Search System Type
Test rankers for relevance loss, freshness gaps, ambiguous intent, typo recovery, and zero-result behavior.
Risks in scope
Search Behavior Trials focus on:
- relevance loss
- freshness gaps
- ambiguous intent
- typo recovery
- zero-result behavior
- personalization drift
- baseline-vs-candidate ranking changes
Supported workflows
Native HTTP and schema-mapped HTTP describe search setup paths. Hosted audit, compare, and generated scenario/persona coverage depend on a search-capable deployment. Confirm availability before using these commands.
shell
determina audit --project-id <project-id> --system-version-id <system-version-id> --scenario time-sensitive-query --output-dir ./search-audit
Generated test case boundary
Search supports generated scenario and persona coverage for audit or compare workflows. Do not present search as having full run-swarm parity with recommender; current public docs keep search generation under advanced/platform-owned workflows.
Generate reusable search packs
Use generated packs as an advanced workflow after the basic search audit path works.