Data Source Verification is the process of ensuring that data is accurate, complete, and consistent before it is processed or stored in a system. While computers can automate many validation checks, certain verification tasks require human judgment, contextual understanding, and pattern recognition that machines cannot reliably replicate.
Computers excel at rule-based checks (e.g., range checks, format checks), but they struggle with:
A human operator compares the original source document (e.g., a paper form, receipt, or ID) with the data entered into the computer system. This catches transcription errors — mistakes made when copying data from one medium to another.
Example: A data entry clerk checks that the name typed into a database matches the name on a handwritten application form.
OCR (Optical Character Recognition) software can scan printed text but often struggles with messy or stylised handwriting. Humans are assigned to review characters flagged as 'low confidence' by the OCR system and correct them using contextual understanding.
Example: A bank employee reviews scanned cheques where the OCR could not confidently read the amount written in words.
Two different operators independently enter the same data into a system. The system then compares both entries:
This technique reduces the chance of undetected transcription errors.
Verifying whether a scanned or digital signature matches a physical ID or reference signature requires visual pattern matching and judgment of authenticity — a task that humans perform more reliably than automated systems in many real-world contexts.
Example: A bank teller compares a customer's signature on a cheque with the signature on file.
Some data cannot be verified by rules alone. Tasks such as content moderation and sentiment analysis verification require humans to determine whether data is:
Computers lack genuine understanding of social context, making humans essential for these tasks.
| Verification Task | Why Humans Are Needed |
|---|---|
| Visual Inspection | Contextual comparison of physical vs. digital data |
| Handwritten Document Verification | Superior pattern recognition over OCR |
| Double Data Entry | Human judgment to resolve mismatches |
| Signature Verification | Visual authenticity judgment |
| Content Moderation | Understanding of cultural/social context |