KG Link Prediction with hierarchical class information

Datasets from the Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction [1] paper presented at NeSy 2026. Here versions of CoDEx [2] and BioKG [3] augmented with class hierarchies are provided. For the AIFB dataset used in the paper, and further instructions and descriptions, go to our GitHub repo. If you find this useful, please cite our paper below.

References:
[1] Kronström, F. and King R.D. (2026). Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction. https://doi.org/10.48550/arXiv.2608.22981
[2] Safavi, T. and Koutra, D. (2020). CoDEx: A Comprehensive Knowledge Graph Completion Benchmark. https://doi.org/10.18653/v1/2020.emnlp-main.669
[3] Walsh, B., Mohamed, S. K., and Nováček, V. (2020). BioKG: A Knowledge Graph for Relational Learning on Biological Data. https://doi.org/10.1145/3340531.3412776

How to cite:
@misc{kronström2026hierarchyawaresemanticlossesknowledge,
      title={Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction}, 
      author={Filip Kronström and Ross D. King},
      year={2026},
      eprint={2608.22981},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2608.22981}, 
}