def troll_bot_analyzer(user, api):
try:
user_data = data_user(user, api)
except tweepy.TweepError:
logging.error("This user is protected or does not exist. His information cannot be accessed")
else:
if len(user_data["tweets"]) == 0:
logging.error("There is not enough information to classify this user")
return False
hashtags_per_tweet = float(user_data["number_hashtags"]) / len(user_data["tweets"])
mentions_per_tweet = float(user_data["number_mentions"]) / len(user_data["tweets"])
percentage_tweet_with_mention = float(user_data["tweet_with_mentions"]) / len(user_data["tweets"])
percentage_tweet_with_hashtag = float(user_data["tweets_with_hashtags"]) / len(user_data["tweets"])
signs_per_char, capitals_per_char, activity, percentage_tweet_with_omg = drama_queen(user_data)
periodicity, answer = periodicity_answer(user_data)
diversity_hashtags = tweet_iteration_hashtags(user_data)
diversity_tweets = tweet_iteration_stemming(user_data)
urls_percentage = tweet_iteration_urls(user_data)
num_stalker, who_stalker = stalker(user_data)
per_drama_queen = percentage_drama_queen(activity, percentage_tweet_with_omg, capitals_per_char, signs_per_char,
percentage_tweet_with_hashtag, percentage_tweet_with_mention,
mentions_per_tweet, hashtags_per_tweet)
per_bot = percentage_bot(periodicity, answer, diversity_tweets)
per_stalker, famous, non_famous = percentage_stalker(num_stalker, who_stalker, mentions_per_tweet, percentage_tweet_with_mention, api)
if per_stalker == 0:
per_stalker = num_stalker
per_spammer = percentage_spammer(diversity_tweets, diversity_hashtags, urls_percentage)
per_hater = (1 - sentiment(user_data)) * 100
max_value = [per_bot, per_drama_queen, per_stalker, per_hater, per_spammer]
index = max_value.index(max(max_value))
labels = ["bot", "drama_queen", "stalker", "hater", "spammer"]
final = labels[index]
return {"user_id": user, "bot": per_bot, "drama_queen": per_drama_queen, "stalker": per_stalker, "hater": per_hater, "spammer": per_spammer, "famous": famous, "non_famous": non_famous, "stalked": who_stalker, "final": final}
def run(self):
while screen_names.count() > users.count():
s_name = screen_names.find_one_and_update(
{"collected": False}, {"$set": {
"collected": True
}})
logging.info("User {} in thread {}.".format(
s_name["_id"], self.getName()))
try:
users.insert_one(data_user(s_name["_id"], self.api))
screen_names.update_one({"_id": s_name["_id"]},
{"$set": {
"completed": True
}})
except TweepError as e:
logging.error(
"Ups, Arrived to API limit in thread {}. Exception: {}".
format(self.getName(), e))
sleep(60 * 15)
except Exception as e:
logging.error(
"Could not store user {}, the Exceception was {}.".format(
s_name["user"], e))
def percentage_stalker(num_stalker, who_stalker, mentions_per_tweet, percentage_tweet_with_mention, api):
famous = 0
non_famous = 0
if num_stalker > 45:
if data_user(str(who_stalker), api)["user_json"]["verified"]:
famous = 1
return num_stalker, famous, non_famous
if num_stalker > 75:
non_famous = 1
result = num_stalker
else:
result = (6*num_stalker+3*100*percentage_tweet_with_mention-1*100*mentions_per_tweet)/8
else:
result = 0
return result, famous, non_famous
def percentage_stalker(num_stalker, who_stalker, mentions_per_tweet,
percentage_tweet_with_mention, api):
famous = 0
non_famous = 0
if num_stalker > 45:
if data_user(str(who_stalker), api)["user_json"]["verified"]:
famous = 1
return num_stalker, famous, non_famous
if num_stalker > 75:
non_famous = 1
result = num_stalker
else:
result = (6 * num_stalker + 3 * 100 * percentage_tweet_with_mention
- 1 * 100 * mentions_per_tweet) / 8
else:
result = 0
return result, famous, non_famous
def run(self):
while screen_names.count() > users.count():
s_name = screen_names.find_one_and_update({"collected": False},
{"$set": {"collected": True}})
logging.info("User {} in thread {}.".format(s_name["_id"], self.getName()))
try:
users.insert_one(data_user(s_name["_id"], self.api))
screen_names.update_one({"_id": s_name["_id"]},
{"$set": {"completed": True}})
except TweepError as e:
logging.error("Ups, Arrived to API limit in thread {}. Exception: {}".format(self.getName(), e))
sleep(60 * 15)
except Exception as e:
logging.error("Could not store user {}, the Exceception was {}.".format(s_name["user"], e))
def troll_bot_analyzer(user, api):
try:
user_data = data_user(user, api)
except tweepy.TweepError:
logging.error(
"This user is protected or does not exist. His information cannot be accessed"
)
else:
if len(user_data["tweets"]) == 0:
logging.error(
"There is not enough information to classify this user")
return False
hashtags_per_tweet = float(user_data["number_hashtags"]) / len(
user_data["tweets"])
mentions_per_tweet = float(user_data["number_mentions"]) / len(
user_data["tweets"])
percentage_tweet_with_mention = float(
user_data["tweet_with_mentions"]) / len(user_data["tweets"])
percentage_tweet_with_hashtag = float(
user_data["tweets_with_hashtags"]) / len(user_data["tweets"])
signs_per_char, capitals_per_char, activity, percentage_tweet_with_omg = drama_queen(
user_data)
periodicity, answer = periodicity_answer(user_data)
diversity_hashtags = tweet_iteration_hashtags(user_data)
diversity_tweets = tweet_iteration_stemming(user_data)
urls_percentage = tweet_iteration_urls(user_data)
num_stalker, who_stalker = stalker(user_data)
per_drama_queen = percentage_drama_queen(
activity, percentage_tweet_with_omg, capitals_per_char,
signs_per_char, percentage_tweet_with_hashtag,
percentage_tweet_with_mention, mentions_per_tweet,
hashtags_per_tweet)
per_bot = percentage_bot(periodicity, answer, diversity_tweets)
per_stalker, famous, non_famous = percentage_stalker(
num_stalker, who_stalker, mentions_per_tweet,
percentage_tweet_with_mention, api)
if per_stalker == 0:
per_stalker = num_stalker
per_spammer = percentage_spammer(diversity_tweets, diversity_hashtags,
urls_percentage)
per_hater = (1 - sentiment(user_data)) * 100
max_value = [
per_bot, per_drama_queen, per_stalker, per_hater, per_spammer
]
index = max_value.index(max(max_value))
labels = ["bot", "drama_queen", "stalker", "hater", "spammer"]
final = labels[index]
return {
"user_id": user,
"bot": per_bot,
"drama_queen": per_drama_queen,
"stalker": per_stalker,
"hater": per_hater,
"spammer": per_spammer,
"famous": famous,
"non_famous": non_famous,
"stalked": who_stalker,
"final": final
}
