Reading Methodology
A transparent breakdown of how our interpretation engine works.
1. Input Sanitization & Safety Filtering
User queries are trimmed, normalized, and evaluated against safety dictionaries. Queries relating to medical emergencies or self-harm immediately trigger supportive crisis resource cards instead of readings.
2. Weighted Intent Classification
Keyword density and grammatical patterns score the query across 11 category modules (e.g. Love, Ex, Vocation, Family, Timing).
3. Deterministic Hashing & Content Selection
A bitwise DJB2 hash algorithm turns the question string into a stable numeric seed. This seed pulls thematic archetypes, energy descriptions, action suggestions, and reflection questions from our curated editorial library.