databrickslabs/ontobricks

Transforms Databricks Unity Catalog tables into a materialized knowledge graph, with ontology design and reasoning exposed as tools via MCP.

What it solves

OntoBricks simplifies the process of turning structured data in Databricks tables into a queryable, materialized Knowledge Graph. It removes the manual effort of designing ontologies, mapping data, and managing triple stores by providing an integrated web application that automates the pipeline from metadata import to graph visualization.

How it works

OntoBricks uses a multi-step pipeline to transform Databricks Unity Catalog tables into a graph:

  1. Ontology Design: Users create ontologies (OWL) via a visual canvas or import industry standards. An LLM-powered assistant can help design entities and relationships through natural language.
  2. Mapping: The system maps ontology entities to tables using R2RML. LLMs can automate the generation of these SQL mappings.
  3. Materialization: Data is materialized into a triple store. The project supports multiple backends: Lakebase (Postgres), Lakehouse (Unity Catalog Delta tables), or Neo4j.
  4. Querying & Reasoning: The resulting graph is exposed via an auto-generated GraphQL API and a visual viewer. Users can run OWL 2 RL inference, SWRL rules, and SHACL validation to reason over the data.

Who it’s for

It is designed for data engineers and architects working within the Databricks ecosystem who need to build knowledge graphs for complex data analysis, data quality validation, or semantic search.

Highlights

  • LLM-Powered Automation: A "four-click" pipeline that uses LLMs to generate ontologies and SQL mappings from metadata.
  • Pluggable Graph Backends: Support for Lakebase (Postgres), Lakehouse (Delta), and Neo4j.
  • Advanced Graph Analytics: Includes data cluster detection (Louvain, Label Propagation) and cohort discovery for grouping entities.
  • Semantic Reasoning: Built-in support for OWL 2 RL, SWRL rules, and SHACL constraints.
  • Unity Catalog Integration: Ability to bind Unity Catalog functions to ontology classes as executable actions on graph nodes.

Related

  • Project
    aws-solutions/generative-ai-application-builder-on-aws

    A web-based management dashboard for deploying and experimenting with customizable Generative AI use cases on AWS, supporting RAG-enabled chatbots and AI agents.

  • Project
    VasiHemanth/tokentelemetry

    A local observability dashboard that tracks token usage, LLM costs, and session traces for AI coding and autonomous agents by reading their local log files.

  • Project
    mikekeith52/scalecast

    A time series forecasting library that provides a uniform interface for ML and deep learning models, simplifying model selection, validation, and pipeline creation.

  • Project
    HewlettPackard/swarm-learning

    A decentralized, privacy-preserving machine learning framework that enables collaborative model training across distributed data sources using blockchain for coordination without sharing raw data.