def gen_wordvector_matrices_fr_source_destination_nodes(self):
print 'generating word vector matrices for parent child nodes ...'
assert self.edgelabel_nodes_map is not None and self.edgelabel_nodes_map
source_vec_matrix_map = {}
destination_vec_matrix_map = {}
for curr_edge_label in self.edgelabel_nodes_map:
nodes_pair_list = self.edgelabel_nodes_map[curr_edge_label]
source_vectors_list = []
destination_vectors_list = []
for curr_nodes_pair in nodes_pair_list:
source = curr_nodes_pair[0]
source_vec = wv.get_wordvector(source)
source = None
destination = curr_nodes_pair[1]
destination_vec = wv.get_wordvector(destination)
destination = None
#
if source_vec is None or destination_vec is None:
continue
#
source_vectors_list.append(source_vec)
destination_vectors_list.append(destination_vec)
assert len(source_vectors_list) == len(destination_vectors_list)
if source_vectors_list and destination_vectors_list:
source_vec_matrix_map[curr_edge_label] = np.array(source_vectors_list)
destination_vec_matrix_map[curr_edge_label] = np.array(destination_vectors_list)
assert source_vec_matrix_map and destination_vec_matrix_map
self.source_vec_matrix_map = source_vec_matrix_map
self.destination_vec_matrix_map = destination_vec_matrix_map
def gen_wordvector_matrices_fr_common_parent_nodes_pair(self):
print 'generating word vector matrices for common parent nodes pair ...'
assert self.common_parent_edgelabel_nodes_pair is not None \
and self.common_parent_edgelabel_nodes_pair
child_i_vec_matrix_map = {}
child_j_vec_matrix_map = {}
for curr_edge_label_pair in self.common_parent_edgelabel_nodes_pair:
if curr_edge_label_pair[0] not in self.source_vec_matrix_map:
assert curr_edge_label_pair[0] not in self.destination_vec_matrix_map
continue
elif curr_edge_label_pair[1] not in self.source_vec_matrix_map:
assert curr_edge_label_pair[1] not in self.destination_vec_matrix_map
continue
#
nodes_pair_list = self.common_parent_edgelabel_nodes_pair[curr_edge_label_pair]
child_i_vectors_list = []
child_j_vectors_list = []
for curr_nodes_pair in nodes_pair_list:
child_i = curr_nodes_pair[0]
child_i_vec = wv.get_wordvector(child_i)
child_i = None
#
child_j = curr_nodes_pair[1]
child_j_vec = wv.get_wordvector(child_j)
child_j = None
#
if child_i_vec is None or child_j_vec is None:
continue
#
child_i_vectors_list.append(child_i_vec)
child_j_vectors_list.append(child_j_vec)
assert len(child_i_vectors_list) == len(child_j_vectors_list)
if child_i_vectors_list and child_j_vectors_list:
if self.is_filter and len(child_i_vectors_list) < self.min_num_common_parent_instances:
continue
child_i_vec_matrix_map[curr_edge_label_pair] = np.array(child_i_vectors_list)
child_j_vec_matrix_map[curr_edge_label_pair] = np.array(child_j_vectors_list)
assert child_i_vec_matrix_map and child_j_vec_matrix_map
self.child_i_vec_matrix_map = child_i_vec_matrix_map
self.child_j_vec_matrix_map = child_j_vec_matrix_map