Knowledge Graph Reasoning

This page contains benchmarks of knowledge graph reasoning methods.

We use the filtered ranking protocol for knowledge graph reasoning. For a test triplet ⟨h, r, t⟩, we rank it against all negative triplets ⟨h, r, t’⟩ or ⟨h’, r, t⟩ that do not appear in the knowledge graph. We report the mean rank (MR), mean reciprocal rank (MRR) and HITS at K (HITS@K) over the test set.

FB15k

FB15k	MR	MRR	HITS@1	HITS@3	HITS@10
`TransE`_	47	0.682	0.585	0.752	0.839
`DistMult`_	43	0.674	0.578	0.743	0.836
`ComplEx`_	38	0.673	0.577	0.738	0.842
`SimplE`_	45	0.689	0.603	0.747	0.836
`RotatE`_	58	0.764	0.703	0.806	0.867
`KBGAT`_	65	0.601	0.489	0.680	0.790
`NeuralLP`_	928	0.660	0.607	0.692	0.757

FB15k-237

FB15k-237	MR	MRR	HITS@1	HITS@3	HITS@10
`TransE`_	189	0.307	0.211	0.343	0.500
`DistMult`_	212	0.271	0.187	0.297	0.436
`ComplEx`_	206	0.271	0.187	0.297	0.437
`SimplE`_	212	0.268	0.185	0.293	0.433
`RotatE`_	202	0.316	0.223	0.349	0.507
`KBGAT`_	222	0.267	0.175	0.298	0.448
`NeuralLP`_	1943	0.259	0.196	0.281	0.384

WN18

WN18	MR	MRR	HITS@1	HITS@3	HITS@10
`TransE`_	297	0.738	0.637	0.809	0.927
`DistMult`_	668	0.744	0.592	0.888	0.932
`ComplEx`_	692	0.744	0.593	0.888	0.933
`SimplE`_	675	0.929	0.922	0.934	0.940
`RotatE`_	406	0.947	0.941	0.951	0.955
`KBGAT`_	231	0.743	0.636	0.830	0.933
`NeuralLP`_	7893	0.740	0.723	0.750	0.771

WN18RR

WN18RR	MR	MRR	HITS@1	HITS@3	HITS@10
`TransE`_	3071	0.220	0.015	0.389	0.529
`DistMult`_	4389	0.447	0.405	0.460	0.541
`ComplEx`_	4635	0.449	0.408	0.458	0.538
`SimplE`_	4935	0.472	0.436	0.488	0.540
`RotatE`_	3917	0.476	0.432	0.494	0.563
`KBGAT`_	3737	0.206	0.015	0.366	0.496
`NeuralLP`_	20950	0.346	0.320	0.354	0.400

Hetionet

Hetionet	MR	MRR	HITS@1	HITS@3	HITS@10
`TransE`_	1088	0.162	0.102	0.173	0.284
`DistMult`_	941	0.187	0.128	0.199	0.304
`ComplEx`_	800	0.235	0.166	0.255	0.374
`SimplE`_	893	0.194	0.134	0.207	0.313
`RotatE`_	744	0.257	0.185	0.282	0.403
`KBGAT`_	1713	0.058	0.023	0.055	0.130
`NeuralLP`_	4017	0.175	0.128	0.182	0.273