Convolution layer of 3D CNN :param input: :param P: weights and bias :return: convolution result

Convolution layer of 3D CNN :param input: :param P: weights and bias :return: convolution result

Metadata files are .csv :param data_folder: :param file_name: :param file_url: :return:

Metadata files are .csv :param data_folder: :param file_name: :param file_url: :return:

Returns the subfolder inside the dataset folder tree which contains the specific data to be loaded This method must be overridden by

Returns the default path in which to save the splitted data # Use default "dataset_name/split_name/original" or "dataset_name/split_n

Creates a temporary folder to be used during the data saving :return:

Loads all URM and ICM :return:

This function is executed before the evaluation of the current model It should ensure the current object "self" can be passed to the

Draw an item uniformly

Draw an item uniformly

Uniformly sample a negative item

Creates a nice printable string from the list of time lengths :param data_list: :param n_decimals: :return:

This function is called when the incremental model is found to have better validation score than the current best one So the current

Update latent factors for a single user or item. Y = |n_interactions|x|n_factors| YtY = |n_factors|x|n_factors|

Adds multiple elements to a given collection(s) :param names: str or list of collections :param values: tensor or list of tensors to add

Read a feature file and bind :param file: feature file :return:

Read a feature file and bind :param file: feature file :return:

Read a feature file and bind :param file: :return:

Read a feature file and bind :param file: :return:

Compute the similarity for the given dataset :param self: :param start_col: column to begin with :param end_col: colu

Construct dataset :param X_user: user structured data :param X_item: item structured data :return:

Construct dataset :param X_user: user structured data :param X_item: item structured data :return:

Score is discounted cumulative gain (dcg) Relevance is positive real values. Can use binary as the previous methods. Returns: Dis

Evaluate the performance (Hit_Ratio, NDCG) of top-K recommendation Return: score of each test rating.

Evaluate the performance (Hit_Ratio, NDCG) of top-K recommendation Return: score of each test rating.

if the last element is the correct one, then index = len(scores[0])-1

test_data = dict([positive, np.array[negatives]])