Duplicate and Similarity Check

Key Capabilities

• Duplicate Detection: Detects repeated submission of identical receipts.
• Shared Receipt Detection: Mitigates collaborative fraud by identifying suspiciously similar receipts across different users.
• Modified Receipt Detection: Catches fraudulent attempts even when receipt images have been slightly altered.

How It Works

After the text is extracted from the receipt, our system analyses the data to determine how similar one receipt is to another. Even if users attempt to manipulate the image, our system can flag these suspicious submissions for further investigation.

1. Data Extraction: Taggun extracts text and supporting data from the image.
2. Similarity Analysis: Extracted data is compared to other receipts in your account (or campaign for receipt validation endpoints).
3. Intelligent Scoring: A similarity score is generated, with scores above 0.9 flagged as potentially suspicious.
4. Merchant-Specific Thresholds: Score thresholds are adjusted per merchant to reduce false positives. This is per merchant to reduce false positives for receipts that are naturally similar.

Setup Process

1. Contact Taggun

Reach out to [email protected] to enable this feature for your account.

2. Update API Requests

Data Extraction API Endpoints

No changes are required. This feature is automatically enabled once activated by Taggun.

Receipt Validation API endpoints

Include the following optional field in your request:

   "fraudDetection": {
       "allowSimilarityCheck": true
   }

3. Submit an API Request

When submitting a receipt for validation, you can include these optional fields for enhanced tracking and fraud detection:

{
  "referenceId": "your_unique_submission_id",
  "userId": "your_system_user_id"
}

Understanding Request Parameters

Understanding Response Properties

Example Response

When no similar receipts are found:

{
  "trackingId": "T-20241001-8439425",
  "entities": {
    "similarReceipts": []
  }
}

When similar receipts are found:

{
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "RCPT-2024-10-07-001",
        "userId": "USER12345",
        "trackingId": "T-20241001-8439425",
        "similarityScore": 0.95
      },
      {
        "referenceId": "RCPT-2024-10-07-002",
        "userId": "USER67890",
        "trackingId": "T-20241001-4801341",
        "similarityScore": 0.95
      }
    ]
  }
}

Receipt Validation Example

A user submits the same receipt twice to claim a reward, and the system flags it as a duplicate. They then try to bypass detection by altering details like the merchant ID and transaction date. Despite the changes, the system detects the similarities and flags them as suspicious, preventing the fraudulent claim.

  "similarReceipts": [
    {
      "referenceId": "REF002",
      "userId": "USER-001",
      "trackingId": "T-20241007-6127420",
      "similarityScore": 1 /*duplicate is found*/
    },
    {
      "referenceId": "REF003",
      "userId": "USER-001",
      "trackingId": "T-20241007-6748290",
      "similarityScore": 0.95 /*slightly modified fraudulent attempt is found*/
}

Common Scenarios

📘

Here are some typical scenarios you may encounter.

See some real-world examples of how Taggun’s similarity detection system identifies different scenarios, ranging from harmless duplicates to sophisticated fraud attempts.

Note that Taggun is equipped to handle many more cases beyond the examples provided here.

1. Two different images of the same receipt

This scenario occurs when two users submit different photos of the same physical receipt. It could be an innocent mistake (e.g., family members both submitting a shared receipt) or a deliberate attempt to claim multiple rewards.

Response

{
  "trackingId": "T-001",
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "R-001",
        "userId": "U-001",
        "trackingId": "T-002",
        "similarityScore": 1
      }
    ]
  }
}

2. Attempted bypass with slight modification

In this case, a user tries to modify the receipt slightly (e.g., changing the transaction date) to bypass detection. The system still detects the high similarity.

Response

{
  "trackingId": "T-003",
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "R-002",
        "userId": "U-002",
        "trackingId": "T-004",
        "similarityScore": 0.98
      }
    ]
  }
}

3. Multiple similar submissions by the same user

This scenario shows a single user submitting multiple similar receipts. It could be accidental (e.g., forgetting about a previous submission) or an attempt to claim multiple rewards.

Response

{
  "trackingId": "T-005",
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "R-003",
        "userId": "U-003",
        "trackingId": "T-006",
        "similarityScore": 1
      },
      {
        "referenceId": "R-004",
        "userId": "U-003",
        "trackingId": "T-007",
        "similarityScore": 1
      }
    ]
  }
}

4. Multiple similar submissions by different users

This indicates a potential syndicated fraud attempt, where multiple users coordinate to submit similar receipts. This could be an organised effort to exploit.

Response

{
  "trackingId": "T-008",
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "R-005",
        "userId": "U-004",
        "trackingId": "T-009",
        "similarityScore": 0.97
      },
      {
        "referenceId": "R-006",
        "userId": "U-005",
        "trackingId": "T-010",
        "similarityScore": 0.96
      },
      {
        "referenceId": "R-007",
        "userId": "U-006",
        "trackingId": "T-011",
        "similarityScore": 0.95
      }
    ]
  }
}

5. Incidental Similarities (e.g., daily coffee purchases)

This scenario represents similar but legitimate receipts, such as purchases from a favourite fastfood joint. These are flagged due to their similarity but are likely genuine transactions.

Response

{
  "trackingId": "T-012",
  "entities": {
    "similarReceipts": [
      {
        "referenceId": "R-008",
        "userId": "U-007",
        "trackingId": "T-013",
        "similarityScore": 0.82
      }
    ]
  }
}

Use Cases

• Promotions: Ensure each participant submits unique, valid receipts.
• Expense Management: Prevent duplicate reimbursement claims.

FAQs

Q: Will Taggun flag two different people submitting the same receipt?

A: Yes. Including the userId in the request facilitates investigation of potential fraudulent behavior across users.

Q: What if receipts from a specific merchant are always very similar?

A: Taggun's engine dynamically adjusts similarity thresholds based on merchant-specific patterns, reducing false alarms for legitimate submissions.

Best Practices

• Always include a referenceId to avoid flagging legitimate resubmissions.
• Use the userId field to track potential collaborative fraud attempts.
• Implement a review process for receipts flagged as suspicious.

Getting Starts

Contact us at [email protected] to enable Similarity Check on your account.