@plust/datasleuth/dist/steps/classify.d.ts

Build LLM-powered research pipelines and output structured data.

/**
 * Entity classification and clustering step implementation
 * This module provides advanced entity recognition, classification and clustering
 * capabilities to organize research findings into a coherent knowledge graph.
 */
import { createStep } from '../utils/steps.js';
/**
 * Options for entity classification
 */
export interface ClassifyOptions {
    /**
     * Whether to identify and classify entities in the research data
     */
    classifyEntities?: boolean;
    /**
     * Whether to cluster identified entities by relationship
     */
    clusterEntities?: boolean;
    /**
     * Minimum confidence threshold for entity classification (0-1)
     */
    confidenceThreshold?: number;
    /**
     * Custom entity types to identify (in addition to standard types)
     */
    customEntityTypes?: string[];
    /**
     * Maximum number of entities to extract
     */
    maxEntities?: number;
    /**
     * Maximum number of clusters to generate
     */
    maxClusters?: number;
    /**
     * Custom instructions for entity clustering algorithm
     */
    clusteringInstructions?: string;
    /**
     * Retry configuration for LLM calls
     */
    retry?: {
        /** Maximum number of retries */
        maxRetries?: number;
        /** Base delay between retries in ms */
        baseDelay?: number;
    };
}
/**
 * Entity object structure
 */
export interface Entity {
    name: string;
    type: string;
    description: string;
    confidence: number;
    mentions: number;
    attributes?: Record<string, any>;
}
/**
 * Relationship between entities
 */
export interface EntityRelationship {
    source: string;
    target: string;
    relationship: string;
    confidence: number;
}
/**
 * Cluster of related entities
 */
export interface EntityCluster {
    name: string;
    description: string;
    entities: string[];
    confidence: number;
}
/**
 * Classification data structure
 */
export interface ClassificationData {
    entities: Record<string, Entity>;
    relationships: EntityRelationship[];
    clusters: Record<string, EntityCluster>;
}
/**
 * Create a classification step
 * @param options - Classification options
 * @returns A configured classification step
 */
export declare function classify(options?: ClassifyOptions): ReturnType<typeof createStep>;