def predict_link(self, graph, graph_name, links, measure_name,
negative_edge_type):
for link in links:
source_node = link[0]
dest_node = link[1]
method_to_call = getattr(LinkPredictionStaticFunctions,
measure_name)
result = method_to_call(graph, source_node, dest_node)
author_guid = str(graph_name + "#" + source_node + "#" +
dest_node).encode('utf-8')
attribute_name = str(graph_name + "_" + measure_name + "_" +
negative_edge_type).encode('utf-8')
af = AuthorFeatures(_author_guid=author_guid,
_window_start=self._window_start,
_window_end=self._window_end,
_attribute_name=attribute_name,
_attribute_value=result)
self._author_features.append(af)
af_2 = AuthorFeatures(_author_guid=author_guid,
_window_start=self._window_start,
_window_end=self._window_end,
_attribute_name="author_screen_name",
_attribute_value=author_guid)
self._author_features.append(af_2)
def create_features_from_calculated_distances(self, authors_features):
# key #value
for author_guid, author_sub_type_paths_dict in self._author_distances.items(
):
author = self._author_dict[author_guid]
for author_sub_type in self._author_types_and_sub_types_in_dictionary:
series = author_sub_type_paths_dict[author_sub_type]
if DistancesFromTargetedClass.MEAN in self._distances_statistics:
mean = series.mean()
if math.isnan(mean):
mean = None
#TODO add parameter attenuation function, that should be applied to the distance
#TODO think of several types of attenuation functions, exponentially decreasing, logarithmically decreasing, etc..
mean_dist_to_author_sub_type_feature = AuthorFeatures(
author[3], self._window_start, self._window_end,
unicode("mean_dist_to_" + author_sub_type), mean)
authors_features.append(
mean_dist_to_author_sub_type_feature)
if DistancesFromTargetedClass.MIN in self._distances_statistics:
min_in_series = series.min()
if math.isnan(min_in_series):
min_in_series = None
min_dist_to_author_sub_type_feature = AuthorFeatures(
author[3], self._window_start, self._window_end,
unicode("min_dist_to_" + author_sub_type),
min_in_series)
authors_features.append(
min_dist_to_author_sub_type_feature)
def _create_tf_idf_author_feature(self, tweet_id,
aggregation_functions_name,
aggregated_tf_idf_score):
author_feature = AuthorFeatures()
author_feature.author_guid = tweet_id
author_feature.attribute_name = self.__class__.__name__ + "_" + aggregation_functions_name
author_feature.attribute_value = aggregated_tf_idf_score
return author_feature
def _add_ground_truth_to_author_features(self, graph_name, links, label):
# label 1.0 = link / label 0.0 = no link
for link in links:
author_guid = str(graph_name + "#" + link[0] + "#" + link[1]).encode('utf-8')
af = AuthorFeatures(_author_guid=author_guid,
_window_start=self._window_start, _window_end=self._window_end,
_attribute_name='has_link', _attribute_value=label)
self._author_features.append(af)
def create_author_feature(feature_name, author_guid, attribute_value, window_start, window_end):
feature = AuthorFeatures()
feature.author_guid = author_guid
feature.attribute_name = feature_name
feature.attribute_value = attribute_value
feature.window_start = window_start
feature.window_end = window_end
return feature
def _create_feature(self, feature, id_val, result, suffix=u''):
author_feature = AuthorFeatures()
author_feature.author_guid = id_val
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
subclass_name = self.__class__.__name__
author_feature.attribute_name = unicode(subclass_name + "_" + feature + suffix)
author_feature.attribute_value = result
return author_feature
def _create_feature(self, feature, id_val, result, suffix=''):
author_feature = AuthorFeatures()
author_feature.author_guid = id_val
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
subclass_name = self.__class__.__name__
author_feature.attribute_name = '{}_{}{}'.format(subclass_name, feature, suffix)
author_feature.attribute_value = str(result)
return author_feature
def _create_author_feature_with_name(self, author_guid, value, feature_name):
author_feature = AuthorFeatures()
author_feature.author_guid = author_guid
author_feature.window_start = date('2010-01-01 00:00:00')
author_feature.window_end = date('2020-01-01 23:59:59')
author_feature.attribute_name = feature_name
author_feature.attribute_value=value
self._authors_to_author_features_dict[author_guid].append(author_feature)
self._db.update_author_features((author_feature))
self._db.session.commit()
def run_and_create_author_feature(self, author, feature, posts, id_val,
feature_name):
try:
result = getattr(self, feature)(posts=posts, author=author)
author_feature = AuthorFeatures()
author_feature.author_guid = id_val
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
author_feature.attribute_name = unicode(feature_name)
author_feature.attribute_value = result
return author_feature
except Exception as e:
info_msg = e.message
logging.error(info_msg + feature)
def run_and_create_author_feature_with_given_value(self, author, value,
feature_name):
try:
result = value
author_feature = AuthorFeatures()
author_feature.author_guid = author
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
author_feature.attribute_name = unicode(feature_name)
author_feature.attribute_value = result
return author_feature
except Exception as e:
info_msg = e.message
logging.error(info_msg + str(value))
def run_and_create_author_feature(self, kwargs, id_val, feature):
try:
result = getattr(self, feature)(**kwargs)
author_feature = AuthorFeatures()
author_feature.author_guid = id_val
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
subclass_name = self.__class__.__name__
author_feature.attribute_name = unicode(subclass_name + "_" + feature)
author_feature.attribute_value = result
return author_feature
except Exception as e:
info_msg = e.message
logging.error(info_msg)
def execute(self):
start_time = time.time()
info_msg = "execute started for " + self.__class__.__name__
logging.info(info_msg)
post_topics_map = self._db.get_single_post_per_author_topic_mapping()
total = len(post_topics_map)
processed = 0
authors_features = []
for tuple in post_topics_map:
author_guid = tuple[0]
for topic in range(1, len(tuple)):
probability = tuple[topic]
author_feature = AuthorFeatures()
author_feature.author_guid = author_guid
author_feature.window_start = self._window_start
author_feature.window_end = self._window_end
author_feature.attribute_name = unicode('probability_topic_' +
str(topic))
author_feature.attribute_value = probability
authors_features.append(author_feature)
processed += 1
print("\r processed authors " + str(processed) + " from " +
str(total),
end="")
if authors_features:
print('\n Beginning merging author_features objects')
counter = 0
for author_features_row in authors_features:
counter += 1
self._db.update_author_features(author_features_row)
if counter == 100:
print("\r " + str(self.__class__.__name__) +
" merging author-features objects",
end="")
self._db.commit()
counter = 0
if counter != 0:
self._db.commit()
print('Finished merging author_features objects')
end_time = time.time()
diff_time = end_time - start_time
print('execute finished in ' + str(diff_time) + ' seconds')