Pattern Normalization & Scoring Pipeline
Canonical Variable Folding • Support Cache • Surprise Metrics
01. RAW INPUTS 02. NORMALIZATION 03. SUPPORT CACHE 04. SURPRISE SCORING vars vars key supp miss store PATTERN_A [RAW] p(X, Y) :- rel(X, Y) var binding: X, Y PATTERN_B [RAW] p(A, B) :- rel(A, B) var binding: A, B VARS_TO_INDICES p($0, $1) :- rel($0, $1) hash: 0x8F4A2E (folded) SUPPORT_CACHE SLOT #042 [HIT] cached supp: 0.142 DB_EVALUATION execute query & count SURPRISE_SCORING METRIC: ISURP score: 2.816 bits
INSPECTOR: PIPELINE OVERVIEW
Two distinct variable-renamed patterns enter the system. The vars_to_indices normalization node collapses them into a single canonical form before probing the support cache and computing unexpectedness.