extractor

package
v0.0.0-beta.24 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Mar 20, 2026 License: MIT Imports: 6 Imported by: 0

Documentation

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type Claim

type Claim struct {
	EntityName string `json:"entity_name"`
	Claim      string `json:"claim"`
	Status     string `json:"status"` // CONFIRMED | REFUTED | SPECULATIVE
}

Claim is a factual covariate extracted from text.

func ExtractClaims

func ExtractClaims(ctx context.Context, provider llm.Provider, chunks []string) ([]Claim, error)

ExtractClaims extracts factual claims from chunks.

type Entity

type Entity struct {
	Name        string `json:"name"`
	Type        string `json:"type"`
	Description string `json:"description"`
}

Entity extracted from document text.

type ExtractOption

type ExtractOption func(*extractOptions)

ExtractOption configures entity extraction.

func WithMaxGleanings

func WithMaxGleanings(n int) ExtractOption

WithMaxGleanings sets the number of gleaning passes (default: 1).

type ExtractionResult

type ExtractionResult struct {
	Entities      []Entity       `json:"entities"`
	Relationships []Relationship `json:"relationships"`
}

ExtractionResult holds entities and relationships for a chunk.

func ExtractEntities

func ExtractEntities(ctx context.Context, provider llm.Provider, chunks []string, opts ...ExtractOption) (*ExtractionResult, error)

ExtractEntities calls the LLM to extract entities and relationships from chunks, with optional gleaning passes to catch missed entities (inspired by Microsoft GraphRAG).

type Relationship

type Relationship struct {
	Source      string  `json:"source"`
	Target      string  `json:"target"`
	Predicate   string  `json:"predicate"`
	Description string  `json:"description"`
	Weight      float64 `json:"weight"`
}

Relationship between two entities.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL