def similarity_from_sparse(matrix_a: sparse.csr_matrix,
matrix_b: sparse.csr_matrix):
intersection = matrix_a.dot(matrix_b.transpose()).toarray()
norm_1 = np.array(matrix_a.multiply(matrix_a).sum(axis=1))
norm_2 = np.array(matrix_b.multiply(matrix_b).sum(axis=1))
union = norm_1 + norm_2.T - intersection
return intersection / union
def filter_dataset(X: sp.csr_matrix, min_ui_count=10, min_iu_count=10):
X = filter_rows(X, min_ui_count)
T_X = X.transpose()
T_X = filter_rows(T_X, min_iu_count)
X = T_X.transpose()
return X
def augmentURM(cls, URM_train: csr_matrix, W_sparse: csr_matrix,
threshold_interactions: int, threshold_similarity: float):
"""
Augmentation of the URM train.
:param threshold_interactions: here a threshold on the similarity is considered.
Similarity matrix W_sparse will be considered for this purpose
:param threshold_similarity: threshold used to insert a new row.
In this case it is specified as the minimum number of interactions required to insert a new
row in the URM train
:param W_sparse: similarity matrix
:param URM_train: URM train that will be augmented
:return: a csr_matrix with augmented interactions according to the threshold
"""
print("Augmenting URM")
URM_train = URM_train.copy()
# Count similarity
count_W_sparse = URM_train.dot(URM_train.transpose())
# Selecting new
print("Selecting new candidates")
users = np.arange(URM_train.shape[0])
new_rows_list = []
for i in range(0, users.size):
if i % 5000 == 0:
print("{} done in {}".format(i, users.size))
candidates = count_W_sparse[i].indices # users candidates
data = count_W_sparse[i].data # data for the candidates
for j, candidate in enumerate(candidates):
if candidate > i and data[
j] > threshold_interactions and W_sparse[
i, candidate] > threshold_similarity:
new_rows_list.append([i, candidate])
print("Candidate list size: {}".format(len(new_rows_list)))
# Creating the new matrix
print("Creating new URM...", end="")
new_URM = None
for candidate in new_rows_list:
new_row = URM_train[[candidate[0], candidate[1]]].sum(axis=0)
new_row = csr_matrix(new_row)
new_row.data[new_row.data > 1] = 1
if new_URM is None:
new_URM = new_row
else:
new_URM = vstack([new_URM, new_row], format="csr")
if new_URM is None:
new_URM = URM_train
else:
new_URM = vstack([URM_train, new_URM], format="csr")
print("Done")
return new_URM
def psinv2(matr: csr_matrix, dtype, reg=0.):
# inverse operation for sparse matrices doesn't seem to exist in cupy
regsize = matr.shape[1]
if reg == 0:
regm = csr_matrix((regsize, regsize))
else:
regm = identity(regsize, dtype=dtype) * (
1. / reg) # if direct cuda, call get_sparse_module()
toinv = matr.transpose().dot(matr) + regm
return get_sparse_module(
inv(toinv)) # if direct cuda, add .get() to inv param
def newAugmentUMR(cls, URM_train: csr_matrix, W_sparse: csr_matrix,
threshold_interactions: int,
threshold_similarity: float):
print("New Augmenting URM")
count_W_sparse = URM_train.dot(URM_train.transpose())
count_mask: csr_matrix = count_W_sparse > threshold_interactions
sim_mask: csr_matrix = W_sparse > threshold_similarity
mask = count_mask.multiply(sim_mask)
mask = triu(mask)
mask = mask.tocoo()
row_user = mask.row
col_user = mask.col
new_mask = row_user != col_user
row_user = row_user[new_mask]
col_user = col_user[new_mask]
new_users = np.array([row_user, col_user])
new_users = np.transpose(new_users)
new_rows_list: list = new_users.tolist()
print("Candidate list size: {}".format(len(new_rows_list)))
# Creating the new matrix
print("Creating new URM...", end="")
new_URM = None
for candidate in new_rows_list:
new_row = URM_train[[candidate[0], candidate[1]]].sum(axis=0)
new_row = csr_matrix(new_row)
new_row.data[new_row.data > 1] = 1
if new_URM is None:
new_URM = new_row
else:
new_URM = vstack([new_URM, new_row], format="csr")
if new_URM is None:
new_URM = URM_train
else:
new_URM = vstack([URM_train, new_URM], format="csr")
print("Done")
return new_URM
def _merge_into_adjacency(
tf_ids: csr_matrix, word_cooccurrences_list: List[Tuple[str, str,
float]],
token_to_int_vocab_map: Dict[str, int]) -> csr_matrix:
"""
Merge the word co-occurence information together with the tf-idf information to create an adjacency matrix
where,
(0, 0) to (|vocab|, |vocab|) - indices describe the word-word interactions
and,
(|vocal|, |vocab|) to (|vocab| + #Docs, |vocab|) - indices describe the word-document interactions.
"""
word_co_row = np.array([
token_to_int_vocab_map[word_cooccurrence[0]]
for word_cooccurrence in word_cooccurrences_list
])
word_co_col = np.array([
token_to_int_vocab_map[word_cooccurrence[1]]
for word_cooccurrence in word_cooccurrences_list
])
word_co_data = np.array([
word_cooccurrence[2] for word_cooccurrence in word_cooccurrences_list
])
word_coocurrences = csr_matrix(
(word_co_data, (word_co_row, word_co_col)),
shape=(len(token_to_int_vocab_map), len(token_to_int_vocab_map)))
# Stack word co-occurences ontop of TF-IDF (Left hand side of adjacency)
adj_lhs = vstack([word_coocurrences, tf_ids])
# Empty (zeros) for doc-doc interactions
zero_csr = csr_matrix(([], ([], [])),
shape=(tf_ids.shape[0], tf_ids.shape[0]))
# Mirror TF-IDFs and stack ontop of doc-doc interactions to create right hand side of the adjacency
adj_rhs = vstack([tf_ids.transpose(), zero_csr])
# Stack side-by-side
adj = hstack([adj_lhs, adj_rhs]) + identity(adj_lhs.shape[0])
assert adj.shape == (
len(token_to_int_vocab_map) + tf_ids.shape[0],
len(token_to_int_vocab_map) + tf_ids.shape[0],
), "Expected {} == {}".format(
adj.shape, (len(token_to_int_vocab_map) + tf_ids.shape[0],
len(token_to_int_vocab_map) + tf_ids.shape[0]))
return adj
def calc_objective_per_iter(w_i, feature_mat: sparse.csr_matrix, empirical_counts, num_h, true_tags, alpha):
"""
Calculate max entropy likelihood for an iterative optimization method
:param alpha: the regularization coefficient
:param num_h: number of histories in the training data
:param empirical_counts: pre computed empirical_counts
:param w_i: weights vector in iteration i
The function returns the Max Entropy likelihood (objective) and the objective gradient
"""
scores = feature_mat.dot(w_i)
scores = scores.reshape((num_h, -1))
exp_scores = np.exp(scores)
sum_exp = np.sum(exp_scores, axis=1)
probs = exp_scores/sum_exp.reshape((num_h, 1))
expected_counts = feature_mat.transpose().dot(probs.reshape(-1)).reshape(-1)
likelihood = np.sum(scores[np.arange(num_h), true_tags] - np.log(sum_exp) - (alpha/2) * (w_i**2))
grad = empirical_counts - expected_counts - alpha*w_i
return (-1) * likelihood, (-1) * grad
def snn_dissimilarity_func(graph: csr_matrix, n_neighbors: int, *args,
**kwargs) -> csr_matrix:
"""Default SNN dissimilarity function
Computes the dissimilarity between two points in terms of shared nearest neighbors
Args:
graph (scipy.sparse.csr_matrix): sparse matrix with dimensions (n_samples, n_samples),
where the element ij represents the distance between the point i and j
n_neighbors (int): number of neighbors in the k-neighborhood search
"""
graph.data[graph.data > 0] = 1
n_samples = graph.shape[0]
# Add the point as its own neighbor
graph += spdiags(np.ones(n_samples), diags=0, m=n_samples, n=n_samples)
matrix = graph * graph.transpose()
matrix.sort_indices()
# The lower the "closer"
matrix.data = n_neighbors - matrix.data
return matrix
def calculate(self, X: sp.csr_matrix, q_X: sp.csr_matrix):
q_X = normalize(q_X, norm="l1")
sim = q_X * X.transpose()
# sim = sim.transpose()
return sim.toarray()[0]
