DICE API Documentation/com.embabel.dice.common.resolver.searcher/VectorCandidateSearcher

VectorCandidateSearcher

class VectorCandidateSearcher(repository: <Error class: unknown class>, autoAcceptThreshold: Double = 0.95, candidateThreshold: Double = 0.7, topK: Int = 10) : CandidateSearcher

Searches using vector/embedding similarity.

Returns a confident match if exactly 1 result exceeds the auto-accept threshold. Otherwise returns all candidates above the minimum threshold for potential LLM arbitration.

Parameters

repository

The repository for vector search

autoAcceptThreshold

Similarity above this returns confident match (if exactly 1)

candidateThreshold

Minimum similarity to be considered a candidate

topK

Maximum candidates to retrieve

Constructors

VectorCandidateSearcher

constructor(repository: <Error class: unknown class>, autoAcceptThreshold: Double = 0.95, candidateThreshold: Double = 0.7, topK: Int = 10)

Types

Companion

object Companion

Properties

resolutionLevel

open override val resolutionLevel: ResolutionLevel

Which ResolutionLevel a confident hit from this searcher counts as, for the per-level telemetry EscalatingEntityResolver logs. Each searcher declares its own tier so the counts stay accurate whatever the chain's length or order. Defaults to ResolutionLevel.HEURISTIC_MATCH; exact-lookup searchers report ResolutionLevel.EXACT_MATCH and vector search reports ResolutionLevel.EMBEDDING_MATCH.

Functions

open override fun search(suggested: SuggestedEntity, schema: <Error class: unknown class>): SearchResult

Search for candidates matching the suggested entity.