QA-03: Prüfgate Klassifizierung & Extraktion (AI) #348

Closed
patrick wants to merge 0 commits from feature/qa-03-pruefgate-klassifizierung-extraktion into main
pull from: feature/qa-03-pruefgate-klassifizierung-extraktion
Owner

Prüfgate für Klassifizierungs-/Extraktionsgüte gegen Referenzkorpus (QA-01) und Korrektur-Feedback-Schleife (baut auf CLS-02/03/04/05, EXT-05).

Ergebnis (echte Inferenz gegen office.sysops.de:11435, gpt-oss:20b):

  • AK1 Klassifizierungsgenauigkeit: 5/5 = 100% ≥ Zielwert 80% -- erfüllt
  • AK2 Extraktionsgenauigkeit: Datum 5/5=100%, Betrag-Vorhandensein 4/5=80% (exakt am begründeten Zielwert 80%) -- erfüllt
  • AK3 Korrektur-Feedback wirkt auf künftige Vorschläge: Vorschlag ändert sich nach 6 Korrekturen von "kuendigung" (Konfidenz 0,86) zu "vertrag" (Konfidenz 1,00) -- erfüllt

Voller Regressionslauf go test ./... im ai-Modul grün.

Protokoll: ai/docs/QA-03-PRUEFPROTOKOLL.md.

Kein Merge, kein Deploy.

🤖 Generated with Claude Code

Prüfgate für Klassifizierungs-/Extraktionsgüte gegen Referenzkorpus (QA-01) und Korrektur-Feedback-Schleife (baut auf CLS-02/03/04/05, EXT-05). **Ergebnis (echte Inferenz gegen office.sysops.de:11435, gpt-oss:20b):** - AK1 Klassifizierungsgenauigkeit: 5/5 = 100% ≥ Zielwert 80% -- erfüllt - AK2 Extraktionsgenauigkeit: Datum 5/5=100%, Betrag-Vorhandensein 4/5=80% (exakt am begründeten Zielwert 80%) -- erfüllt - AK3 Korrektur-Feedback wirkt auf künftige Vorschläge: Vorschlag ändert sich nach 6 Korrekturen von "kuendigung" (Konfidenz 0,86) zu "vertrag" (Konfidenz 1,00) -- erfüllt Voller Regressionslauf `go test ./...` im `ai`-Modul grün. Protokoll: `ai/docs/QA-03-PRUEFPROTOKOLL.md`. Kein Merge, kein Deploy. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
patrick added 1 commit 2026-10-17 10:59:59 +02:00
QA-03: pruefgate klassifizierung & extraktion
CI / DMS-Modul (Lint/Build/Test/Coverage/Artefakt) (push) Canceled after 0s
CI / DMS-Modul (Lint/Build/Test/Coverage/Artefakt) (pull_request) Canceled after 0s
Core-Schnittstellen-Vertragstests / contract-tests (pull_request) Canceled after 0s
CI / DMS Security-Test-Gate (AC5, Merge-Gate) (pull_request) Canceled after 0s
CI / Connect Security-Test-Gate (QA-01 AC4, Merge-Gate) (pull_request) Canceled after 0s
CI / Core-Modul (Build/Vet/Test) (push) Canceled after 0s
CI / DMS Security-Test-Gate (AC5, Merge-Gate) (push) Canceled after 0s
CI / Connect Security-Test-Gate (QA-01 AC4, Merge-Gate) (push) Canceled after 0s
CI / DMS-Modul Supply-Chain-Scan (govulncheck) (push) Canceled after 0s
CI / Web (dms-browser) (push) Canceled after 0s
CI / Web (pdf-viewer) (push) Canceled after 0s
CI / Web (shl) (push) Canceled after 0s
Supply-Chain-Scan (Go) / govulncheck (push) Canceled after 0s
AI-Pflichttest-Gate / pflichttest-gate (pull_request) Canceled after 0s
CI / Core-Modul (Build/Vet/Test) (pull_request) Canceled after 0s
CI / DMS-Modul Supply-Chain-Scan (govulncheck) (pull_request) Canceled after 0s
CI / Web (dms-browser) (pull_request) Canceled after 0s
CI / Web (pdf-viewer) (pull_request) Canceled after 0s
CI / Web (shl) (pull_request) Canceled after 0s
Supply-Chain-Scan (Go) / govulncheck (pull_request) Canceled after 0s
Supply-Chain-Scan (npm) / npm-audit (pull_request) Canceled after 0s
Core-Pflichttest-Gate / pflichttest-gate (pull_request) Canceled after 0s
RBAC-Rollenhierarchie-Pflichtgate / rbac-policy-invariant (pull_request) Canceled after 0s
CI / Web-Barrierefreiheitspruefung (axe, Tastatur, Reflow, Textabstaende) (push) Canceled after 0s
CI / Web-Barrierefreiheitspruefung (axe, Tastatur, Reflow, Textabstaende) (pull_request) Canceled after 0s
8eb07c1797
Neues Paket ai/internal/qa03gate misst Klassifizierungs- und
Extraktionsgenauigkeit am QA-01-Referenzkorpus real gegen den externen
Ollama-Server und belegt die Wirkung von Korrektur-Feedback auf
kuenftige Vorschlaege (historysuggest), ohne bestehende Pakete zu
aendern. Ergebnisse in ai/docs/QA-03-PRUEFPROTOKOLL.md.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Author
Owner

Inhalt bereits in main enthalten (geprueft per merge_all_prs.sh: leerer Merge-Diff). Geschlossen ohne Merge.

Inhalt bereits in main enthalten (geprueft per merge_all_prs.sh: leerer Merge-Diff). Geschlossen ohne Merge.
patrick closed this pull request 2026-10-19 12:30:00 +02:00

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: scripte/nexarch#348