{
  "source": "https://data.ntsb.gov/avdata/FileDirectory/DownloadFile?fileID=avall.zip",
  "source_directory": "https://data.ntsb.gov/avdata/FileDirectory/",
  "snapshot_file": "avall.zip",
  "snapshot_directory_date": "2026-10-01",
  "downloaded_at": "2026-10-26T20:47:42.150086+00:00",
  "archive_sha256": "bc74fefb1b39e1b86700a4be82b8ff6014bb1f6d81008524de73f2ddbd38b052",
  "database_sha256": "c891b8a6fe55651294a0cb0fa328e5648833caf5ac62b1c21d6caf3ea5e99bb7",
  "source_events": 31303,
  "source_narrative_rows": 28516,
  "indexed_unit": "One event; unique narrative sections from all aircraft merged by ev_id",
  "prepared_records": 28043,
  "events_without_usable_narrative": 3260,
  "records_sha256": "5f8d2ff952964f21252748e31de2b0214875f394cbd2bfe4ed011d92f1eb5b12",
  "date_range": {
    "min": "2008-01-01",
    "max": "2026-09-17"
  },
  "narrative_fields_in_order": [
    "narr_cause",
    "narr_accf",
    "narr_accp",
    "narr_inc"
  ],
  "narrative_field_record_counts": {
    "narr_cause": 24418,
    "narr_accf": 24442,
    "narr_accp": 21296
  },
  "preparation": "Decode HTML entities, remove markup, normalize whitespace, discard placeholder narratives and values shorter than 8 characters, deduplicate exact narrative sections within each event. No sampling.",
  "model": "sentence-transformers/all-MiniLM-L6-v2",
  "model_revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41",
  "dimensions": 384,
  "distance": "Cosine",
  "vector_field": "body",
  "text_field": "text",
  "collection": "ntsb_aviation_minilm",
  "text_storage": "Full prepared narrative text retained; the embedding model uses its native token limit. Probable cause and final narrative summary precede the full factual narrative.",
  "text_lengths": {
    "min": 234,
    "median": 2151,
    "p95": 16687,
    "max": 162357
  },
  "event_types": {
    "ACC": 26598,
    "INC": 1445
  },
  "embedding_max_tokens": 256,
  "indexed_records": 28043,
  "indexed_at": "2026-10-26T20:53:29.811312+00:00"
}