Module evaluation 

Evaluation framework for query results and pipeline validation Evaluation framework for GraphRAG system

This module provides two complementary evaluation approaches:

1. LLM-based Query Result Evaluation

Evaluate GraphRAG query results using LLM-based metrics:

• Relevance: How relevant is the answer to the query?
• Faithfulness: Is the answer grounded in the retrieved context?
• Completeness: Does the answer address all aspects of the query?
• Coherence: Is the answer well-structured and easy to understand?
• Groundedness: Are entity names and relationships correctly mentioned?

2. Pipeline Phase Validation

Validate each phase of the GraphRAG pipeline:

• Document Processing: Chunking and enrichment validation
• Entity Extraction: Entity quality and coverage checks
• Relationship Extraction: Relationship validity and connectivity
• Graph Construction: Overall graph structure validation

Re-exports

pub use pipeline_validation::PhaseValidation;

pub use pipeline_validation::ValidationCheck;

pub use pipeline_validation::PipelineValidationReport;

pub use pipeline_validation::DocumentProcessingValidator;

pub use pipeline_validation::EntityExtractionValidator;

pub use pipeline_validation::RelationshipExtractionValidator;

pub use pipeline_validation::GraphConstructionValidator;

Modules

pipeline_validation

Pipeline validation framework

Structs

DimensionScore

Score for a single evaluation dimension

EvaluableQueryResult

A query result from GraphRAG that can be evaluated

EvaluableQueryResultBuilder

Builder for creating evaluable query results

LLMEvaluation

Parsed LLM evaluation result

LLMEvaluationPrompt

LLM evaluation prompt template for query results

ResultMetadata

Metadata about how the result was generated