Toolboxes & Code Repositories

INK

Library for the generation of human-comprehensible binary feature representations from nodes within a knowledge graphs

software
Description:

INK is a knowledge graph representation method that generates human-comprehensible binary features for entities, bridging the gap between deep learning performance and model explainability. INK is built to be transparent: it transforms nodes of interest into interpretable feature vectors that can be used with classical machine learning techniques for classification tasks.

More Information:

  • Originates from the AI-Driven Data Science research challenge (FAIR 2019–2023).

  • Responsible Research Lead: Femke Ongenae

  • Authors: Bram Steenwinckel, Filip De Turck, Femke Ongenae

  • Programming Language: Python

  • Link to the repository

  • Categories: Networks and graphs

Publications:

  • Weyns, M., Blyau, T., Steenwinckel, B., De Turck, F., Van Hoecke, S., & Ongenae, F. (2025). Explainable knowledge graph embeddings for industrial process monitoring & control. INFORMATION FUSION, 123. https://doi.org/10.1016/j.inffus.2025.103242

  • Steenwinckel, B., Vandewiele, G., Weyns, M., Agozzino, T., De Turck, F., & Ongenae, F. (2022). INK : knowledge graph embeddings for node classification. DATA MINING AND KNOWLEDGE DISCOVERY, 36(2), 620–667. https://doi.org/10.1007/s10618-021-00806-z