This page defines the required schema for receipt data used in the OCR Data Integrity experiments. Ensuring consistent data structure is essential for benchmarking error detection, correction, and validation steps.

Receipt Data Structure:

{
  "id": "string",            // Unique receipt identifier (required)
  "abn": "string",           // Supplier ABN (optional, required in Exp 2: ABN Validation)
  "supplier_name": "string",  // Supplier name
  "date": "YYYY-MM-DD",      // Receipt date (required)
  "amount_total": number,    // Total amount (required)
  "gst": number,             // GST amount (optional, checked in Exp 3: Arithmetic Consistency)
  "gst_included": boolean,   // Whether GST is included
  "items": [                  // Array of goods/services
    {
      "description": "string",
      "amount": number,
      "gst": number
    }
  ]
}

Field Requirements

• Required fields: id, date, amount_total
• Optional fields: abn, supplier_name, gst, gst_included, items

Usage Notes

• id must be unique per receipt.
• date must follow ISO format YYYY-MM-DD.
• amount_total should equal the sum of items.amount (when items are provided).
• abn is required in Experiment 2: ABN Validation.
• items enable line-level testing in Experiment 3: Arithmetic Consistency.
• gst and gst_included are used in both Experiment 3 and Experiment 4 for hybrid correction.

This way the page is not just technical JSON, but clearly shows the link between fields and experiments, making it auditor-friendly and developer-ready.