UCKG

Unified Cybersecurity Knowledge Graph

Software Requirements

• Docker Desktop - https://www.docker.com/products/docker-desktop/

Usage

• First, pull the repository into your file system

$  git clone https://github.com/edogdu/UCKG.git ./UCKG

• Next, switch to root directory of project

$  cd UCKG

• Lastly, run docker-compose to start project

$  docker-compose up --build

• To shutdown the UCKG, follow these steps
  • type Ctrl+C to stop server
  • use docker-compose to clean up images

$  docker-compose down

Configuration Options

GPU Acceleration (Optional)

The embedding generation process can utilize NVIDIA GPUs for faster processing:

• To run with GPU (default): The GPU configuration is enabled by default. Requires NVIDIA drivers and Docker GPU support.

• To disable GPU support: Comment out the GPU-related lines in docker-compose.yml if you don't have an NVIDIA GPU:

  # In the ollama service section, comment out these lines:
  environment:
    # - NVIDIA_VISIBLE_DEVICES=all
    # - NVIDIA_DRIVER_CAPABILITIES=compute,utility
  # deploy:
  #   resources:
  #     reservations:
  #       devices:
  #         - driver: nvidia
  #           count: all
  #           capabilities: [gpu]

Embedding Process Control

The system includes semantic embedding generation for enhanced search capabilities:

• To enable embeddings (default): The embedding generation is enabled by default with EMBED_ENV: "true".

• To disable embeddings: Change EMBED_ENV to "false" in docker-compose.yml to skip embedding generation and save processing time:

  # In uckg-scripts service environment section:
  uckg-scripts:
    environment:
      EMBED_ENV: "false"  # Change from "true" to "false" to disable

Note: Disabling embeddings will skip the semantic search capabilities but the core knowledge graph functionality will still work.

Resources

• A copy of our paper outlining the project is available in the root directory as uckg_paper.pdf
• A web-based visualization of the Unified Cybersecurity Ontology can be accessed at this url: https://service.tib.eu/webvowl/#iri=http://purl.org/cyber/uco
• Docker has a tendency to have hanging resources that can take up alot of diskspace, I found the following commands useful
  • In Docker Desktop Application, ensure no containers are running
  • In Command Line or Bash run Docker prune commands

  $  docker image prune -f

  $  docker builder prune

  • In Docker Desktop Application, navigate to Troubleshoot section (bug button) and select Clean/Purge data
  • If diskspace utilization is not changing after running these steps, try restarting your computer

License

This project is licensed under the MIT License.