Search Filtering Logic

InfraLynx search applies filters after the result corpus is assembled and scored.

Matching Rules

• query text is normalized to lowercase
• whitespace is collapsed before tokenization
• every query token must match somewhere in the record corpus
• title matches score higher than location, summary, or tag matches
• result ordering is stable by score, then domain, then title, then identifier

Filter Rules

• all searches every supported domain
• domain filters narrow results to one domain without changing tokenization
• grouped results always preserve domain boundaries even when the filter is all

Result Grouping

Results are grouped in fixed domain order:

1. Core Platform
2. IPAM
3. DCIM
4. Operations
5. Automation

This order is intentional and deterministic. It prevents relevance ties from producing inconsistent group ordering between runs.

UI State Rules

• empty query keeps search idle
• non-empty query triggers centralized fetch
• failed requests produce a retryable error state
• selected results remain explicit by result ID

Future Extension

Later indexing engines may improve ranking, but they must preserve:

• explicit domain grouping
• domain filter semantics
• deterministic tie-breaking