Skip to main content

Advances, Systems and Applications

Table 3 Inference results on common datasets

From: Defect knowledge graph construction and application in multi-cloud IoT

Dataset

Metric

ComplEx

DistMult

ConvE

NeuralLP

DRUM

MINERVA

GRULR

 

HITS@1

0.754

0.808

0.697

0.475

0.183

0.605

0.568

KINSHIP

HITS@3

0.910

0.942

0.886

0.707

0.378

0.812

0.824

 

HITS@10

0.980

0.979

0.974

0.912

0.675

0.924

0.912

 

MRR

0.838

0.878

0.797

0.619

0.335

0.720

0.715

 

HITS@1

0.823

0.916

0.894

0.643

0.358

0.728

0.730

UMLS

HITS@3

0.962

0.967

0.964

0.869

0.699

0.900

0.893

 

HITS@10

0.995

0.992

0.992

0.962

0.854

0.968

0.963

 

MRR

0.894

0.944

0.933

0.778

0.548

0.825

0.814

 

HITS@1

0.410

0.390

0.400

0.368

0.369

0.413

0.412

WN18RR

HITS@3

0.460

0.440

0.440

0.386

0.388

0.456

0.471

 

HITS@10

0.510

0.490

0.520

0.408

0.410

0.513

0.522

 

MRR

0.440

0.430

0.430

0.381

0.382

0.448

0.450

 

HITS@1

0.158

0.155

0.237

0.173

0.174

0.217

0.245

FB15K-237

HITS@3

0.275

0.263

0.356

0.259

0.261

0.329

0.360

 

HITS@10

0.428

0.419

0.501

0.361

0.364

0.456

0.497

 

MRR

0.247

0.241

0.325

0.237

0.238

0.293

0.329

  1. Bold numbers are the results obtained by the best method in the table