Table 3 Median WAF scores for the combinations of knowledge graph embeddingss with Cosine similarity, RF or XGB for the different knowledge graphs using the Hadamard operator. Best result for each knowledge graph embeddings and machine learning algorithm or CS is bold. Results that are statistically significantly different when compared to HPf are underlined

CS

HPf

0.687

0.671

0.664

0.699

HPf+GO

0.682

0.670

0.660

0.678

HPs+GO+LD

0.689

0.677

0.656

0.701

HPs+GO+Map

0.682

0.667

0.668

0.693

HPs+GO+LD+Map

0.681

0.676

0.669

0.695

RF

HPf

0.737

0.756

0.690

0.727

HPf+GO

0.753

0.761

0.696

0.717

HPs+GO+LD

0.749

0.770

0.716

0.729

HPs+GO+Map

0.745

0.771

0.703

0.728

HPs+GO+LD+Map

0.742

0.775

0.711

0.721

XGB

HPf

0.732

0.748

0.689

0.728

HPf+GO

0.743

0.758

0.690

0.716

HPs+GO+LD

0.737

0.768

0.706

0.734

HPs+GO+Map

0.735

0.765

0.698

0.727

HPs+GO+LD+Map

0.733

0.765

0.702

0.726