Expand description
Text integrity detection pipeline Integrity detection pipeline — full orchestration
Tier: T3 | Primitives: σ Sequence, → Causality, ∂ Boundary
Modules§
- aggregation
- Signal aggregation — normalize → Beer-Lambert → Hill
- bloom
- Bloom taxonomy → threshold mapping
- burstiness
- Burstiness coefficient — inter-arrival time analysis
- chemistry
- Inlined chemistry primitives — Beer-Lambert, Hill, Arrhenius
- classify
- Classification via Arrhenius threshold
- entropy
- Sliding window Shannon entropy
- perplexity
- Perplexity variance — per-sentence entropy variance
- tokenize
- Text tokenization — word splitting + TTR
- zipf
- Zipf’s Law Deviation — log-log regression
Structs§
- Pipeline
Result - Flat pipeline result for server→client transfer.
Constants§
- DEFAULT_
WINDOW_ SIZE - Default entropy window size.
- DEFAULT_
WINDOW_ STEP - Default entropy window step.
- MIN_
TOKENS - Minimum token count for reliable analysis.
Functions§
- run_
pipeline - Run the full integrity detection pipeline.