def __init__(self):
self.ego = EgoNetwork()
class MatchUsers:
def __init__(self):
self.ego = EgoNetwork()
# discover top 5 similar users for a specific user
def match_user(self, target_ID, target_value):
score = []
# iterate users
file_en_filter_entity = open('/Users/shengjing/Desktop/Thesis-Project/DATASET/twitter/twitter_en_filter_Alchemy.csv','r')
for line in file_en_filter_entity:
line = line.strip('\r')
line = line.strip('\n')
tokens = line.split('\t')
#location = tokens[1]
source_value = tokens[3]
#source_value is a string data type, pay attention
source_Id = tokens[0]
# test case
# if source_Id == '14417192':
# score_item= self.similarity_metric(source_value, target_value)
# print score_item
if source_Id != target_ID:
#compare similarity with the rest of users
score_item= self.similarity_metric(source_value, target_value)
temp = [target_ID, source_Id, score_item]
score.append(temp)
file_en_filter_entity.close()
return score
#compare topical category
def similarity_metric(self, v1, v2):
# in case of topical value
v = v1.replace(']', '').strip()
v = v.replace('[', '').strip()
if v != '':
# normalizeValue
normalize_V1 = self.normalize_value(v)
# jaccardSimilarity
score = self.ego.jaccard_similarity(normalize_V1, v2)
return score
else:
return 0.0
def normalize_value(self, v):
# topic data type
item_set = set()
#v = v.encode('ascii','ignore')
v = v.replace('\'', '').strip()
tokens = v.split(',')
for token in tokens:
term = token.split('/')
for item in term:
if item != ' ':
item_set.add(item)
return item_set
#rank the users based on similarity score of topical category in their bio-description
def rank(self, data):
# return l.sort(key=lambda tup: tup[2])
return sorted(data, key=lambda tup: tup[2])
#get top 5 users
def top(self, n, data):
print data[-n:]
