What detection methods does Siberson Veriket Data Discovery use?

Question

Accepted Answer

Single-method discovery (e.g., regex-only) produces high false-positive rates. Siberson Veriket Data Discovery applies a multi-layer detection model where multiple methods can vote on a single finding, raising precision dramatically. Method How It Works Best For Regular Expressions Pattern matching against text streams, with optional check-digit validation (e.g., Luhn for credit cards, TC Kimlik mod-11) Well-formatted identifiers — cards, IDs, phone numbers Dictionary / Keyword Match against curated lists, with regional packs for Turkish, English, German, French, Spanish, Russian, Arabic, and more Domain terminology — medical, legal, defense Exact Data Match (EDM) Hash-fingerprinting of customer master records (e.g., HR or customer DB exports) for one-to-one identification Customer lists, employee data, IP documents AI / ML Classifiers Trained models that recognize document categories (contracts, invoices, source code) and content sensitivity Unstructured content where pattern matching is too brittle Document Fingerprinting Identifies entire documents and their derivatives (copy, rename, partial extract) via Siberson Verikor DLP's fingerprint engine Sensitive corporate documents, s

Method	How It Works	Best For
Regular Expressions	Pattern matching against text streams, with optional check-digit validation (e.g., Luhn for credit cards, TC Kimlik mod-11)	Well-formatted identifiers — cards, IDs, phone numbers
Dictionary / Keyword	Match against curated lists, with regional packs for Turkish, English, German, French, Spanish, Russian, Arabic, and more	Domain terminology — medical, legal, defense
Exact Data Match (EDM)	Hash-fingerprinting of customer master records (e.g., HR or customer DB exports) for one-to-one identification	Customer lists, employee data, IP documents
AI / ML Classifiers	Trained models that recognize document categories (contracts, invoices, source code) and content sensitivity	Unstructured content where pattern matching is too brittle
Document Fingerprinting	Identifies entire documents and their derivatives (copy, rename, partial extract) via Siberson Verikor DLP's fingerprint engine	Sensitive corporate documents, source files, legal templates
OCR-then-detect	Runs Tesseract OCR over images and scanned PDFs, then feeds extracted text through every other method	Scanned IDs, faxed contracts, screenshot leaks