cognee-mcp

{ "cells": [ { "cell_type": "markdown", "id": "25cf0a40e669a70", "metadata": {}, "source": [ "# Understanding Ontologies with Cognee\n", "\n", "This notebook demonstrates how to work with ontologies in scientific research using the Cognee framework. We'll explore how ontologies can enhance our understanding and querying of scientific papers.\n", "\n", "## What is an Ontology?\n", "\n", "An ontology is a formal representation of knowledge that defines:\n", "- Concepts within a domain\n", "- Relationships between concepts\n", "- Properties and attributes\n", "- Rules and constraints\n", "\n", "Key terms:\n", "- **Classes**: Categories or types (e.g., Disease, Symptom)\n", "- **Instances**: Specific examples of classes (e.g., Type 2 Diabetes)\n", "- **Properties**: Relationships between classes/instances (e.g., hasSymptom)\n", "- **Axioms**: Logical statements defining relationships" ] }, { "cell_type": "markdown", "id": "441248da37f2b901", "metadata": {}, "source": [ "## Setup\n", "\n", "First, let's install the required packages and set up our environment:" ] }, { "cell_type": "code", "id": "8cf7ba29f9a150af", "metadata": {}, "source": [ "# Install required package\n", "# !pip install cognee" ], "outputs": [], "execution_count": null }, { "cell_type": "code", "id": "abb86851", "metadata": {}, "source": [ "import os\n", "\n", "# Set up OpenAI API key (required for Cognee's LLM functionality)\n", "if \"LLM_API_KEY\" not in os.environ:\n", " os.environ[\"LLM_API_KEY\"] = \"your-api-key-here\" # Replace with your API key" ], "outputs": [], "execution_count": null }, { "cell_type": "code", "id": "d825d126b3a0ec26", "metadata": {}, "source": [ "# Import required libraries\n", "import cognee\n", "from cognee.shared.logging_utils import get_logger\n", "\n", "cognee.config.set_llm_model(\"gpt-4o-mini\")\n", "cognee.config.set_llm_provider(\"openai\")\n", "from cognee.api.v1.search import SearchType\n", "\n", "logger = get_logger()" ], "outputs": [], "execution_count": null }, { "cell_type": "markdown", "id": "6af350837e86b7a1", "metadata": {}, "source": [ "## Creating the Pipeline\n", "\n", "Let's create a pipeline that will:\n", "1. Clean existing data\n", "2. Process scientific papers\n", "3. Apply ontological knowledge" ] }, { "cell_type": "code", "id": "4d0e4a58e4207a7d", "metadata": {}, "source": [ "async def run_pipeline(config=None):\n", " # Clean existing data\n", " await cognee.prune.prune_data()\n", " await cognee.prune.prune_system(metadata=True)\n", " \n", " # Set up path to scientific papers\n", " scientific_papers_dir = os.path.join(\n", " os.path.dirname(os.path.dirname(os.path.abspath(\".\"))), \n", " \"cognee\",\n", " \"examples\",\n", " \"data\", \n", " \"scientific_papers/\"\n", " )\n", " \n", " # Add papers to the system\n", " await cognee.add(scientific_papers_dir)\n", " \n", " # Cognify with optional ontology\n", " return await cognee.cognify(config=config)\n", "\n", "async def query_pipeline(questions):\n", " answers = []\n", " for question in questions:\n", " search_results = await cognee.search(\n", " query_type=SearchType.GRAPH_COMPLETION,\n", " query_text=question,\n", " )\n", " answers.append(search_results)\n", " return answers" ], "outputs": [], "execution_count": null }, { "cell_type": "markdown", "id": "c87c21a75d6f4d79", "metadata": {}, "source": [ "## Running the Demo\n", "\n", "Let's test our system with some medical questions, comparing results with and without ontological knowledge:" ] }, { "cell_type": "code", "id": "1363772d2b48f5c0", "metadata": {}, "source": [ "from cognee.modules.ontology.rdf_xml.RDFLibOntologyResolver import RDFLibOntologyResolver\n", "from cognee.modules.ontology.ontology_config import Config\n", "# Test questions\n", "questions = [\n", " \"What are common risk factors for Type 2 Diabetes?\",\n", " \"What preventive measures reduce the risk of Hypertension?\",\n", " \"What symptoms indicate possible Cardiovascular Disease?\",\n", " \"What diseases are associated with Obesity?\"\n", "]\n", "\n", "# Path to medical ontology\n", "ontology_path = \"../examples/python/ontology_input_example/enriched_medical_ontology_with_classes.owl\" # Update with your ontology path\n", "\n", "config: Config = {\n", " \"ontology_config\": {\n", " \"ontology_resolver\": RDFLibOntologyResolver(ontology_file=ontology_path)\n", " }\n", " }\n", "\n", "# Run with ontology\n", "print(\"\\n--- Results WITH ontology ---\\n\")\n", "await run_pipeline(config=config)\n", "answers_with = await query_pipeline(questions)\n", "for q, a in zip(questions, answers_with):\n", " print(f\"Q: {q}\\nA: {a}\\n\")" ], "outputs": [], "execution_count": null }, { "cell_type": "code", "id": "3aa18f4cdd5ceff6", "metadata": {}, "source": [ "# Run without ontology\n", "print(\"\\n--- Results WITHOUT ontology ---\\n\")\n", "await run_pipeline()\n", "answers_without = await query_pipeline(questions)\n", "for q, a in zip(questions, answers_without):\n", " print(f\"Q: {q}\\nA: {a}\\n\")" ], "outputs": [], "execution_count": null }, { "cell_type": "markdown", "id": "c60533d2423acdb0", "metadata": {}, "source": [ "## Visualizing the Knowledge Graph\n", "\n", "Let's visualize how our ontology connects different medical concepts:" ] }, { "cell_type": "code", "id": "36ee2a360f47a054", "metadata": {}, "source": [ "import webbrowser\n", "import os\n", "from cognee.api.v1.visualize.visualize import visualize_graph\n", "html = await visualize_graph()\n", "home_dir = os.path.expanduser(\"~\")\n", "html_file = os.path.join(home_dir, \"graph_visualization.html\")\n", "display(html_file)\n", "webbrowser.open(f\"file://{html_file}\")" ], "outputs": [], "execution_count": null }, { "cell_type": "markdown", "id": "ff39326921b75273", "metadata": {}, "source": [ "## Understanding the Results\n", "\n", "The demonstration above shows how ontologies enhance our analysis by:\n", "\n", "1. **Making Connections**: \n", " - Linking related medical concepts even when not explicitly stated\n", " - Identifying relationships between symptoms, diseases, and risk factors\n", "\n", "2. **Standardizing Terms**: \n", " - Unifying different ways of referring to the same medical condition\n", " - Ensuring consistent terminology across documents\n", "\n", "3. **Enabling Inference**: \n", " - Drawing conclusions based on ontological relationships\n", " - Discovering implicit connections in the data\n", "\n", "## Next Steps\n", "\n", "To learn more about Cognee and ontologies:\n", "1. Check out the [Cognee documentation](https://docs.cognee.ai/)\n", "2. Explore more examples in the `examples` directory\n", "3. Try creating your own domain-specific ontology\n", "\n", "Remember to:\n", "- Place your scientific papers in the appropriate directory\n", "- Update the ontology path to point to your .owl file\n", "- Replace the API key with your own OpenAI key" ] }, { "cell_type": "code", "id": "8d2a0fe555a7bc0f", "metadata": {}, "source": [ "# Only exit in interactive mode, not during GitHub Actions\n", "import os\n", "\n", "# Skip exit if we're running in GitHub Actions\n", "if not os.environ.get('GITHUB_ACTIONS'):\n", " print(\"Exiting kernel to clean up resources...\")\n", " os._exit(0)\n", "else:\n", " print(\"Skipping kernel exit - running in GitHub Actions\")" ], "outputs": [], "execution_count": null }, { "metadata": {}, "cell_type": "code", "source": "", "id": "adb6601890237b6a", "outputs": [], "execution_count": null } ], "metadata": { "kernelspec": { "display_name": ".venv", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.12.7" } }, "nbformat": 4, "nbformat_minor": 5 }