sejm_interpellations data files (scoped fix for #24)

artifacts/source_findings.md

@@ -541,3 +541,36 @@ 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.
+
+---
+
+## 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.

data/sejm_interpellations/sejm_interpellations.md

@@ -0,0 +1,120 @@
+# sejm_interpellations
+
+Interpelacje i zapytania pisemne z 10. kadencji Sejmu, wraz z odpowiedziami
+ministerstw dostępnymi jako HTML.
+
+## Dataset description
+
+- **Source (upstream):** https://api.sejm.gov.pl/
+- **Source documentation:** https://api.sejm.gov.pl/sejm.html
+- **Domain:** political/written
+- **Language:** Polish (`pl`)
+- **License:** `public-domain (official documents)`
+- **Created (range):** 2023-11-15, 2026-09-03
+- **Added:** 2026-09-05
+- **OCR:** no
+
+## Licensing — traceable basis
+
+Official parliamentary materials are excluded from copyright under Polish
+Copyright Act art. 4 pkt 2
+(https://eli.gov.pl/api/acts/DU/2025/24/text/O/D20250024.pdf). Reuse of
+public-sector information is governed by the Polish Open Data Act arts. 2(12),
+5, 14 and 17 (https://eli.gov.pl/api/acts/DU/2023/1524/text.html). Same basis
+as the shipped `sejm_api` shard. No Creative Commons grant is asserted.
+
+## What the texts are
+
+Written questions from MPs to ministers (`interpellations`,
+`writtenQuestions`) and the ministry replies that the API serves as HTML.
+Spoken floor speeches are **not** included — those are `sejm_api` (2023–) and
+[#18](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/18)
+(2011–2022). Same kadencja, different genre.
+
+PDF-only replies (`onlyAttachment: true`) and keyless prolongation stubs are
+skipped. Term 9 listed `X-Total-Count: 0` on both collections at fetch time
+and is not in this shard. Terms 7–8 were not fetched (PPC already covers
+interpellations through 2019).
+
+## Provenance and reproducibility
+
+```bash
+python3 src/fetch_sejm_interpellations.py --out ~/speakleash --terms 9,10 --workers 4
+python3 src/build_dynaword.py --sources sejm_interpellations --speakleash-dir ~/speakleash --out .
+```
+
+`src/fetch_sejm_interpellations.py` pulls HTML `/body` endpoints, strips
+header chrome (nr / recipient / title / signatory / date) into `meta`, then
+runs pattern PII redaction (the `scrub_pii` pass from #20 at build time —
+phones → `[Telefon]`; e-mail / PESEL / NIP / REGON / IBAN → `[PII]`). That
+in-repo scrubber was removed in #38; PII now goes through
+[SlayerLab/NERGAL](https://huggingface.co/SlayerLab/NERGAL) and **this shard
+must be re-run through NERGAL before it is admitted to a release** (it quotes
+constituency casework naming private persons). Names of public officials stay.
+`build_dynaword.py` applies the usual gates (strip, `< 200` chars, Polish
+diacritic ratio, exact sha1).
+
+## Why include it
+
+The corpus already has spoken Sejm (`parliamentary`, `sejm_api`). This adds
+the written Q&A register for the current term: longer, formal, ministry-voiced
+replies that do not appear in the speech shards — the argumentative
+question/answer pair is a document type the corpus otherwise lacks entirely.
+Caveats for downstream users: single term (10th, 2023-11-15 – 2026-09-03), and
+interpellations quote constituency casework that can name private citizens (see
+Privacy below — a NERGAL pass is a release gate).
+
+## Statistics
+
+| Measure | Value |
+|---|---:|
+| JSONL rows read | 26,503 |
+| Exact duplicate texts removed | 553 |
+| Documents published | 25,950 |
+| Characters | 78,607,831 |
+| Tokens (`cl100k_base` proxy) | 28,180,895 |
+| Documents with author attribution | 25,950 |
+
+JSONL mix before dedup: 19,413 interpellations, 3,889 written questions,
+2,730 interpellation replies, 471 written-question replies. All term 10.
+
+## Filters applied (build_dynaword.py)
+
+- drop documents < 200 chars: **0**
+- drop non-Polish (diacritic ratio): **0**
+- exact sha1 dedup: **553**
+- OCR alpha-ratio: **0**
+- read 26,503 → kept 25,950
+
+## Sample
+
+Question (author `Daniel Milewski`):
+
+```
+Szanowny Panie Ministrze,
+
+latem 2022 roku na podlaskiej części granicy z Białorusią ukończono zaporę
+fizyczną, czyli 5,5-metrowy płot ze stalowych przęseł zwieńczony drutem
+żyletkowym. Ogrodzenie stoi na 186 kilometrach. […]
+```
+
+Reply (author `sekretarz stanu w Ministerstwie Spraw Zagranicznych
+Władysław Teofil Bartoszewski`):
+
+```
+Szanowny Panie Marszałku,
+
+w odpowiedzi na zapytanie poselskie nr 3472 pana posła Bronisława Foltyna
+w sprawie Instytutu Polskiego w Szwajcarii, uprzejmie informuję, co
+następuje. […]
+```
+
+## Privacy and limitations
+
+Regex PII scrub at fetch time is not a guarantee of zero personal data.
+Constituency casework in interpellations can name private persons; those
+names are left in (officials acting in capacity are not treated as PII).
+Downstream users must review applicable privacy and source terms. Per
+AGENTS.md the shard must be re-run through
+[SlayerLab/NERGAL](https://huggingface.co/SlayerLab/NERGAL) before release;
+merge with `release: None` is fine meanwhile.

data/sejm_interpellations/sejm_interpellations.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a309b9300933ee772d3c8fcda7546461cbf072807ea37939b6101fc2eb33bead
+size 27168758

data/sejm_interpellations/sejm_interpellations.stats.json

@@ -0,0 +1,16 @@
+{
+  "read": 26503,
+  "kept": 25950,
+  "drop_short": 0,
+  "drop_lang": 0,
+  "drop_dup": 553,
+  "drop_ocr": 0,
+  "chars": 78607831,
+  "tokens": 28180895,
+  "licenses": {
+    "public-domain (official documents)": 25950
+  },
+  "authors_with_value": 25950,
+  "secs": 1.8,
+  "license": "public-domain (official documents)"
+}