Stage 2 — Structured extraction

2.5 Stage 2 — Structured extraction

This stage converts the document-reading output into a ReceiptExtraction object. The public contract is the schema and stage behavior; model providers, prompt text, routing policy, token budgets, and retry conditions are managed in the internal operations layer.

Model-routing boundary

Yumo Yumo runs structured extraction behind a model-independent interface. Operational policy can choose the appropriate engine based on language, document complexity, health state, and quality signals. The weights, ordering, and fallback behavior of that policy remain private.

Structured output

// ReceiptExtraction
{
  "merchant": {
    "name_raw": "MIGROS T.A.S.",
    "tax_id_raw": "1234567890",
    "address_raw": "...",
    "phone_raw": null
  },
  "captured_at_raw": "17/05/2026 14:23",
  "currency": "TRY",
  "totals": {
    "subtotal": 234.50,
    "tax_total": 42.21,
    "grand_total": 276.71
  },
  "tax_lines": [
    { "rate_pct": 18.0, "base": 200.0, "amount": 36.0 },
    { "rate_pct": 8.0, "base": 77.50, "amount": 6.20 }
  ],
  "payment_method": "credit_card",
  "line_items": [
    {
      "raw_text": "SUT 1L PINAR",
      "qty": 2,
      "unit_price": 23.50,
      "line_total": 47.00,
      "tax_rate_pct": 8.0
    }
  ],
  "quality_band": "medium",
  "extraction_notes": "tax_total reconstructed from tax_lines"
}

The schema reduces model output into a form that the rules layer can validate. Totals, date, currency, line items, and tax fields are checked again in the next stage.

Consistency handling

If extraction carries a low quality band or the rules layer finds inconsistency, the pipeline can send the result to review or reprocessing. Path selection is managed through operational parameters.