Evaluator

class mowl.evaluation.Evaluator(dataset, device='cpu', batch_size=16)[source]

Bases: object

Base evaluation class for ontology embedding methods.

Parameters:

  • dataset (mowl.datasets.base.Dataset) – mOWL dataset object. Required to obtain the ontology entities (classes, individuals, object properties, etc.).

  • device (str, optional) – Device to use for the evaluation. Defaults to ‘cpu’.

  • batch_size (int, optional) – Batch size for evaluation. Defaults to 16.

Changed in version 1.0.0: Updated Evaluator with a new API.

Attributes Summary

deductive_closure_tuples

Methods Summary

create_tuples(ontology)

evaluate(*args[, include_deductive_closure, ...])

evaluate_base(model, eval_tuples[, mode, ...])

get_logits(batch)

Attributes Documentation

deductive_closure_tuples

Methods Documentation

create_tuples(ontology)[source]
evaluate(*args, include_deductive_closure=False, exclude_testing_set=False, filter_deductive_closure=False, **kwargs)[source]
Parameters:

  • include_deductive_closure (bool, optional) – Whether to evaluate using deductive closure axioms as positives. Defaults to False.

  • exclude_testing_set (bool, optional) – Whether to exclude the testing set from the evaluation. Defaults to False.

  • filter_deductive_closure (bool, optional) – Whether to filter deductive closure axioms from the evaluation. Defaults to False.

evaluate_base(model, eval_tuples, mode='test', include_deductive_closure=False, exclude_testing_set=False, filter_deductive_closure=False, **kwargs)[source]
get_logits(batch)[source]