Inductive WN18RR and FB15k-237

Daniel Daza, Michael Cochez, Paul GrothZenodo 2021 · DOI · OpenAlex

Abstract

This repository contains knowledge graphs based on the WN18RR and FB15k-237 datasets. We generate new training, validation, and test splits for the <em>inductive</em> setting, where some entities are removed from the training set. The splits are used in the experiments described in the paper <strong>"Inductive Entity Representations from Text via Link Prediction"</strong>. To generate inductive splits, we remove nodes so that no other node becomes isolated, and the number of edges of a particular relation type does not drop below 100. The following are statistics for the datasets. <pre><code>| | WN18RR-ind | FB15k-237-ind | |-----------|------------|---------------| | Relations | 11 | 237 | | | Training | | Entities | 32,755 | 11,633 | | Triples | 69,585 | 215,082 | | | Validation | | Entities | 4,094 | 1,454 | | Triples | 11,381 | 42,164 | | | Test | | Entities | 4,094 | 1,454 | | Triples | 12,087 | 52,870 |</code></pre> The splits for each dataset are called ind-train.tsv, ind-dev.tsv, and ind-test.tsv. We also include textual descriptions for each entity, as well as type information.
BibTeX
@inproceedings{daza2021inductive,
  author = {Daniel Daza and Michael Cochez and Paul Groth},
  title = {{Inductive WN18RR and FB15k-237}},
  booktitle = {Zenodo},
  year = {2021},
  doi = {10.5281/zenodo.4501272},
}

Loading related content...

Figure