Papers › Rot-Pro: Modeling Transitivity by Projection in Knowledge Graph Embedding

Rot-Pro: Modeling Transitivity by Projection in Knowledge Graph Embedding

27 Oct 2021NeurIPS 2021 12arXiv:2110.14450archive 2025-07-28

Tengwei Song, Jie Luo, Lei Huang

Knowledge graph embedding models learn the representations of entities and relations in the knowledge graphs for predicting missing links (relations) between entities. Their effectiveness are deeply affected by the ability of modeling and inferring different relation patterns such as symmetry, asymmetry, inversion, composition and transitivity. Although existing models are already able to model many of these relations patterns, transitivity, a very common relation pattern, is still not been fully supported. In this paper, we first theoretically show that the transitive relations can be modeled with projections. We then propose the Rot-Pro model which combines the projection and relational rotation together. We prove that Rot-Pro can infer all the above relation patterns. Experimental results show that the proposed Rot-Pro model effectively learns the transitivity pattern and achieves the state-of-the-art results on the link prediction task in the datasets containing transitive relations.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

tewiSong/Rot-Pro officialmentioned on GitHubpytorchBSD-2-Clause report
Tigter/ogblwikikg2-RotPro mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Graph EmbeddingKnowledge Graph EmbeddingKnowledge GraphsLink Prediction

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Link Prediction FB15k-237 Rot-Pro Hits@1 0.246 #41 of 75 Archive leaderboard report
Link Prediction FB15k-237 Rot-Pro Hits@10 0.540 #41 of 75 Archive leaderboard report
Link Prediction FB15k-237 Rot-Pro Hits@3 0.383 #41 of 75 Archive leaderboard report
Link Prediction FB15k-237 Rot-Pro MRR 0.344 #41 of 75 Archive leaderboard report
Link Prediction WN18RR Rot-Pro Hits@1 0.397 #32 of 75 Archive leaderboard report
Link Prediction WN18RR Rot-Pro Hits@10 0.577 #32 of 75 Archive leaderboard report
Link Prediction WN18RR Rot-Pro Hits@3 0.482 #32 of 75 Archive leaderboard report
Link Prediction WN18RR Rot-Pro MRR 0.457 #32 of 75 Archive leaderboard report
Link Prediction YAGO3-10 Rot-Pro Hits@1 0.443 #13 of 18 Archive leaderboard report
Link Prediction YAGO3-10 Rot-Pro Hits@10 0.699 #13 of 18 Archive leaderboard report
Link Prediction YAGO3-10 Rot-Pro Hits@3 0.596 #13 of 18 Archive leaderboard report
Link Prediction YAGO3-10 Rot-Pro MRR 0.542 #13 of 18 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 Rot-Pro Ext. data No #19 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 Rot-Pro Number of params 1000669602 #19 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 Rot-Pro Test MRR 0.5602 ± 0.0016 #19 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 Rot-Pro Validation MRR 0.5740 ± 0.0008 #19 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 RotPro Ext. data No #23 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 RotPro Number of params 1000669602 #23 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 RotPro Test MRR 0.4277 ± 0.0008 #23 of 30 Archive leaderboard report
Link Property Prediction ogbl-wikikg2 RotPro Validation MRR 0.4174 ± 0.0058 #23 of 30 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections