HomoglyphDetector™
Visual Lookalike Character Identification System
HomoglyphDetector™ is a specialized sub-component of WatermarkScanner™ focused exclusively on identifying Unicode homoglyph substitutions — characters from different scripts that are visually identical but different code points. These substitutions are used to embed hidden identifiers, track authorship, or evade text processing tools.
How It Works
Script-Aware Parsing
Identifies the expected Unicode script for each word based on the document's primary language. For an English document, all characters should be U+0041–U+007A (Basic Latin). Characters from other scripts that visually resemble Latin are flagged.
Homoglyph Table Lookup
Checks flagged characters against a 500+ entry homoglyph lookup table covering: Cyrillic/Latin pairs (а/a, с/c, е/e, о/o, р/p, х/x, у/y), Greek/Latin pairs (α/a, ο/o, κ/k), and other confusable Unicode pairs.
Context Validation
Runs context checks to reduce false positives: mixed-script words in multilingual documents are not flagged if the context indicates legitimate language switching (e.g., a Cyrillic word in a Russian passage within an English document).
Substitution Reporting
Generates a character-level report listing: original character, Unicode code point, detected homoglyph target, word position, and confidence score. Also provides the cleaned text with all homoglyphs replaced by their standard equivalents.
Key Metrics
Use Cases
- Detecting authorship-tracking homoglyph marks in shared documents
- NLP pipeline normalisation — ensuring consistent character encoding
- Security audit of text that may contain hidden script injections
- Cleaning text before AI processing to prevent encoding attacks
Try HomoglyphDetector™ Now
Powered by our proprietary technology stack. Available directly in the TextHumanize tools and via REST API.