powerbi-ontology-extractor 0.1.1

Extract semantic intelligence from Power BI .pbix files and convert to formal ontologies

PowerBI Ontology Extractor

Transform 20 million Power BI dashboards into AI-ready ontologies

Installation • Quick Start • Features • Documentation • Contributing

---

🎯 The Problem

Enterprises have 20+ million Power BI semantic models that are actually informal ontologies trapped in proprietary .pbix files.

• The Challenge: Each Power BI model contains entities, relationships, and business logic—but AI agents can't access this semantic intelligence
• The Cost: Enterprises spend $50K-$200K per semantic definition to reconcile conflicts across dashboards
• The $4.6M Mistake: A logistics company lost $4.6M when an AI agent used a renamed column (Warehouse_Location → FacilityID) because there was no semantic binding validation

💡 The Solution

PowerBI Ontology Extractor unlocks the hidden ontologies in your Power BI dashboards and transforms them into formal, AI-ready ontologies.

┌─────────────────────┐     ┌──────────────────────┐     ┌─────────────────────────────┐
│   Power BI .pbix    │────▶│  Ontology Extractor  │────▶│       OntoGuard             │
│  (20M+ dashboards)  │     │  (this project)      │     │  Semantic Firewall          │
└─────────────────────┘     └──────────────────────┘     └─────────────────────────────┘
                                     │                              │
                                     │ OWL/Fabric IQ                │ Semantic Validation
                                     ▼                              ▼
                            ┌──────────────────────┐     ┌─────────────────────────────┐
                            │   Semantic Contract  │────▶│  Universal Agent Connector  │
                            │   (permissions)      │     │  AI Agent Infrastructure    │
                            └──────────────────────┘     └─────────────────────────────┘
                                                                    │
                                                                    ▼
                                                         ┌─────────────────────────────┐
                                                         │       AI Agents             │
                                                         │  (Claude, GPT, etc.)        │
                                                         └─────────────────────────────┘

30-minute workflow:

Power BI (.pbix) → Ontology Extractor → OntoGuard → Universal Agent Connector → AI Agent
     10 min           10 min            5 min            3 min               2 min

---

🚀 Quick Start

Installation

# Install from PyPI (recommended)
pip install powerbi-ontology-extractor

Or install from source:

git clone https://github.com/vpakspace/powerbi-ontology-extractor.git
cd powerbi-ontology-extractor
pip install -e .

Basic Usage

from powerbi_ontology import PowerBIExtractor, OntologyGenerator

# Step 1: Extract semantic model from Power BI
extractor = PowerBIExtractor("path/to/dashboard.pbix")
semantic_model = extractor.extract()

# Step 2: Generate formal ontology
generator = OntologyGenerator(semantic_model)
ontology = generator.generate()

print(f"✅ Extracted {len(ontology.entities)} entities")
print(f"✅ Found {len(ontology.relationships)} relationships")
print(f"✅ Generated {len(ontology.business_rules)} business rules")

# Step 3: Export to OWL for OntoGuard
from powerbi_ontology.export import OWLExporter

exporter = OWLExporter(ontology)
exporter.save("ontology.owl")

Visual Ontology Editor (No-Code UI)

# Start Streamlit UI
streamlit run ontology_editor.py --server.port 8503

Features:

• 📂 Load from .pbix files or JSON
• 📦 Edit entities with properties and constraints
• 🔗 Manage relationships between entities
• 🔐 Configure permission matrix (RBAC)
• 📜 Add business rules with classification
• 🦉 Preview and export OWL
• 🔀 Diff & Merge ontology versions
• 💬 AI Chat - Ask questions about your ontology!

---

🔥 Key Features

1. Automatic Extraction (PBIXRay)

• ✅ Reads Power BI .pbix files (binary DataModel via PBIXRay)
• ✅ Extracts tables, columns, relationships, hierarchies
• ✅ Parses DAX measures and calculated columns
• ✅ Captures Row-Level Security (RLS) rules
• ✅ Fallback to JSON model.bim for legacy files

2. DAX to Business Rules

• ✅ Parses DAX formulas automatically
• ✅ Extracts conditional logic (IF, SWITCH, CALCULATE)
• ✅ Converts filters to business rules
• ✅ Classifies measure types (aggregation, conditional, time intelligence)

3. Ontology Generation (70% Automated)

• ✅ Entities from tables
• ✅ Properties from columns (with data types)
• ✅ Relationships from foreign keys (with cardinality)
• ✅ Business rules from DAX measures
• ✅ Constraints (required, unique, range, regex, enum)
• ✅ Pattern detection (date tables, dimensions, facts)

4. Multi-Format Export

Format	Use Case
OWL/RDF	OntoGuard semantic validation
Fabric IQ	Microsoft Fabric deployment
JSON	Universal agent connector
Semantic Contract	Role-based AI agent permissions

5. Schema Drift Detection (Prevents $4.6M Mistakes!)

• ✅ Validates schema bindings
• ✅ Detects column renames/deletions
• ✅ Type normalization (varchar→text, int→integer)
• ✅ Severity levels: CRITICAL, WARNING, INFO
• ✅ Auto-fix suggestions

6. Multi-Dashboard Semantic Debt Analysis

• ✅ Analyzes multiple Power BI dashboards
• ✅ Detects conflicting definitions ("Revenue" defined differently)
• ✅ 5 conflict types: MEASURE, TYPE, ENTITY, RELATIONSHIP, RULE
• ✅ Generates consolidation reports

7. Ontology Diff & Merge

• ✅ Git-like diff between ontology versions
• ✅ Detect added/removed/modified elements
• ✅ Three-way merge (base, ours, theirs)
• ✅ Conflict detection and resolution strategies

8. Collaborative Review Workflow

• ✅ Comments on entities/properties/rules
• ✅ Reply and resolve threads
• ✅ Approval workflow: draft → review → approved → published
• ✅ Audit trail of all actions

9. CLI Tool for Automation

# Extract single .pbix file
pbix2owl extract -i dashboard.pbix -o ontology.owl

# Batch process directory (8 parallel workers)
pbix2owl batch -i ./dashboards/ -o ./ontologies/ -w 8 --recursive

# Analyze semantic debt
pbix2owl analyze -i ./ontologies/ -o report.md

# Compare versions (diff)
pbix2owl diff -s v1.json -t v2.json -o changelog.md

10. AI-Powered Ontology Chat 🆕

• ✅ Ask questions about loaded ontology in natural language
• ✅ OpenAI API integration (gpt-4o-mini)
• ✅ Role-based context (Admin/Analyst/Viewer)
• ✅ Bilingual support (Russian/English)
• ✅ Suggested questions based on ontology content

Example questions:

• "What entities exist in the ontology?"
• "How are Customer and Sales related?"
• "Show all DAX measures"
• "What permissions does Analyst role have?"

---

📊 Real-World Example

Tested with Microsoft official samples:

File	Size	Entities	Relationships	DAX Measures	OWL Triples
Sales_Returns_Sample.pbix	6.3 MB	15	9	58	1,734
Adventure_Works_DW_2020.pbix	7.8 MB	11	13	0	1,083

from powerbi_ontology import PowerBIExtractor, OntologyGenerator
from powerbi_ontology.export import OWLExporter

# Extract from Power BI
extractor = PowerBIExtractor("Sales_Returns_Sample.pbix")
model = extractor.extract()

# Generate ontology
ontology = OntologyGenerator(model).generate()

# Export to OWL (for OntoGuard)
exporter = OWLExporter(ontology, default_roles=["Admin", "Analyst", "Viewer"])
exporter.save("sales_ontology.owl")

# Summary
summary = exporter.get_export_summary()
print(f"Classes: {summary['classes']}")
print(f"Properties: {summary['datatype_properties']}")
print(f"Action Rules: {summary['action_rules']}")  # CRUD per entity × role

---

🔗 Integration Ecosystem

OntoGuard (Semantic Firewall)

from powerbi_ontology.export import OWLExporter

exporter = OWLExporter(ontology)
exporter.save("ontology.owl")

# Use with OntoGuard for AI agent validation
# github.com/vpakspace/ontoguard-ai

Universal Agent Connector (MCP)

from powerbi_ontology import ContractBuilder
from powerbi_ontology.export import ContractToOWLConverter

# Create semantic contract for AI agent
builder = ContractBuilder(ontology)
contract = builder.build_contract(
    agent_name="SalesAnalyst",
    permissions={
        "read": ["Customer", "Sales", "Product"],
        "write": {"Sales": ["Status"]},
        "execute": ["GenerateReport"]
    }
)

# Export for MCP
converter = ContractToOWLConverter(contract)
converter.save("sales_agent_contract.owl")

# Use with Universal Agent Connector
# github.com/vpakspace/universal-agent-connector

Microsoft Fabric IQ

from powerbi_ontology.export import FabricIQExporter

exporter = FabricIQExporter(ontology)
fabric_json = exporter.export()

# Deploy as Ontology Item to OneLake

---

🧪 Testing

# Run all tests (340 tests, 82% coverage)
pytest

# Run with coverage report
pytest --cov=powerbi_ontology --cov-report=html

# Run specific test module
pytest tests/test_owl_exporter.py -v

Test Statistics:

• 340 tests passing
• 82% code coverage
• E2E tests with real .pbix files
• OntoGuard integration tests

---

📁 Project Structure

powerbi-ontology-extractor/
├── powerbi_ontology/
│   ├── __init__.py
│   ├── extractor.py           # PowerBIExtractor
│   ├── ontology_generator.py  # OntologyGenerator
│   ├── pbix_reader.py         # PBIXRay integration
│   ├── dax_parser.py          # DAX formula parsing
│   ├── semantic_debt.py       # Multi-dashboard analysis
│   ├── ontology_diff.py       # Diff & Merge
│   ├── review.py              # Collaborative review
│   ├── chat.py                # AI Chat (OpenAI)
│   ├── cli.py                 # CLI commands
│   ├── mcp_server.py          # MCP Server for Claude Code
│   ├── export/
│   │   ├── owl.py             # OWL/RDF export
│   │   ├── fabric_iq.py       # Fabric IQ export
│   │   ├── fabric_iq_to_owl.py
│   │   └── contract_to_owl.py
│   └── utils/
│       ├── visualizer.py
│       └── validators.py
├── ontology_editor.py         # Streamlit UI (1300+ lines)
├── examples/
│   ├── sample_pbix/           # Microsoft official samples
│   └── sample_ontology.json
├── tests/                     # 340 tests
├── requirements.txt
└── README.md

---

📊 Project Status

Feature	Status	Coverage
PBIX Extraction (PBIXRay)	✅ Complete	51%
DAX Parser	✅ Complete	73%
Ontology Generator	✅ Complete	83%
OWL Exporter	✅ Complete	95%
Fabric IQ Exporter	✅ Complete	97%
Contract Builder	✅ Complete	98%
Schema Drift Detection	✅ Complete	84%
Semantic Debt Analysis	✅ Complete	84%
Ontology Diff & Merge	✅ Complete	84%
Review Workflow	✅ Complete	93%
CLI Tool	✅ Complete	60%
MCP Server (Claude Code)	✅ Complete	85%
Visual Editor (Streamlit)	✅ Complete	-
AI Chat (OpenAI)	✅ Complete	-

Overall: 370 tests, 82% coverage

PyPI: https://pypi.org/project/powerbi-ontology-extractor/

---

🤖 MCP Server (Claude Code Integration)

Use PowerBI Ontology Extractor directly in Claude Code via MCP protocol.

Setup

1. Install the package:

pip install powerbi-ontology-extractor

2. Add to ~/.claude.json:

{
  "mcpServers": {
    "powerbi-ontology": {
      "command": "python",
      "args": ["-m", "powerbi_ontology.mcp_server"]
    }
  }
}

Optional: Add "env": {"OPENAI_API_KEY": "..."} for AI chat feature.

3. Restart Claude Code

Available MCP Tools

Tool	Description
pbix_extract	Extract semantic model from .pbix file
ontology_generate	Generate ontology from model data
export_owl	Export to OWL format (xml/turtle)
export_json	Export to JSON format
analyze_debt	Analyze semantic debt across ontologies
ontology_diff	Compare two ontology versions
ontology_merge	Merge ontologies (three-way)
ontology_chat_ask	AI Q&A about ontology

Usage Examples in Claude Code

# Extract and generate ontology
"Extract ontology from sales.pbix and export to OWL"

# Ask questions about ontology
"What entities are in the Sales_Returns ontology?"

# Compare versions
"Compare v1 and v2 ontologies and show differences"

---

🛠️ Development Setup

# Clone repository
git clone https://github.com/vpakspace/powerbi-ontology-extractor.git
cd powerbi-ontology-extractor

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt
pip install -e .

# Run tests
pytest

# Start Streamlit UI
streamlit run ontology_editor.py --server.port 8503

Environment Variables

Create .env file for AI Chat:

# Required for Ontology Chat
OPENAI_API_KEY=your-openai-api-key

# Optional: Model selection (default: gpt-4o-mini)
# OPENAI_MODEL=gpt-4o-mini

# Optional: Local models via Ollama
# OLLAMA_BASE_URL=http://localhost:11434/v1

---

🤝 Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

Ways to contribute:

• 🐛 Report bugs via GitHub Issues
• 💡 Suggest features
• 📝 Improve documentation
• 🔧 Submit pull requests
• ⭐ Star the repository

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🔗 Related Projects

Project	Description
OntoGuard AI	Semantic Firewall for AI Agents
Universal Agent Connector	MCP Infrastructure + Streamlit UI

---

📞 Contact

• 🐛 Issues: GitHub Issues
• 💬 Discussions: GitHub Discussions

---

Ready to unlock the semantic intelligence in your Power BI dashboards? 🚀

git clone https://github.com/vpakspace/powerbi-ontology-extractor.git
cd powerbi-ontology-extractor
pip install -r requirements.txt
streamlit run ontology_editor.py

Star ⭐ this repo if you find it useful!