Experiment 1 – ABN Validation

This experiment tests automatic correction and validation of ABNs extracted from OCR data, using checksum algorithms, API lookups, and error correction techniques.

Experiment Objective

Validate and correct supplier ABNs obtained through OCR by applying:

• Checksum verification
• OCR error correction (O→0, I→1, S→5 substitutions, etc.)
• Cross-checks against the Australian Business Register (ABR)
• Company name fallback search when ABN validation fails

Working Logic

• Validation – Official ABN checksum verification
• OCR Correction – Fix common OCR scanning errors automatically
• API Integration – Retrieve authoritative records from the ABR
• Caching – Store verified results locally to reduce redundant API calls
• Name Search – Use company name lookup if ABN cannot be validated directly

Validation Progress

• Dataset loaded: 50 records (baseline test set)
• Actions: [✔ Data Loaded] [Start Validation] [Pause]
• Live tracker displays progress: % validated, % corrected, % invalid, % missing