predict_target

predict_target(model, *, head=None, relation=None, tail=None, triples_factory=None, targets=None, mode=None)[source]

Get predictions for the head, relation, and/or tail combination.

Note

Exactly one of head, relation and tail should be None. This is the position which will be predicted.

Parameters:
  • model (Model) – A PyKEEN model

  • head (Union[None, int, str]) – the head entity, either as ID or as label. If None, predict heads

  • relation (Union[None, int, str]) – the relation, either as ID or as label. If None, predict relations

  • tail (Union[None, int, str]) – the tail entity, either as ID or as label. If None, predict tails

  • targets (Union[None, LongTensor, Sequence[Union[int, str]]]) – restrict prediction to these targets. None means no restriction, i.e., scoring all entities/relations.

  • triples_factory (Optional[TriplesFactory]) – the training triples factory; required if head/relation/tail are given as string, and used to translate the label to an ID.

  • mode (Optional[Literal[‘training’, ‘validation’, ‘testing’]]) – The pass mode, which is None in the transductive setting and one of “training”, “validation”, or “testing” in the inductive setting.

Return type:

Predictions

Returns:

The predictions, containing either the \(k\) highest scoring targets, or all targets if \(k\) is None.