DICE API Documentation/com.embabel.dice.text2graph.support

Package-level declarations

Types

object AcceptRecommendedRelationshipMergePolicy : RelationshipMergePolicy

Always adds new entities and ignores existing or vetoed entities.

class LlmSourceAnalyzer(ai: <Error class: unknown class>, llmOptions: <Error class: unknown class> = LlmOptions()) : SourceAnalyzer

Source analyzer that uses an LLM to suggest entities and relationships.

MultiPassKnowledgeGraphBuilder

class MultiPassKnowledgeGraphBuilder(sourceAnalyzer: SourceAnalyzer, entityResolver: EntityResolver, relationshipResolver: RelationshipResolver = AcceptSuggestionRelationshipResolver, entityMergePolicy: EntityMergePolicy = UseNewEntityMergePolicy, relationshipMergePolicy: RelationshipMergePolicy = AcceptRecommendedRelationshipMergePolicy, diceEventListener: DiceEventListener = DiceEventListener.DEV_NULL) : KnowledgeGraphBuilder

Strategy-based builder for knowledge graphs. Responsible for analyzing source chunks and building a knowledge graph delta and then calling the knowledge graph updater to apply the delta.

ParallelSourceAnalyzer

class ParallelSourceAnalyzer(analyzers: List<SourceAnalyzer>, executor: ExecutorService = Executors.newFixedThreadPool(analyzers.size), config: ParallelSourceAnalyzer.Config = Config()) : SourceAnalyzer

Runs multiple source analyzers in parallel on the same chunk and merges results. Entities suggested by multiple analyzers are weighted higher (agreement scoring).

SourceAnalyzerProperties

data class SourceAnalyzerProperties(val llms: List<<Error class: unknown class>> = listOf(LlmOptions()), val minAgreement: Int = 1)

Configuration for source analyzers. Supports multiple LLMs for parallel analysis with agreement voting.

UseNewEntityMergePolicy

object UseNewEntityMergePolicy : EntityMergePolicy

Always adds new entities and ignores existing or vetoed entities. This is probably not want you want in production!