Expand description
Piracy detection and watermarking for .ald format (§9.3)
Provides first-class support for detecting stolen datasets and tracing leaks.
§Features
- Entropy Analysis: Detect watermark presence without seller key
- Watermark Embedding: Buyer-specific fingerprints in LSB of floats
- Watermark Extraction: Recover buyer identity with seller key
- Legal Evidence: Generate cryptographic proof for proceedings
Structs§
- Buyer
Identity - Extracted buyer identity
- Column
Entropy - Entropy analysis results for a column
- Column
Evidence - Per-column evidence
- Detection
Result - Result of watermark detection analysis
- Entropy
Analysis - Full entropy analysis of a dataset
- Legal
Evidence - Legal evidence for proceedings
- Piracy
Detector - Piracy detector for analyzing datasets
- Watermark
- Watermark configuration for embedding
- Watermark
Embedder - Watermark embedder for protecting datasets
Constants§
- AUTOCORRELATION_
THRESHOLD - Autocorrelation threshold for pattern detection Watermarks repeat every 256 bits, creating detectable autocorrelation
- DETECTION_
CONFIDENCE_ THRESHOLD - Minimum confidence for watermark detection
- LSB_
NATURAL_ THRESHOLD - Natural LSB entropy threshold for clean data