Aletheia¶
Aletheia (Greek: truth, disclosure) - A GraphRAG evaluation framework that transforms structured data into queryable knowledge graphs.
Aletheia is built on Graphiti. It provides a pluggable architecture for building knowledge graphs from domain-specific data and rigorous evaluation using RAGAS metrics.
Introduction Video¶
What Aletheia Does¶
Quick Example¶
# Build a knowledge graph
aletheia build-knowledge-graph \
--use-case terrorist_orgs \
--knowledge-graph my_graph \
--schema-mode graph-hybrid
# Evaluate retrieval quality
aletheia evaluate-ragas \
--knowledge-graph my_graph \
--questions questions.json \
--grounding-mode strict
-
Build your first knowledge graph in minutes
-
Domain-aware LLM servers with Cypher analytics
-
Sanctions, aviation safety, and more
-
RAGAS metrics with grounding verification
Key Features¶
| Feature | Description |
|---|---|
| Use Case Architecture | Self-contained data domains with parsers, episode builders, and ontologies |
| Schema Inference | Six modes from free extraction to strict ontology adherence |
| Ontology Support | Load domain ontologies for consistent entity extraction |
| FTM Support | First-class support for FollowTheMoney investigative data format |
| RAGAS Evaluation | Context precision, recall, faithfulness, and answer similarity |
| Grounding Verification | Detect and prevent parametric knowledge leakage |
| MCP Connectors | Domain-aware MCP servers with schema introspection, Cypher analytics, and cross-graph workflows |