SimplE

class SimplE(triples_factory, embedding_dim=200, loss=None, preferred_device=None, random_seed=None, regularizer=None, clamp_score=None, entity_initializer=None, relation_initializer=None)

Bases: pykeen.models.base.EntityRelationEmbeddingModel

An implementation of SimplE [kazemi2018].

SimplE is an extension of canonical polyadic (CP), an early tensor factorization approach in which each entity \(e \in \mathcal{E}\) is represented by two vectors \(\textbf{h}_e, \textbf{t}_e \in \mathbb{R}^d\) and each relation by a single vector \(\textbf{r}_r \in \mathbb{R}^d\). Depending whether an entity participates in a triple as the head or tail entity, either \(\textbf{h}\) or \(\textbf{t}\) is used. Both entity representations are learned independently, i.e. observing a triple \((h,r,t)\), the method only updates \(\textbf{h}_h\) and \(\textbf{t}_t\). In contrast to CP, SimplE introduces for each relation \(\textbf{r}_r\) the inverse relation \(\textbf{r'}_r\), and formulates its the interaction model based on both:

\[f(h,r,t) = \frac{1}{2}\left(\left\langle\textbf{h}_h, \textbf{r}_r, \textbf{t}_t\right\rangle + \left\langle\textbf{h}_t, \textbf{r'}_r, \textbf{t}_h\right\rangle\right)\]

Therefore, for each triple \((h,r,t) \in \mathbb{K}\), both \(\textbf{h}_h\) and \(\textbf{h}_t\) as well as \(\textbf{t}_h\) and \(\textbf{t}_t\) are updated.

See also

• Official implementation: https://github.com/Mehran-k/SimplE

• Improved implementation in pytorch: https://github.com/baharefatemi/SimplE

Initialize the entity embedding model.

See also

Constructor of the base class pykeen.models.Model

Attributes Summary

hpo_default	The default strategy for optimizing the model’s hyper-parameters
loss_default_kwargs	The default parameters for the default loss function class
regularizer_default_kwargs	The power sum settings used by [trouillon2016] for SimplE

Methods Summary

score_h(rt_batch)	Forward pass using left side (head) prediction.
score_hrt(hrt_batch)	Forward pass.
score_t(hr_batch)	Forward pass using right side (tail) prediction.

Attributes Documentation

hpo_default: ClassVar[Mapping[str, Any]] = {'embedding_dim': {'high': 256, 'low': 16, 'q': 16, 'type': <class 'int'>}}

The default strategy for optimizing the model’s hyper-parameters

loss_default_kwargs: ClassVar[Mapping[str, Any]] = {}

The default parameters for the default loss function class

regularizer_default_kwargs: ClassVar[Mapping[str, Any]] = {'normalize': True, 'p': 2.0, 'weight': 20}

The power sum settings used by [trouillon2016] for SimplE

Methods Documentation

score_h(rt_batch)

Forward pass using left side (head) prediction.

This method calculates the score for all possible heads for each (relation, tail) pair.

Parameters

rt_batch (LongTensor) – shape: (batch_size, 2), dtype: long The indices of (relation, tail) pairs.

Return type

FloatTensor

Returns

shape: (batch_size, num_entities), dtype: float For each r-t pair, the scores for all possible heads.

score_hrt(hrt_batch)

Forward pass.

This method takes head, relation and tail of each triple and calculates the corresponding score.

Parameters

hrt_batch (LongTensor) – shape: (batch_size, 3), dtype: long The indices of (head, relation, tail) triples.

Raises

NotImplementedError – If the method was not implemented for this class.

Return type

FloatTensor

Returns

shape: (batch_size, 1), dtype: float The score for each triple.

score_t(hr_batch)

Forward pass using right side (tail) prediction.

This method calculates the score for all possible tails for each (head, relation) pair.

Parameters

hr_batch (LongTensor) – shape: (batch_size, 2), dtype: long The indices of (head, relation) pairs.

Return type

FloatTensor

Returns

shape: (batch_size, num_entities), dtype: float For each h-r pair, the scores for all possible tails.