{
  "title": "Electric distribution outage records for simulated AI review",
  "source_page": "https://fiund.com/industries/electric-distribution-outage-review-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "Separate missing monitoring data from confirmed service evidence",
    "context": "A detached, invented timeline tests retrospective status interpretation. It includes no customer outage history, feeder information or connection to operational systems.",
    "records": [
      [
        "ALERT-A1",
        "Practice monitoring alert",
        "Area alias A has a missing observation at 10:00."
      ],
      [
        "CHECK-A2",
        "Independent status note",
        "At 10:02, the authorized practice reviewer records service present for A during that interval."
      ],
      [
        "ALERT-B1",
        "Second practice alert",
        "Area alias B has a missing observation at 10:00."
      ],
      [
        "CHECK-B2",
        "Independent status note",
        "At 10:04, the reviewer records an interruption for B from 09:58 to 10:03."
      ]
    ],
    "question": "Which record supports an interruption, and can both missing observations be labeled outages?",
    "answer": "CHECK-B2 independently records B's interruption and its interval. ALERT-B1 alone would not establish it. For A, the missing observation and CHECK-A2's stated service evidence must remain distinct; the supplied record does not support labeling the data gap itself an outage. Report what each source establishes without inferring a cause or an operational response.",
    "failureModes": [
      "Converting every absent monitoring sample into a service interruption.",
      "Using B's independently confirmed event as ground truth for A.",
      "Extending a reviewed status beyond the interval it names."
    ],
    "fields": [
      [
        "Area alias",
        "Match independent evidence to the correct synthetic case."
      ],
      [
        "Observation and event time",
        "Keep a missing sample separate from an interruption interval."
      ],
      [
        "Evidence role",
        "Distinguish an alert from the supplied independent status record."
      ]
    ],
    "holdout": "Omit CHECK-A2 in a new case. The answer should leave A's service state unresolved rather than infer either normal service or an outage.",
    "boundary": "Use fabricated records or an authorized detached simulation. Exclude real maps, switching information, customers, credentials and precise asset locations. No control action is requested."
  }
}