Eval_Task

link_prediction

neuralkg_ind.eval_task.link_prediction.link_predict(batch, model, prediction='all', model_name=None)

The evaluate task is predicting the head entity or tail entity in incomplete triples.

Parameters:

• batch – The batch of the triples for validation or test.

• model – The KG model for training.

• predicion – mode of link prediction.

Returns:

The rank of the triple to be predicted.

Return type:

ranks

neuralkg_ind.eval_task.link_prediction.ind_predict(batch, model)

Getting the ranking of positive samples in the other 50 negative samples.

Parameters:

• batch – The batch of sample for test.

• model – The inductive model for training.

Returns:

The rank of the triple to be predicted.

Return type:

ranks

neuralkg_ind.eval_task.link_prediction.head_predict(batch, model)

Getting head entity ranks.

Parameters:

• batch – The batch of the triples for validation or test

• model – The KG model for training.

Returns:

The rank of the head entity to be predicted, dim [batch_size]

Return type:

tensor

neuralkg_ind.eval_task.link_prediction.tail_predict(batch, model)

Getting tail entity ranks.

Parameters:

• batch – The batch of the triples for validation or test

• model – The KG model for training.

Returns:

The rank of the tail entity to be predicted, dim [batch_size]

Return type:

tensor

neuralkg_ind.eval_task.link_prediction.calc_ranks(idx, label, pred_score)

Calculating triples score ranks.

Parameters:

• idx ([type]) – The id of the entity to be predicted.

• label ([type]) – The id of existing triples, to calc filtered results.

• pred_score ([type]) – The score of the triple predicted by the model.

Returns:

The rank of the triple to be predicted, dim [batch_size].

Return type:

ranks

neuralkg_ind.eval_task.link_prediction.link_predict_SEGNN(batch, kg, model, prediction='all')

The evaluate task is predicting the head entity or tail entity in incomplete triples.

Parameters:

• batch – The batch of the triples for validation or test.

• model – The KG model for training.

• predicion – mode of link prediction.

Returns:

The rank of the triple to be predicted.

Return type:

ranks

neuralkg_ind.eval_task.link_prediction.head_predict_SEGNN(batch, ent_emb, rel_emb, model)

Getting head entity ranks.

Parameters:

• batch – The batch of the triples for validation or test

• model – The KG model for training.

Returns:

The rank of the head entity to be predicted, dim [batch_size]

Return type:

tensor

neuralkg_ind.eval_task.link_prediction.tail_predict_SEGNN(batch, ent_emb, rel_emb, model)

Getting tail entity ranks.

Parameters:

• batch – The batch of the triples for validation or test

• model – The KG model for training.

Returns:

The rank of the tail entity to be predicted, dim [batch_size]

Return type:

tensor

neuralkg_ind.eval_task.link_prediction.calc_ranks_SEGNN(idx, filter_label, pred_score)

Calculating triples score ranks.

Parameters:

• idx ([type]) – The id of the entity to be predicted.

• label ([type]) – The id of existing triples, to calc filtered results.

• pred_score ([type]) – The score of the triple predicted by the model.

Returns:

The rank of the triple to be predicted, dim [batch_size].

Return type:

ranks

triple_classification

neuralkg_ind.eval_task.triple_classification.classification(batch, model)

Calculating triple classification score.

Parameters:

• batch – Positive sample and negative sample.

• model – The model for testing.

Returns:

The positive sample score and the negative sample score.

Return type:

score