forms: form-183
This data as json
| id | name | damage | class | injection | layer | status | organisation | campaign | cancellable_by_code | reachable_by_deletion | judgeable_by | observations | observed_absent_in | name_fr | legacy_line | version | record |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| form-183 | Half of the in-force acts have no structured text in the official dump | MISSING | SILENT_FALSE | DELETE | ② reading | proposed | Loxyn SAS, Lyon | September 2026 — EUR-Lex collection for the Answer Key (Alexandrie) | 0 | no | registry | 1 | 1 | {"id": "form-183", "version": 1, "name": "Half of the in-force acts have no structured text in the official dump", "origin": {"organisation": "Loxyn SAS, Lyon", "campaign": "September 2026 — EUR-Lex collection for the Answer Key (Alexandrie)", "contact": "contact@loxyn.ai"}, "damage": "MISSING", "class": "SILENT_FALSE", "layer": "② reading", "injection": "DELETE", "status": "proposed", "seen": [{"corpus": "EUR-Lex data dump of the Publications Office, 2026-09-06 (LEG_MTD 58,686 works; LEG_EN_FMX 123,983 files; LEG_FR_FMX 121,500 files), inventoried 2026-09-09", "document": "27,792 of 58,686 works without any Formex file; 22,293 .tif files on the EN side", "date": "2026-09-09", "observer": "Alexandrie (LLM agent, collection session), from the EUR-Lex collection manifest", "organisation": "Loxyn SAS, Lyon", "excerpt": "Inventaire du dump du 06-09-2026, mesuré le 09-09-2026 en lecture de flux : 58 686 œuvres au MTD, 30 047 avec un texte Formex EN, 29 892 avec un texte FR ; 27 792 œuvres (47 %) sans aucun Formex — côté EN, 22 293 fichiers .tif, des actes anciens scannés jamais passés en XML. Un dump de métadonnées complet et un dump de textes à moitié.", "excerpt_en": "Inventory of the 2026-09-06 dump, measured 2026-09-09 by streaming the archives: 58,686 works in the MTD, 30,047 with an EN Formex text, 29,892 with FR; 27,792 works (47 %) without any Formex — on the EN side, 22,293 .tif files, old scanned acts never converted to XML. A complete metadata dump and a half text dump.", "lang": "fr", "probe": "probes/2026-09/form-183-dump-coverage.py"}], "specimens": {"cases": [], "counter_examples": []}, "prevention": {"cancellable_by_code": false, "note": "a missing text cannot be produced by code; it can be declared"}, "repair": {"reachable_by_deletion": "no", "note": "deletion cannot restore this; classified by judgment, not measured — the EUR-Lex loop will measure it"}, "judgeable_by": ["registry"], "related": ["form-045", "form-097"], "history": [{"date": "2026-09-09", "event": "proposed from the EUR-Lex collection reports of 2026-09-09; effect on a graph: A graph built from the dump's texts covers half of the acts in force and silently lacks the other half; the metadata says an act exists, no text says what it contains. Same family as scanned pages, observed here on an official registry. Prevalence: measured: 27,792 / 58,686 (47 %).", "by": "Alexandrie (LLM agent, collection session), from the EUR-Lex collection manifest"}, {"date": "2026-09-09", "event": "converted to a record; class set from the collector's own refusal clause where one exists; to be reviewed by a second reader", "by": "Fault Atlas"}, {"date": "2026-09-09", "event": "exact probe attached and re-run: probes/2026-09/form-183-dump-coverage.py", "by": "Fault Atlas maintainer"}], "probes": [{"file": "probes/2026-09/form-183-dump-coverage.py", "kind": "python", "corpus_id": "corpus-2026-09-eurlex", "written": "2026-09-09", "run": "python3 probes/2026-09/form-183-dump-coverage.py <dir holding the three dump .zip>", "result": {"date": "2026-09-09", "count": 27792, "of": 58686, "by": "Fault Atlas maintainer (Loxyn), on the SSD copy of the dump"}, "matches_excerpt": true, "note": "Re-run on the 2026-09-06 archives (stream read, nothing extracted): 58,686 works; 30,047 with EN Formex; 29,892 with FR; 27,792 without any (47 %); 22,293 .tif in the EN archive. Same figures."}]} |