ExtractionResult

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "#/components/schemas/ExtractionResult",
  "title": "ExtractionResult",
  "type": "object",
  "description": "The data extraction results for a processed document, including all extracted fields, tables, and their confidence scores.",
  "properties": {
    "document_id": {
      "type": "string",
      "format": "uuid",
      "description": "The identifier of the source document"
    },
    "template_id": {
      "type": "string",
      "format": "uuid",
      "description": "The extraction template that was applied"
    },
    "classification": {
      "type": "string",
      "description": "The classified document type"
    },
    "extraction_method": {
      "type": "string",
      "description": "The AI method used for extraction",
      "enum": [
        "ocr",
        "nlp",
        "generative_ai",
        "template_based",
        "hybrid"
      ]
    },
    "overall_confidence": {
      "type": "number",
      "description": "The overall confidence score for the extraction (0.0 to 1.0)",
      "minimum": 0.0,
      "maximum": 1.0
    },
    "fields": {
      "type": "array",
      "items": {
        "$ref": "#/components/schemas/ExtractedField"
      },
      "description": "The list of extracted fields with values and confidence scores"
    },
    "tables": {
      "type": "array",
      "items": {
        "$ref": "#/components/schemas/ExtractedTable"
      },
      "description": "The list of extracted tables"
    },
    "requires_review": {
      "type": "boolean",
      "description": "Whether any extracted values fall below the confidence threshold and require manual review"
    },
    "extracted_at": {
      "type": "string",
      "format": "date-time",
      "description": "The date and time the extraction was performed"
    }
  }
}