Testing #8412
closedVendor Invoice Verification with Product Matching & AI-Based Similar Product Handling
Description
Implement and validate the workflow for verifying vendor invoices by matching invoice products with existing products in the system. The system should leverage AI-based extraction to identify and suggest similar products. Based on user/client validation, the system should either map to an existing product or create a new one.
Business Requirement:
The client requires an intelligent invoice verification mechanism where:
Products from vendor invoices are extracted using AI.
Extracted products are compared against the existing product database.
If similar products are found:
Suggest them to the user.
Allow user to accept or reject the suggestion.
If no suitable match is found or user rejects suggestions:
Allow creation of a new product.
Workflow / Functional Flow:
Invoice Upload
User uploads vendor invoice (PDF/Image).
AI extraction process is triggered.
AI Data Extraction
Extract fields:
Product Name
Quantity
Unit Price
Total Price
Vendor Details (optional)
Product Matching Logic
System compares extracted product names with existing products using:
Exact match
Partial match (string similarity)
AI similarity model
Similar Product Suggestion
If similar products found:
Display list of suggested products.
Show confidence score (if available).
User Decision Handling
Accept Suggestion
Map invoice product to selected existing product.
Reject Suggestion
Proceed to create new product.
New Product Creation
Auto-fill fields from extracted data.
Allow user to edit before saving.
Invoice Verification Completion
All products must be mapped (existing or new).
Mark invoice as "Verified".
Acceptance Criteria:
ID Criteria
AC1 Invoice upload should trigger AI extraction successfully
AC2 Extracted product data should be displayed correctly
AC3 System should suggest similar products when matches exist
AC4 User should be able to accept a suggested product
AC5 User should be able to reject suggestions and create a new product
AC6 New product form should be pre-filled with extracted data
AC7 Invoice cannot be marked verified unless all products are mapped
AC8 System should handle duplicate/similar product names correctly
AC9 Confidence score (if available) should be visible for suggestions
AC10 No data loss should occur during mapping or creation
Test Scenarios (QA Coverage):
Positive Scenarios
Upload valid invoice → AI extracts correctly → suggestions shown
Accept suggested product → mapping successful
Reject suggestion → create new product → saved successfully
Negative Scenarios
AI extraction fails → error handling
No similar products found → direct new product flow
Incorrect AI suggestion → user rejects and proceeds
Edge Cases
Similar product names with slight variations (e.g., “Rose Plant” vs “Red Rose Plant”)
Multiple similar matches returned
Duplicate products in invoice
Special characters / OCR errors in product names
UI Expectations:
Highlight extracted products clearly
Show “Suggested Products” section
Provide:
Accept button
Reject/Create New button
Editable form for new product creation
Dependencies:
AI Extraction Service
Product Database
Matching Algorithm / Similarity Engine
Subtasks
Related issues