load_fb15k_237

ampligraph.datasets.load_fb15k_237(check_md5hash=False, clean_unseen=True, add_reciprocal_rels=False, return_mapper=False)

Load the FB15k-237 dataset (with option to load human labeled test subset).

FB15k-237 is a reduced version of FB15K. It was first proposed by [TCP+15].

Warning

FB15K-237’s validation set contains 8 unseen entities over 9 triples. The test set has 29 unseen entities, distributed over 28 triples.

The FB15k-237 dataset is loaded from file if it exists at the AMPLIGRAPH_DATA_HOME location. If AMPLIGRAPH_DATA_HOME is not set, the default ~/ampligraph_datasets is checked. If the dataset is not found at either location, it is downloaded and placed in AMPLIGRAPH_DATA_HOME or ~/ampligraph_datasets.

The dataset is divided in three splits:

  • train: 272,115 triples

  • valid: 17,535 triples

  • test: 20,466 triples

It also contains a subset of the test set with human-readable labels, available here:

  • test-human

  • test-human-ids

Dataset

Train

Valid

Test

Test-Human

Entities

Relations

FB15K-237

272,115

17,535

20,466

273

14,541

237
Parameters:

  • check_md5hash (bool) – If True check the md5hash of the files (default: False).

  • clean_unseen (bool) – If True, filters triples in validation and test sets that include entities not present in the training set.

  • add_reciprocal_rels (bool) – Flag which specifies whether to add reciprocal relations. For every <s, p, o> in the dataset this creates a corresponding triple with reciprocal relation <o, p_reciprocal, s> (default: False).

  • return_mapper (bool) – Whether to return human-readable labels in a form of dictionary in X["mapper"] field (default: False).

Returns:

splits – The dataset splits: {‘train’: train, ‘valid’: valid, ‘test’: test, ‘test-human’:test_human, ‘test-human-ids’: test_human_ids}. Each split is a ndarray of shape (n, 3).

Return type:

dict

Example

>>> from ampligraph.datasets import load_fb15k_237
>>> X = load_fb15k_237()
>>> X["train"][2]
array(['/m/07s9rl0', '/media_common/netflix_genre/titles', '/m/0170z3'],
  dtype=object)