class ModelLinkPrediction(number_of_trials: int, device: torch.device = 'cuda', loss_name: str = 'APP', emb_conv_name: str = 'SAGE')

Bases: object

Model for Link Prediction task with unsupervised embeddings

Parameters:

• dataset – (Graph): Input Graph

• (int) (number_of_trials) – Number of trials for optuna tuning embeddings

• device – (device): Device ‘cuda’ or ‘cpu’

• emb_conv_name – (str): Name of convolution for embedding learning

• loss_name – (str): Name of loss function for embedding learning

test(clf: sklearn.base.BaseEstimator, test_edges: List[List[int]], neg_samples_test: List[List[int]]) → float

Calculate f1 measure for test edges

Param:

cl (BaseEstimator)

Parameters:

• test_edges – (List): List of existing edges to test on

• neg_samples_test – (List): List of negative samples to test on

Returns:

(float): Value of f1 measure

train_cl(train_edges: List[List[int]], neg_samples_train: List[List[int]]) → sklearn.base.BaseEstimator

Train classifier for link prediction

Parameters:

• train_edges – (List): List of existing edges

• neg_samples_train – (List): List of negative samples to train

Returns:

(BaseEstimator): Classifier which support fit predict notation

train_test_edges(dataset: Graph) → Tuple[List[Any], List[int], List[Any], List[int]]

Split dataset to train and test and calculate negative samples

Parameters:

dataset – (Graph): Data to split on train, test and negatives

Returns:

(Tuple): Tuple of four lists of train edges, negativу train samples, test and negative test samples edges