Interaction

class Interaction(*args, **kwargs)

Bases: Module, Generic[HeadRepresentation, RelationRepresentation, TailRepresentation], ABC

Base class for interaction functions.

Initializes internal Module state, shared by both nn.Module and ScriptModule.

Attributes Summary

entity_shape	The symbolic shapes for entity representations
is_complex	whether the interaction is defined on complex input
relation_shape	The symbolic shapes for relation representations
tail_entity_shape	Return the symbolic shape for tail entity representations.
value_range	the interaction's value range (for unrestricted input)

Methods Summary

forward(h, r, t)	Compute broadcasted triple scores given broadcasted representations for head, relation and tails.
full_entity_shapes()	Return all entity shapes (head & tail).
get_dimensions()	Get all of the relevant dimension keys.
head_indices()	Return the entity representation indices used for the head representations.
reset_parameters()	Reset parameters the interaction function may have.
score(h, r, t[, slice_size, slice_dim])	Compute broadcasted triple scores with optional slicing.
score_h(all_entities, r, t[, slice_size])	Score all head entities.
score_hrt(h, r, t)	Score a batch of triples.
score_r(h, all_relations, t[, slice_size])	Score all relations.
score_t(h, r, all_entities[, slice_size])	Score all tail entities.
tail_indices()	Return the entity representation indices used for the tail representations.

Attributes Documentation

entity_shape: Sequence[str] = ('d',)

The symbolic shapes for entity representations

is_complex: ClassVar[bool] = False

whether the interaction is defined on complex input

relation_shape: Sequence[str] = ('d',)

The symbolic shapes for relation representations

tail_entity_shape

Return the symbolic shape for tail entity representations.

Return type:

Sequence[str]

value_range: ClassVar[ValueRange] = ValueRange(lower=None, lower_inclusive=False, upper=None, upper_inclusive=False)

the interaction’s value range (for unrestricted input)

Methods Documentation

abstract forward(h, r, t)

Compute broadcasted triple scores given broadcasted representations for head, relation and tails.

Parameters:

• h (~HeadRepresentation) – shape: (*batch_dims, *dims) The head representations.

• r (~RelationRepresentation) – shape: (*batch_dims, *dims) The relation representations.

• t (~TailRepresentation) – shape: (*batch_dims, *dims) The tail representations.

Return type:

FloatTensor

Returns:

shape: batch_dims The scores.

full_entity_shapes()

Return all entity shapes (head & tail).

Return type:

Sequence[str]

classmethod get_dimensions()

Get all of the relevant dimension keys.

This draws from Interaction.entity_shape, Interaction.relation_shape, and in the case of ConvEInteraction, the Interaction.tail_entity_shape.

Return type:

Set[str]

Returns:

a set of strings representting the dimension keys.

head_indices()

Return the entity representation indices used for the head representations.

Return type:

Sequence[int]

reset_parameters()

Reset parameters the interaction function may have.

score(h, r, t, slice_size=None, slice_dim=1)

Compute broadcasted triple scores with optional slicing.

Note

At most one of the slice sizes may be not None.

# TODO: we could change that to slicing along multiple dimensions, if necessary

Parameters:

• h (~HeadRepresentation) – shape: (*batch_dims, *dims) The head representations.

• r (~RelationRepresentation) – shape: (*batch_dims, *dims) The relation representations.

• t (~TailRepresentation) – shape: (*batch_dims, *dims) The tail representations.

• slice_size (Optional[int]) – The slice size.

• slice_dim (int) – The dimension along which to slice. From {0, …, len(batch_dims)}

Return type:

FloatTensor

Returns:

shape: batch_dims The scores.

score_h(all_entities, r, t, slice_size=None)

Score all head entities.

Parameters:

• all_entities (~HeadRepresentation) – shape: (num_entities, d_e) The head representations.

• r (~RelationRepresentation) – shape: (batch_size, d_r) The relation representations.

• t (~TailRepresentation) – shape: (batch_size, d_e) The tail representations.

• slice_size (Optional[int]) – The slice size.

Return type:

FloatTensor

Returns:

shape: (batch_size, num_entities) The scores.

score_hrt(h, r, t)

Score a batch of triples.

Parameters:

• h (~HeadRepresentation) – shape: (batch_size, d_e) The head representations.

• r (~RelationRepresentation) – shape: (batch_size, d_r) The relation representations.

• t (~TailRepresentation) – shape: (batch_size, d_e) The tail representations.

Return type:

FloatTensor

Returns:

shape: (batch_size, 1) The scores.

score_r(h, all_relations, t, slice_size=None)

Score all relations.

Parameters:

• h (~HeadRepresentation) – shape: (batch_size, d_e) The head representations.

• all_relations (~RelationRepresentation) – shape: (num_relations, d_r) The relation representations.

• t (~TailRepresentation) – shape: (batch_size, d_e) The tail representations.

• slice_size (Optional[int]) – The slice size.

Return type:

FloatTensor

Returns:

shape: (batch_size, num_entities) The scores.

score_t(h, r, all_entities, slice_size=None)

Score all tail entities.

Parameters:

• h (~HeadRepresentation) – shape: (batch_size, d_e) The head representations.

• r (~RelationRepresentation) – shape: (batch_size, d_r) The relation representations.

• all_entities (~TailRepresentation) – shape: (num_entities, d_e) The tail representations.

• slice_size (Optional[int]) – The slice size.

Return type:

FloatTensor

Returns:

shape: (batch_size, num_entities) The scores.

tail_indices()

Return the entity representation indices used for the tail representations.

Return type:

Sequence[int]