SemKGE

A python package that extends LibKGE with methods for learning Knowledge Graph Embeddings that include Semantics.

Jupyter Notebook

The Jupyter Notebook also serves as an example for the installation process and how to use the package .

Implemented Methods

  • TransT

Results

FB15K-237 (Freebase)

MRR Mean Rank Hits@1 Hits@3 Hits@10 Config file
TransE 0.313 - 0.221 0.347 0.497 config.yaml
TransT-type (ours/best) 0.324 159 0.231 0.357 0.517 todo
TransT-type-only (ours/best) todo

WN18RR (Wordnet)

MRR Mean Rank Hits@1 Hits@3 Hits@10 Config file
TransE 0.228 - 0.053 0.368 0.520 config.yaml
TransT-type (ours/best) 0.191 4617 0.012 0.358 0.438 todo
TransT-type-only (ours/best) todo

FB15K (Freebase)

MRR Mean Rank Hits@1 Hits@3 Hits@10 Config file
TransE 0.676 - 0.542 0.787 0.875 config.yaml
TransE (Ma et al.) - 143 - - 0.621 -
TransT-type (ours/best) todo
TransT-type (ours/repro) todo
TransT-multiple (ours/repro) todo
TransT-multiple-type (ours/repro) todo
TransT-type (Ma et al.) - 72 - - 0.823 -
TransT-multiple (Ma et al.) - 62 - - 0.836 -
TransT-multiple-type (Ma et al.) - 46 - - 0.854 -

WN18 (Wordnet)

MRR Mean Rank Hits@1 Hits@3 Hits@10 Config file
TransE 0.553 - 0.315 0.764 0.924 config.yaml
TransE (Ma et al.) - 251 - - 0.892 -
TransT-type (ours/repro) todo
TransT-type (ours/best) todo
TransT-multiple-type (Ma et al.) - 130 - - 0.974 -