Link Prediction On Wn18

評価指標

Hits@1

Hits@10

Hits@3

MRR

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名	Hits@1	Hits@10	Hits@3	MR	MRR	Paper Title	Repository
QuatE	0.945	0.959	0.954	162	0.95	Quaternion Knowledge Graph Embeddings	-
ComplEx NSCaching	-	0.9398	-	1072	0.9355	NSCaching: Simple and Efficient Negative Sampling for Knowledge Graph Embedding	-
Quaternion	0.931	0.956	0.950	-	0.941	Analyzing Knowledge Graph Embedding Methods from a Multi-Embedding Interaction Perspective	-
pRotatE	0.942	0.957	0.950	254	0.947	RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space	-
RotatE	0.944	0.959	0.952	309	0.949	RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space	-
SimplE	0.944	0.954	0.950	412	0.948	GraphVite: A High-Performance CPU-GPU Hybrid System for Node Embedding	-
LineaRE	0.947	0.961	0.955	170	0.952	LineaRE: Simple but Powerful Knowledge Graph Embedding for Link Prediction	-
DistMult (after variational EM)	-	-	-	-	0.911	Augmenting and Tuning Knowledge Graph Embeddings	-
MEI-BTD	0.946	0.957	0.952	-	0.950	Multi-Partition Embedding Interaction with Block Term Format for Knowledge Graph Completion	-
ConvR	0.947	0.958	0.955	-	0.951	Adaptive Convolution for Multi-Relational Learning	-
ConvE	0.935	0.956	0.946	374	0.943	Convolutional 2D Knowledge Graph Embeddings	-
HypER	0.947	0.958	0.955	-	0.951	Hypernetwork Knowledge Graph Embeddings	-
KBGAN (TransD + ComplEx)	-	0.948	-	-	0.779	KBGAN: Adversarial Learning for Knowledge Graph Embeddings	-
TuckER	0.949	0.958	0.955	-	0.953	TuckER: Tensor Factorization for Knowledge Graph Completion	-
LibKGE (ruffinelli2020you)	-	-	-	-	-	PyKEEN 1.0: A Python Library for Training and Evaluating Knowledge Graph Embeddings	-
QuatDE	0.944	0.961	0.954	120	0.95	QuatDE: Dynamic Quaternion Embedding for Knowledge Graph Completion	-
HolE	0.930	0.949	0.945	-	0.938	Holographic Embeddings of Knowledge Graphs	-
TransD	-	0.922	-	212	-	-	-
HHolE	.931	.951	.945	183	.939	Augmenting Compositional Models for Knowledge Base Completion Using Gradient Representations	-
ParTransH	-	0.668	-	215	-	Efficient Parallel Translating Embedding For Knowledge Graphs	-

0 of 37 row(s) selected.