ontobuilder 0.1.0

Build, explore, and export ontologies from data — no PhD required.

OntoBuilder

Build, explore, and export ontologies from data — no PhD required.

OntoBuilder is a Python toolkit that turns raw data (CSV, JSON) into formal ontologies with concepts, relations, properties, and OWL/RDF exports. It comes with a CLI, a web UI, LLM-powered assistance, and a beginner-friendly glossary.

---

Features

• Data → Ontology pipeline — analyze CSV/JSON files, get concept and relation suggestions, build ontologies automatically or interactively
• Rich CLI — manage concepts, properties, relations, instances, and exports from the terminal
• OWL/RDF export — Turtle and RDF-XML via rdflib, with built-in reasoning and consistency checks
• RAG-friendly exports — JSON-LD, Schema Card, and system prompt text formats
• LLM integration — AI-powered interview mode, natural language chat, and structure inference
• Domain templates — pre-built starters for healthcare, e-commerce, and more
• Web UI — visual ontology builder with Streamlit
• Graph backends — NetworkX (built-in) and Neo4j
• Educational glossary — learn ontology terms as you build

---

Installation

# Core (CLI + OWL export + reasoning)
pip install ontobuilder

# With AI features
pip install ontobuilder[llm]

# With web UI
pip install ontobuilder[web]

# Everything
pip install ontobuilder[all]

For development:

git clone https://github.com/iksun/ontobuilder.git
cd ontobuilder
pip install -e ".[dev,llm,web]"

---

Quick Start

Python API

from ontobuilder import Ontology

onto = Ontology("Pet Store", description="A pet store domain model")

# Add concepts with hierarchy
onto.add_concept("Animal", description="A living creature")
onto.add_concept("Dog", parent="Animal", description="A domestic dog")
onto.add_concept("Cat", parent="Animal", description="A domestic cat")
onto.add_concept("Customer", description="A person who buys pets")

# Add properties
onto.add_property("Animal", "name", data_type="string", required=True)
onto.add_property("Animal", "age", data_type="int")
onto.add_property("Dog", "breed", data_type="string")

# Add relations
onto.add_relation("buys", source="Customer", target="Animal")

# Add instances
onto.add_instance("Rex", concept="Dog", properties={"name": "Rex", "breed": "Labrador"})

print(onto.print_tree())

CLI

# Create a new ontology
onto init "Hospital Booking"

# Add concepts
onto concept add Patient --description "A person receiving care"
onto concept add Surgeon --parent Provider
onto concept add SurgeryBooking

# Add relations
onto relation add assigned_surgeon --source SurgeryBooking --target Surgeon

# View the ontology
onto info

# Export to OWL
onto owl export --format turtle

# Run consistency checks
onto owl reason

# Query the ontology
onto owl query describe SurgeryBooking
onto owl query path SurgeryBooking --target Hospital

Build from Data

# Analyze a data file
onto tool analyze data.csv

# Get ontology suggestions
onto tool suggest data.csv

# Auto-build ontology from data
onto tool build data.csv

# Interactive mode — review each suggestion step by step
onto tool build -i data.csv

Web UI

pip install ontobuilder[web]
streamlit run streamlit_app.py

---

CLI Commands

Command	Description
onto init	Create a new ontology project
onto info	Show ontology summary
onto concept add/list/remove	Manage concepts
onto relation add/list/remove	Manage relations
onto save / load	Save or load .onto.yaml files
onto export	Export to various formats
onto owl export	Export as OWL/RDF (Turtle or RDF-XML)
onto owl reason	Run OWL inference and consistency checks
onto owl query	Structured queries (classes, relations, describe, path)
onto tool analyze	Analyze a data file
onto tool suggest	Show ontology suggestions from data
onto tool build	Build ontology from data (auto or interactive)
onto interview	AI-powered interview to build an ontology
onto chat	Chat with your ontology in natural language
onto workspace	Full workspace: analyze, build, chat, export
onto learn	Learn ontology terms
onto suggest	Get next-step suggestions
onto domains list/apply	Browse and apply domain templates
onto configure	Set up LLM API keys and models

---

Architecture

src/ontobuilder/
├── core/           # Data model — Concept, Property, Relation, Instance, Ontology
├── cli/            # Typer CLI with Rich formatting
├── serialization/  # YAML, JSON, JSON-LD, Schema Card, Prompt exporters
├── owl/            # OWL/RDF export, reasoning, structured queries (rdflib)
├── llm/            # LLM integration (LiteLLM, OpenAI, Instructor)
├── chat/           # Natural language chat and workspace
├── tool/           # Data analysis → ontology building pipeline
├── graph/          # Graph backends (NetworkX, Neo4j)
├── domains/        # Domain templates (healthcare, e-commerce)
└── education/      # Ontology glossary for beginners

---

Examples

See the examples/ directory:

File	Description
quickstart.py	Build a Pet Store ontology with the Python API
hospital_surgery_booking.onto.yaml	Hospital surgery booking ontology
hospital_surgery_bookings.csv	Sample hospital surgery data
real_ecommerce_orders.csv	E-commerce orders dataset
bookstore.csv	Bookstore dataset
concept_university.csv	University concepts dataset

# Run the quickstart
python examples/quickstart.py

# Build from hospital data
onto tool build -i examples/hospital_surgery_bookings.csv

---

Running Tests

pip install -e ".[dev]"
pytest

---

License

MIT