 2026-10-05. Inferred: pre-June-2023 text was under 3.0, whose §4(b) allows an adaptation under 4.0. Quoted lines are used on Wiktionary under Art. 29, which does not pass to a corpus. Evidence in data/wiktionary_examples/wiktionary_examples.license-evidence.json.

Blockers — none to merge. Quotations pasted without a source marker cannot be detected; NERGAL pass before release.

Unmarked-quotation sample — kept (decided 2026-10-05) after a web check of 100 random released lines (seed 20261005, of 7,749): no line was found copied from another source (95% upper bound 3.7%). One line occurs on a 2017 forum post, copied from Wiktionary, where it has been since 2007; one is a paraphrase of a 2014 news quote, not verbatim. Web search cannot see unindexed print. Per line in data/wiktionary_examples/wiktionary_examples.quotation-sample.jsonl.

200-character gate — kept (decided 2026-10-05). It removes 29,265 of the 31,094 entries with a kept example; several headwords per document is not done.

Lesson — (1) a template-only citation filter misses about 1% of quotations, which contributors attribute by hand in parentheses; survey the uncited lines before trusting it. (2) An end-of-section regex for {{przykłady}} must be anchored to line start, or an inline {{tłum}} cuts the section and hides cited lines.

nkjp1m — NKJP 1-million-word subcorpus (licence confirmed)

Investigated — 2026-10-05 (October 2026 rights audit). Held for v0.2.6 until IPI PAN replied to the maintainer's question.

Verdict — added, relabelled CC-BY → CC-BY-4.0. Text unchanged: 18,381 documents; only the license and author columns differ from v0.2.5.

License evidence — verified: an email from IPI PAN for the NKJP Consortium (2026-10-05 22:12, Europe/Warsaw):

the CC BY licence covers the texts as well as the annotations;
agreements were signed with the publishers or authors of all texts, spoken ones included;
the texts may be distributed without annotations.

The version is CC BY 4.0: stated on nkjp.pl (index.php?page=14, read 2026-10-05) and confirmed in the consortium's second reply (2026-10-05). The template in the first email named CC BY 3.0 Polska; another company had proposed it to IPI PAN for its own product. Evidence in data/nkjp1m/nkjp1m.license-evidence.json.

Attribution — data/nkjp1m/ATTRIBUTION.md adapts the formula IPI PAN forwarded, which was written for a model trained on the corpus, to redistribution of the text, and lists the changes made. The consortium agreed to it on 2026-10-05; its licence line was changed to CC BY 4.0 as the same reply suggested.

Blockers — none to merge. NERGAL pass before release.

Lesson — when the download page names a licence without a version ("CC-BY"), check the project's other official pages before asking: nkjp.pl stated 4.0. A template the licensor forwards may have been written by a third party; check its licence version.

openstax_pl — withdrawn (2026-10-05)

Verdict — removed. Never released (release: None, not in the v0.2.5 tag). 1,422 documents / 4,632,374 tokens dropped. The data, the registry entry, the fetcher, its test and requirements, and artifacts/openstax_pl_ontology_manifest.json were removed. A legal-note stub (the HPLT pattern) was not kept, because nothing was released. Git history holds the code if OpenStax Poland ever gives written permission.

Reason — on 2026-10-05, the "Cytowanie i udostępnianie" block on all 8 live /pages/przedmowa pages allowed reuse only for non-commercial purposes. It also barred LLM training without OpenStax Poland's prior written consent. The OpenStax CMS lists all 8 Polish books as CC BY-NC-SA 4.0. The sources contradict each other: the same block labels itself "(CC BY)", and openstax.pl's FAQ and catalogue say CC BY 4.0. Whether the book-level wording was live at our 2026-09-05 fetch is not established; the last archived captures predate it. The audit read this as unresolved and restrictive, which is not enough to ship under a CC BY label. References: rights_audit_2026-10/reports/academic.md L3-9; reports/citations/openstax_cv.md A1-A14 and Corrections 1-5.

Cause — the fetcher read only the foreword body and hard-coded CC-BY-4.0. It never read the licence block, the site footer or the CMS licence field.

Lesson — the licence on the badge or in the body is not the whole grant. Read every rights statement the publisher shows, plus its machine-readable licence field, and stop on any NC, ND or AI-training term (.agents/skills/source-rights-review, check 4).

fandom_pl — withdrawn (2026-10-05)

Verdict — removed, whole source (all 12 wikis). Never released (release: None, not in the v0.2.5 tag, not in the README totals). 50,653 documents / 45,790,009 tokens dropped. The data, the registry entry, src/build_fandom_pl.py, its test and requirements, and artifacts/fandom_pl_ontology_manifest.json were removed. No legal-note stub was kept, because nothing was released. This decision supersedes the earlier audit action "drop 6 wikis, sample 4".

Reason — (1) The Fandom Terms of Use (revision 2025-12-19, read from a Wayback capture of 2026-09-05) bar three things without express written permission: scraping and automated access, bypassing robot-exclusion headers, and using content to train an ML/AI system. The AI clause is not in the 2023 revision and has no carve-out. Whether it overrides the CC BY-SA grant is a legal question we have not resolved. (2) The pages were fetched live from each wiki's MediaWiki API on 2026-10-02, after that revision. The builder has no robots.txt handling. PiotrSty/fandom-pl is the preparer's own repository, not an independent snapshot. (3) In 40-document samples, the wikis carry third-party text under the CC BY-SA label. Affected documents: dragonage 22/40 (game codex, letters, poems; EA/BioWare), elderscrolls 17/40 (full English in-game books; Bethesda), wiedzmin 11/40 (CD PROJEKT RED bars text and data mining). naruto has lyrics pages. pokemon includes Bulbapedia translations, and Bulbapedia is CC BY-NC-SA 2.5. creepypasta has 240 labelled translations of stories whose origins are mostly untraced. (4) All 12 wikis report an unversioned CC-BY-SA in rightsinfo. References: rights_audit_2026-10/reports/wikis_ugc.md L12-31; reports/citations/fandom.md.

Lesson — a site's CC licence for user contributions does not settle how its content may be fetched. Read the host's terms for scraping and AI-training clauses, and the revision date, before a live API crawl. Fan wikis also quote the works they cover; sample them for game text, lyrics and translations before trusting the site licence.

common_voice_pl — withdrawn (2026-10-05)

Verdict — removed, whole source. Never released (release: None, not in the v0.2.5 tag). 45,043 sentences / 994,922 tokens dropped. The data, the registry entry, src/fetch_common_voice_pl.py, its contract test, 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.

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 counts: 24,709 (54.9%) are in selected-europarl-v7-pl.txt, 18,569 (41.2%) are in sentence-collector.txt, 37 are in both, and 1,802 (4.0%) are in neither. The counts reproduce only with NFKC + casefold + removal of all non-word characters (openstax_cv.md B.3). The recipe first stated in the audit gives different numbers.

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 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.

References: rights_audit_2026-10/reports/citations.md L35; reports/citations/openstax_cv.md section B, Corrections 6-13.

Lesson — a dataset-level CC0 is only as good as the per-item provenance behind it. Match the items against the platform's own source files before trusting the label. A public-sector text with an attribution condition cannot be relabelled CC0.

wikiquote — withdrawn (2026-10-05)

Verdict — removed, whole source, from main and from v0.2.6. Released in v0.2.0 and in every tag through v0.2.5, which stay as published. 30,363 documents / 31,896,591 tokens dropped. As with HPLT, the registry keeps its release: "0.2.0" entry, now with a legal_note; make_docs.py skips it (no stats file). plwikiquote was removed from src/fetch_wikimedia.py, so a default run no longer fetches it. The audit's alternative, screening per page for authors and translators who died in 1955 or earlier, was not chosen; its 10-20% yield estimate was not reproduced (Q14).

Reason — most of the text quotes other people's works. CC BY-SA covers the contributors' selection and wording, not the quotes. The quotes rest on the Art. 29 quotation right (t.j. Dz.U. 2025 poz. 24), which is purpose-bound and is not a licence to downstream users; its reach to bulk reuse is inferred, not verified.

Random 200-page sample (seed 1), audit's own labels: 158 in copyright (79%, Wilson 72.8-84.1%), 20 public domain, 13 mixed, 7 translation-only, 2 unknown. The citation pass checked only the arithmetic, not the labelling (Q13).
Example: wikiquote_2429270 is the page on Alice Sebold (b. 1963), quoted in modern translations (Q10-Q12).
The 1,000-character quote cap is a length guideline with exceptions up to 1,800 characters, not a copyright rule (Q1-Q3). No pl.wikiquote copyright-policy page was found (Q8).

References: rights_audit_2026-10/reports/wikis_ugc.md L1-4; reports/citations.md L11; reports/citations/wikimedia.md section 3 (Q1-Q14); sample in rights_audit_2026-10/wikis/wq_sample.json.

Lesson — a project licence covers what the project's contributors wrote. On a 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 tag through v0.2.5, which stay as published. 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 its upstream is now pl.wikisource.org rather than the dumps it never came from. count_vocab.py and sample_quality.py no longer list it. There was no fetch script to remove: the data was a SpeakLeash pull.

Reason — all rows carry one CC-BY-SA-3.0 label, and the card said the underlying texts are public-domain works. Wikisource's own guidance says "mostly" (W1).

Among the 12 audited rows, several are in-copyright post-war imprints that their rights holders released through consent programmes or OTRS, each page under its own licence template: PIW 1971 and 1972, Czytelnik 1962 and 1966, WL 1974, a 1995 sailing manual, 1990-91 regional press (W11-W21). The templates differ (CC BY 3.0 PL, CC BY-SA 3.0 PL, CC BY-SA 3.0, GFDL, CC BY 4.0), so no single relabel fixes them (W17).
One row carries a translator's footnote for a translation still protected until 2045; the row-to-page mapping is inferred (W8-W10).
The rows keep neither the consent, the licence version nor an author (W22). The SpeakLeash selection is undocumented and rows have no title or page id, so the rows cannot be screened afterwards. The audit's crude filter (372 rows, 1.7%) was not reproduced; the in-copyright share is unknown (W23).

References: rights_audit_2026-10/reports/citations.md L10; reports/citations/wikimedia.md section 2 (W1-W23); REVIEW.html (wikisource block).

Lesson — a project-wide label on an aggregator pull hides per-page licences. Pull from the official dump with page ids and licence templates, so each page can be 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 → 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 fails on any other value. recover_attribution.py no longer counts public-domain as clean.

Reason — the 3 rows (books_2142716, books_2147851, articles_2097363) carry the platform's "Domena publiczna" label. Verified 2026-10-06 against bibliotekanauki.pl/api/{type}/{id}: licence id 22, creativeCommons: false, URL creativecommons.org/publicdomain/mark/1.0/. Creative Commons describes the Mark as "a tag or a label" for works already free of copyright; it grants nothing. All three are original texts (reprintPublication and translationPublication false) first published 2019, 2021 and 2022 by named authors (Jerzy Zając, Witold Rakowski, Beata Michalec), so life+70 has not run out. The 2019 book's own colophon reads "Copyright by Jerzy Zając 2019 / Copyright by Muzeum Niepodległości w Warszawie 2019". That the authors are alive is inferred.

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 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 reserved" in their text (publication years 2010-2021); regex count only, not reviewed.

Lesson — a platform's "public domain" label can be the PD Mark. Check the URL behind the label, and check the date: a 2019 work cannot be public domain by term in Poland.

govpl — PAP reprints dropped (2026-10-06)

Verdict — kept, minus 188 PAP reprints (slayerlabs/polish-dynaword-v2#51). 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. fetch_govpl.normalize() applies the same rule (pap_rule) to future crawls.

Reason — gov.pl's copyright page grants CC BY-SA 4.0 for its text but carves out works whose copyright belongs to third parties; their use needs the owners' consent. Articles that credit PAP as their source are such works.

The issue's rule (Źródło: line naming PAP) matches 154 rows. A hand-check of 20 (seed 51) found 20 text source lines and no photo credits; 14 credit PAP jointly with the ministry, so the text may be partly the ministry's. PAP's archive is not public, so verbatim PAP authorship was not confirmed.
A scan of the other rows added 34 verified text attributions: misspelt or colon-less source lines (9), Na podst. … PAP (22, all uw-podlaski), a lone (PAP) sign-off line (2) and one PAP interview transcript (prokuratura-krajowa). Photo credits (fot. PAP) and rows that quote or mention PAP stay.
Left in, flagged: a mid-text (PAP) after one paragraph (govpl_30227, govpl_58869), a footnote crediting PAP and the health ministry (govpl_81037), a KRRiT report quoting two PAP questions (govpl_86914).
Not reviewed: 710 kept rows carry a Źródło: line naming some other source (mostly government bodies; some name WHO, universities, journals).

References: rights_audit_2026-10/reports/govweb.md (govpl block); https://www.gov.pl/web/gov/prawa-autorskie (read 2026-10-06).

Lesson — an open-licence footer covers the publisher's own text. On a news-style site, check source lines for agency copy before taking the footer at face value.

samorzad_gov_pl — third-party slices dropped (2026-10-06)

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 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 authors. All 69 hand-checked.
9 IMGW-PIB notices carrying "Opracowanie niniejsze i jego format…", which forbids changes and requires naming IMGW-PIB. All 9 hand-checked.
66 articles whose Źródło: line credits a non-governmental source (Wikipedia, media, associations including volunteer fire brigades, foundations, clubs, rural housewives' circles, companies, named books or brochures), unless a photo or film marker precedes it on the line. All 66 hand-checked as third-party text. The rule was revised on the same matches (first pass 61 of 67 correct), so precision is in-sample and recall is unknown; one known miss is a Fundacja Santander credit. Part of the regex lists work titles found in that review.
Not dropped: contest entries. The issue's 112 is not reproducible; the closest regex (awarded/winning/distinguished works) gives 111 documents, and a 40-document sample (seed 49) contains no entry text: they are results announcements. Only 3 documents publishing entries were found; most entries are images, not text.

Reason — gov.pl's copyright page says the service also carries third-party works whose use requires the owners' consent; the footer licence cannot cover them. Credits matter here: 810 documents have an Źródło: line, but in 405 every credit is 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; 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).

Lesson — de-identification that replaces URLs also removes the provenance needed to screen reposts. Record the credited domain before masking URLs.

global_voices — partner republications dropped (2026-10-06)

Verdict — admitted with 114 posts withheld (polish-dynaword-v2 #50). 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 checks that no shipped row matches the rule.

Reason — Global Voices licenses "content created by Global Voices" under CC BY 3.0. A post that republishes a partner's article under a content-sharing agreement or with permission stays the partner's work.

Where the rule looks. It reads only the first 700 and last 350 characters, where the editorial note sits. A rule fires only on a sentence about the post itself that does not name Global Voices, so news reporting ("raport opublikowany przez UNESCO") and GV's own projects do not count. Sentences whose subject is an image are skipped too.
Rule counts (a post can match several):
- originally: 69
- agreement: 56
- published_by: 51
- partnership: 11
- permission: 5
- republished: 5
- written_for: 3
- non_commercial: 3
- listed by hand: 3 (21466, 27005, 32352)
Hand check.
- Every one of the 114 drops was read, and the English original was consulted where the note was unclear.
- 25 rule matches are kept by hand with a stated reason:
  - 12 cross-posts from the author's own blog;
  - 7 posts GV wrote for the ISN blog;
  - 3210, a GV post that another outlet republished;
  - 2 notes dating GV's own English original;
  - 3 news mentions.
- All 43 posts where the gating removed a raw match were checked. They are image credits, news and GV projects.
- 15 partner-organisation accounts with no republication note (e.g. 350.org 22567, Metamorphosis 33071) were checked against the English page and kept.
Outlets.
- PRI/The World 21
- The Stand News 8
- Hong Kong Free Press 7
- Agência Pública 6
- Metamorphosis/Meta.mk 5
- 350.org 5
- Eurasianet 3
- Freemuse 3
- The Irrawaddy 3
- 2 each: Balkan Diskurs, Ensia, Kurtural, Loa, MigraMundo, NACLA, Nepali Times/Himalkhabar, Premium Times, Syria Untold, Teplitsa, Transitions
- 1 each: 88 Bar, APC.org, Afroféminas, Azerbaijan Internet Watch, Baynana, Bumaga, CHRI, Cari-Bois, Chaikhana, Common Dreams, Connectas, Doha News, Fundación MEPI, GCHR, Hive-Mind/EJO, Iran Voices, New America Media, OC Media, Periodismo de Barrio, Promundo, Russia Post, Sin Embargo, Syria Deeply, TJournal, Tea After Twelve, The Isaan Record, The Moscow Times, VideoVolunteers, WITNESS, inmediahk.net, openDemocracy.
Delta against the audit's 64. The audit used outlet names plus an agreement regex, and it recorded neither the list nor the script. Its "17 PRI" is an undercount: all 21 posts credited to PRI/PRX's The World carry a republication note. Most of the remaining difference is outlets the audit's name list did not cover, such as The Stand News (8) and Agência Pública (6). The Hub PR has per-document evidence.
Borderline; dropped and flagged for review.
- Freemuse ×3: the note allows republication "by non-commercial media" only.
- Bumaga 30003: a shorter version of the author's article for Bumaga.
- Hive-Mind/EJO 34820.
- openDemocracy 29544: "under a Creative Commons licence", version not stated.
- Transitions 34277: "powstał dzięki umowie partnerskiej", possibly a co-production.
- 32352: built on two Meta.mk texts.

Lesson — the licence covers what the publisher created. A per-post editorial note is the only signal of a partner's work, so the filter has to read the note and not only the byline.

`1000_novels` rescreened: unknown authors fail closed, translator death years cited

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.

Verified

Screen defect: a known public-domain translator cleared an author whose death year was unknown. 1000_novels_0733 (author "Roux", unidentified) shipped in the 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, 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 d. 1922, Stefan Frycz d. 1960.
"tłum. A. W." on 29 Chekhov texts is Antoni Lange (d. 1929): BN catalogues their source edition (Śmierć urzędnika, Bibljoteka Groszowa 1926) with "nazwa właśc. tł.: Antoni Lange".
1000_novels_0981 (Seneca's letter, "PRZEKŁAD", no name) is Żuławski's own translation in his 1914 collection Bajka o człowieku szczęśliwym. It is now credited to Seneca with Żuławski as translator.
"Roux" is Saint-Pol-Roux (d. 1940), so 0733 stays in on a cited death year.
1000_novels_0462 (Wola mocy): BN credits Stefan Frycz (d. 1960) as co-translator. It is dropped although its Wolne Lektury footer says public domain. The screen's footer cross-check allowlists it by name.
Net: 53 unresolved and 10 Wolne-Lektury-footer rows move to public domain by term; 1 Wolne-Lektury-footer row moves to in copyright.

Not established

"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).

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.

Update (2026-10-06, maintainer decision): the screen no longer keeps a row on a Wolne Lektury public-domain footer. "A. L." (1000_novels_0734) now fails closed as unresolved, so the release keeps 887 texts and drops 18 in copyright and 95 unresolved. src/clean_1000_novels.py pins the screen CSV by SHA-256 (7fefc75f8fcbd07e5ec98842276765ffef9520b6acb4e24b4751c67cc3d2081c, also in 1000_novels.license-evidence.json).

wolne_lektury — eltec_pol duplicates dropped (2026-10-06)

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 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 cut first (no WL row has one). Containment is used, not Jaccard, because WL splits novels into tomes: Chłopi has 4, Faraon, Nad Niemnem and Stara baśń have 3.
Threshold 0.9, in an empty stretch: no document scores between 0.7588 and 0.9636. All 63 documents above it pair with an ELTeC novel by the same author and title (Żuławski's two WL documents are headed with the trilogy's title).
Hand-checked: all 3 documents between 0.3 and 0.9, the 3 between 0.9 and 0.99, and 10 sampled at ≥ 0.99 (seed 72). In 12 of the 13 drops checked, the unmatched shingles form short scattered runs of at most 27 words (title head, spelling variants). The exception is Dzieje grzechu (0.9879): its WL copy ends with a "Dodatek" of draft chapters, about 1,800 words that ELTeC lacks. It is dropped with the novel.
Kept and listed as keep_partial: both WL tomes of Mniszek's Trędowata (0.7588 and 0.6473 against POL0051). The ELTeC text is shorter: about 121,000 words against WL's 162,000. 95% of the ELTeC shingles are in WL's tomes, and the WL text missing from ELTeC comes in runs of up to 5,418 words. The shared part stays duplicated across the two sources. Also kept: Kochanowski's Pieśń IX from Pieśni, księgi wtóre (0.4267), which Łoziński's Oko proroka quotes.
Below 0.3 (not listed, not dropped): 6 documents at 0.10–0.22, all poems of at most 2,321 tokens by Mickiewicz, Kochanowski and Miciński; their overlap with the paired novels was not inspected. Nothing scores between 0.2203 and 0.4267.

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.

References: polish-dynaword-v2 #72 (decision), #45 (method); 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.

Source investigation findings

How to add an entry

zpe.gov.pl — Zintegrowana Platforma Edukacyjna

Enumeration — solved

License evidence — the blocker

Consequences for design

Extraction hazards (seen in a single sampled page)

What would unblock this

Lesson

ParlaMint-PL — parliamentary debates 2015–2022

Size — measured, not guessed

Overlap with the existing parliamentary shard — the real issue

Version note

Verified

Why it is still worth doing

Drift found while checking

Sejm API — interpelacje and zapytania poselskie

What happened

Unverified — what to check on retry

Why it is on the list

CLARIN-PL DSpace (PELCRA Spokes, and the 1000_novels upstream)

Verified

Side effect on shipped data

Expected concern when it returns

NIK — Najwyższa Izba Kontroli audit reports

License evidence — verified

Enumeration — unsolved

Verdict rationale

Billion-scale survey — what can actually move the totals

SAOS — biggest clean win, already half-built

FineWeb-2 Polish — only billion-scale contemporary prose

Dedup is not the bottleneck — this was checked, not assumed

The actual blocker: the HPLT gate cannot be ported

epfml/FineWeb2-HQ — the better target

quality_score → token yield

Recommended operating point

Work order if this is built

PleIAs/Polish-PD — good on paper, rough in practice

HPLT is exhausted — correction to an earlier assumption

Conclusion

common_voice_pl — Common Voice Polish validated sentences (v26.0)

sejm_interpellations — 10th-term written Q&A (interpellations + replies)

Edukacja Medialna — media-literacy lessons (edukacja_medialna_pl)

ted_pl — TED Polish public-procurement notices

kultura_bez_barier_pl — cultural-accessibility guides (FKzB)

plwiki_talk — Polish Wikipedia talk pages (ns 1+5)

plwiki_user_talk — Polish Wikipedia user talk, ns 3

open_agh_chemistry_pl — Open AGH chemistry textbooks

openstax_pl — OpenStax Poland textbooks (8 volumes)

wiadomosci_statystyczne_pl — Wiadomości Statystyczne / The Polish Statistician (GUS)

sejm_committee_transcripts — Sejm committee full transcripts (terms 9–10)

Studia BAS — Kancelaria Sejmu research & public-policy journal (2021–2025)

Archive enumeration & fetch route — verified

License evidence — verified cutoff from issue 2(66)/2021

PDF characteristics, text extraction & non-OCR status

Observed yield (2021–2025)

Author, affiliation & PII handling

Inclusion rationale & quality/bias caveats

Cross-source exact dedup — verified 2026-08-31

open_icm_pl — OPEN ICM open-access publications (pilot)

1000_novels rebuilt as its public-domain subset

Verified

Not established

wiktionary_examples — Polish Wiktionary usage examples (rebuild)

nkjp1m — NKJP 1-million-word subcorpus (licence confirmed)

openstax_pl — withdrawn (2026-10-05)

fandom_pl — withdrawn (2026-10-05)

common_voice_pl — withdrawn (2026-10-05)

wikiquote — withdrawn (2026-10-05)

wikisource — withdrawn (2026-10-05)

biblioteka_nauki — public-domain-label records dropped (2026-10-06)

govpl — PAP reprints dropped (2026-10-06)

samorzad_gov_pl — third-party slices dropped (2026-10-06)

global_voices — partner republications dropped (2026-10-06)

1000_novels rescreened: unknown authors fail closed, translator death years cited

Verified

Not established

wolne_lektury — eltec_pol duplicates dropped (2026-10-06)

Overlap with the existing `parliamentary` shard — the real issue

CLARIN-PL DSpace (PELCRA Spokes, and the `1000_novels` upstream)

`epfml/FineWeb2-HQ` — the better target

Edukacja Medialna — media-literacy lessons (`edukacja_medialna_pl`)

`1000_novels` rebuilt as its public-domain subset

`1000_novels` rescreened: unknown authors fail closed, translator death years cited