Table 2 The metrics of link prediction results for different models on integrated KG, ADInt, by time slicing evaluation.

From: Repurposing non-pharmacological interventions for Alzheimer's disease through link prediction on biomedical literature

 

TransE

RotatE

DistMult

ComplEX

RGCN

CompGCN

Hits@1

0.1770

0.1786

0.1109

0.1062

0.2656

0.1520

Hits@3

0.3242

0.3055

0.2586

0.2467

0.5058

0.3227

Hits@10

0.5996

0.5340

0.5921

0.5854

0.8496

0.6585

MRR

0.3109

0.2987

0.2547

0.2479

0.4390

0.3033

MR

8.8607

10.1095

9.2785

9.3799

5.2054

7.8198

  1. The best values for each metric are in [bold].
Back to article page