{
  "title": "Specialty canning data for AI closure-inspection records",
  "source_page": "https://fiund.com/industries/specialty-canning-closure-evidence-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "Before- and after-adjustment gauge readings must not be averaged into one condition",
    "context": "An original inert container fixture uses abstract inspection regions and practice units. It contains no closure parameter, process instruction, real inspection limit or food decision.",
    "records": [
      [
        "SCC1",
        "Matched sample and image",
        "Lot L, sample S, readable source label. Original image A links S to region R before abstract setup change C."
      ],
      [
        "SCC2",
        "Measurement records",
        "S/R before C reads 8 practice units. Later matched S/R after C reads 9. Distinct condition labels and exact sample links preserved."
      ],
      [
        "SCC3",
        "Loose image",
        "Image B has handwritten time near the second reading but no sample or lot ID. No independent review linking it to S."
      ]
    ],
    "question": "Which readings can be attributed to L/S, and should the evaluator average them or assign the loose image to the later condition?",
    "answer": "SCC1 and SCC2 support eight units before C and nine after C for the same named practice region. Keep them separate because the condition changed; an average would obscure the distinction the task asks for. SCC3's nearby handwritten time does not establish sample identity or support an S association. Report the two labeled measurements and unresolved image without claiming closure adequacy for this item or any batch.",
    "failureModes": [
      "Averaging readings from different setup conditions as if they were repeated under one basis.",
      "Assigning an unlinked image from timestamp proximity or treating one observation as a batch-wide adequacy finding."
    ],
    "fields": [
      [
        "Lot, sample and image region",
        "Defines the inspected object."
      ],
      [
        "Gauge reading, unit and condition",
        "Preserves interpretation."
      ],
      [
        "Change event and unresolved image",
        "Limits associations."
      ]
    ],
    "holdout": "Use a new original closure drawing and reverse record order. Require condition-based grouping rather than first/last row assumptions.",
    "boundary": "Use original inert props and arbitrary readings. Exclude actual limits, equipment drawings and operating settings; qualified staff retain inspection decisions."
  }
}