We present KGMark, the first watermarking method for knowledge graph embeddings that ensures high detectability, transparency, and robustness across various graph modifications.
uv pip install -r requirements.txt- dataset paths:
datasets/* - use
notebooks/dataprocess.ipynbfor processing the raw triples to KGEs respectively
- check
notebooks/pipeline.ipynb
- check
method/baseline/