How to use Taggun's Feedback API Endpoint

Why Submit Feedback?

Providing feedback helps Taggun's machine learning models learn from your corrections, improving recognition of receipt types, formats, and vendor details. Over time, this reduces the need for manual corrections.

Helpful Links

• API Reference
• Endpoint: POST https://api.taggun.io/api/account/v1/feedback

Implementation Steps

1. Initial Upload

Before uploading a receipt or invoice for data extraction, please ensure the following setup is complete:

1. Include a unique referenceId generated by your system

2. Set incognito: false to allow the receipt to be used for training

For Example

{
  "referenceId": "your-unique-id-123",
  "incognito": false,
  // ... other upload parameters
}

2. Submit Verified Data

After human verification in your system, submit the corrected data to the feedback endpoint:

Example Request

curl --request POST \
     --url https://api.taggun.io/api/account/v1/feedback \
     --header 'accept: application/json' \
     --header 'apikey: 12345678abcdef1234567890ijklmnop' \
     --header 'content-type: application/json' \
     --data '
{
  "referenceId": "\"your-unique-id-123\"",
  "totalAmount": 25,
  "taxAmount": 2,
  "merchantName": "Walmart",
  "currencyCode": "USD",
  "date": "2023-12-31T00:00:00Z"
}
'

The referenceId links the feedback to the original uploaded receipt (you must assign a referenceId in the API request for the original Data Extraction) - allowing our system to compare the AI results with the verified data.

Important Notes

• The referenceId must be consistent between the initial upload for data extraction and feedback submission
• Only receipts that were originally uploaded for data extraction with incognito: false can be used for training
• Submit feedback as soon as possible after human verification to maximise the training benefit