context-ontology-accelerator vs neocarta
AWS's ontology-based context layer: scan your sources, induce ontologies, then serve validated context to agents over MCP — SPARQL federation, a virtual knowledge graph and OWL reasoning. — versus — Neo4j Labs' semantic layer for data agents: ingest warehouse schema, business glossary, metrics and query history into one graph, then serve it over MCP so agents route queries and write grounded SQL.
AWS's take: induce ontologies from your sources and serve validated context to agents over MCP with SPARQL federation and OWL reasoning. Neocarta skips the ontology layer and grounds agents in observed schema, FKs and real query history instead.
| context-ontology-accelerator | neocarta | |
|---|---|---|
| Stars | 667 | 98 |
| Forks | 66 | 26 |
| Language | Python | Python |
| License | Apache-2.0 | Apache-2.0 |
| Last activity | 2 days ago | 4 days ago |
| Topics | knowledge-graphs | knowledge-graphs, data |
| Curated connections | 6 | 5 |
context-ontology-accelerator — the curator's take
The serious option when an agent's answers have to be defensible: ontology induction and OWL reasoning (HermiT/ELK) sit between your data and the model, an Ontop virtual knowledge graph federates SPARQL without copying anything, and namespace isolation plus platform roles govern who sees what. The cost is equally serious — it deploys as AWS CDK stacks across a Smithy-generated monorepo wanting Python 3.12, Node 22, Java 17, Docker and Nx, so this is a platform-team project, not a weekend install. Note the governance too: published as a read-only mirror with no pull requests accepted, and you're told to start from a release tag rather than `main`.
neocarta — the curator's take
The honest fix for Text2SQL: the model is not bad at SQL, it is blind to your data landscape. Neocarta pulls schema metadata, foreign keys, sample values, glossary terms, governed metric definitions and real query history into a Neo4j graph — only metadata crosses over, data stays in the source — then serves it to agents over MCP with full-text, vector and hybrid search that returns columns, types, example values and the FK references needed to build a join. Embeddings are optional; catalog search works from schema alone. It ships a runnable LangGraph + BigQuery agent so you can see the routing loop end to end. Read the label though: Neo4j Labs, explicitly experimental, not a supported product, 98 stars, and it assumes you are willing to stand up and maintain a Neo4j instance next to your warehouse.