Fundamental

← all recipes

Gravity core

Search Relevance Field

let search results settle by relevance, confidence, and recency

frame — bodies 0 relationships 0

Running the actual recipe live via applyRecipe() — not a mock.

"Search Relevance Field" — let search results settle by relevance, confidence, and recency. It has a attract layer (draw matter into a focus / well), a memory layer (a decaying trail / hysteresis), and a repel layer (carve a void / keep-clear region). The active render stack is particles, heatmap.

Relevance becomes attract strength so confident results sink into deeper wells; weak results drift out (repel); previously opened results retain memory.

Runtime tokens
attract memory repel
Concepts
relevance
Metrics
density memory recency density is a host-supplied lane — drive it with data-field-<metric> (or a domain model); without that, its --field-* stays inert.
Diagnostics
potential prediction heatmap
Conditions
related
Render
particles · heatmap
Reduced motion
results re-rank into a static ordered list; memory shows as a "seen" tick
Without motion
ranking is a numbered list; relevance and recency are labelled

Copy this recipe

<field-root></field-root>
  <div data-body="attract" data-strength="1" data-range="280" data-feedback></div>
  <div data-body="memory" data-strength="0.6" data-range="320"></div>
  <div data-body="repel" data-strength="0.5" data-range="200"></div>

See also