Knowledge Fingerprint (KFP) — Algebraic Epistemological Identity, Reconstructable Hashes, and Self-Healing Knowledge Bases

A deterministic system for characterizing the epistemological identity of knowledge units and AI agents. Features a 7-dimensional taxonomy, a formally proven Meet-Semilattice composition algebra, dual-fingerprinting for metadata inconsistency detection (82% misclassification found), Reconstructable Knowledge Hashes (RKH), Knowledge DNA (~19 KB for 191 articles), epistemological RAG filtering (86% unsafe chunk reduction), agent behavioral profiling with drift detection, and a PubMedQA-inspired benchmark (66% rule-based, 82% LLM accuracy). Standard-library Python, 914 tests, PolyForm Noncommercial 1.0.