{"_id": {"$oid": "69d4eb12b877594bc5718cc9"}, "test_id": "T0697", "benchmark": "fraktur_adverts", "date": "2026-03-17", "tags": {"document-type": ["book-page"], "writing": ["printed"], "century": [18, 19], "language": ["de"], "layout": ["prose"], "script-style": ["fraktur"], "task": ["transcription"]}, "contributors": ["ina_serif", "maximilian_hindermann"], "hidden": false, "config": {"provider": "openai", "model": "gpt-5.3-codex", "dataclass": "Document", "temperature": 0.0, "role_description": "You are a historian with keyword knowledge", "prompt_file": "prompt_optimized.txt", "rules": "{\"api_style\":\"responses\"}", "legacy_test": false}, "prompt": "## IDENTITY AND PURPOSE\n\nYou are an OCR and information extraction system trained to process historical newspaper pages printed in 18th-century German using Fraktur type. The pages contain mostly classified advertisements. Your task is to identify and extract each advertisement *exactly as printed*, including historical spellings, typographic errors, punctuation, and formatting.\n\n\n## INSTRUCTIONS\n\n- Extract **all advertisements** from the input image, one after the other, following the sequence on the page.\n- Maintain the **original spelling**, capitalization, and any **typos or non-standard forms**.\n- Follow these transcription rules: \n  - the long s (\u017f) is transcribed as \"s\"\n  - \"/\" is transcribed as \",\"\n- Use the masthead of the newspaper only to extract the date, ignore other content.\n- The layout is typically **two-column**; extract ads from both columns, including the ad number.\n- Return the result as a **JSON object** in the specified format and **nothing else** (no explanations, summaries, or additional text).\n- For each advertisement, include:\n  - `\"date\"`: the publication date of the page in ISO 8061 format (YYYY-MM-DD)\n  - `\"tags_section\"`: the heading under which the advertisement appears\n  - `\"text\"`: the full advertisement text\n\n## EXAMPLE OUTPUT\n\n{\n  \"advertisements\": [\n    {\n      \"date\": \"1731-01-02\",\n      \"tags_section\": \"Es werden zum Verkauff offeriert\",\n      \"text\": \"5. Ein kleines, jedoch listiges Lehrbuch der Zauberkunst, lange im Gebrauche des jungen Bartolomeus Simpson.\"\n    },\n    {\n      \"date\": \"1731-01-02\",\n      \"tags_section\": \"Es werden zum Verkauff offeriert\",\n      \"text\": \"6. Ein rarer, mit Edelsteinen besetzter Saxophon-Kasten, aus dem Besitze der Jungfer Lisa Simpson.\"\n    },\n    {\n      \"date\": \"1731-01-02\",\n      \"tags_section\": \"Es werden zu Entleihen begehrt\",\n      \"text\": \"7. Ein gar prachtvoller, jedoch etwas zerlesener Band mit Rezepten von Margaretha Simpsonin.\"\n    }\n  ]\n}\n", "results": {"text": "", "model": "gpt-5.3-codex", "provider": "openai", "finish_reason": "error", "usage": {"input_tokens": 0, "output_tokens": 0, "total_tokens": 0}, "duration": 9.796094417572021, "timestamp": "2026-03-17T22:12:23.850318", "score": {"fuzzy": 0.0, "cer": 1.0}}, "scoring": {"fuzzy": 0.57, "cer": 0.442, "cost_summary": {"total_input_tokens": 7134, "total_output_tokens": 4159, "total_tokens": 11293, "input_cost_usd": 0.0124845, "output_cost_usd": 0.058226, "total_cost_usd": 0.07071050000000001}}, "normalized_score": 55.8}