NVOpen source
NVIDIA Auto Ontology
Build a semantic catalog from enterprise metadata and give agents consistent business definitions and query context.
Ontology & Semantic Modeling
EvoOntology
Expose a versioned ontology layer to data agents and evolve it using execution evidence and paired evaluations.
DOCUMENTATION-BASED SELECTION
Why it is a Featured Tool
An explicit build, use, evolve, and evaluate lifecycle provides a distinct starting point for studying how an agent ontology layer changes with workload evidence.
Before you choose: This is a specialized semantic layer. Review the representation and evaluation design; it does not replace document parsing.
See the shortlist and criteria →At a glance
- Input
- Tables, files, databases, tasks and execution traces
- Output
- Ontology workspace, semantic objects, MCP tools
- Provenance
- Evidence-linked semantics; not default page coordinates
- Scanned pages
- Not applicable; parse documents upstream
- Deployment
- Self-hosted
- Availability
- Open source
- Costs to plan for
- Compute or model API costs
Semantic modeling profile
Agent ontology layer
Representation and interoperability
Project-specific typed semantic objects with content, schema, and tool layers.
Updates and versioning
Parent and candidate are evaluated under the same workload before publication.
Limitations to check
Uses its own ontology representation. Candidate improvements depend on workload evidence and evaluation; verify interchange requirements and failure cases.
Build a meaningful sample
Test ambiguous terms, conflicting metric definitions, source mappings, and schema changes. Check how revisions affect the same workload.