Fig. 7From: BioBLP: a modular framework for learning on multimodal biomedical knowledge graphsDifference in Average MRR between RotatE and BioBLP-D predictions against node degree of disease (left) or non-disease (right) being predicted, together with the degree distribution of predicted entities. A negative \(\Delta\)MRR means that BioBLP-D outperforms RotatE, while a positive value means that RotatE has higher performanceBack to article page