def getFullInfluenceScore():
partialScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
fbfriends = getFriendsCollection()
fullScoreCollection = getFullInfluenceScoreCollection()
for friend in fbfriends.find():
fid = friend['id']
partial = partialScoreCollection.find_one({'_id': fid})
contribution = userClusterContributionCollection.find_one({'_id': fid})
score = []
for i in range(len(contribution['contribution'])):
avg = 1.0
try:
avg *= sum(partial['cluster'][str(i)]) / len(
partial['cluster'][str(i)])
except ZeroDivisionError:
avg = 0.0
score.append(avg * contribution['contribution'][i])
if len(score) > 0:
document = {}
document['_id'] = fid
document['name'] = friend['name']
document['score'] = score
fullScoreCollection.insert(document)
print friend['name'], score
def getFullInfluenceScore():
partialScoreCollection = getClusterInfluencerScoreCollection()
userClusterContributionCollection = getUserClusterContributionCollection()
fbfriends = getFriendsCollection()
fullScoreCollection = getFullInfluenceScoreCollection()
for friend in fbfriends.find():
fid = friend['id']
partial = partialScoreCollection.find_one({'_id' : fid})
contribution = userClusterContributionCollection.find_one({'_id' : fid})
score = []
for i in range(len(contribution['contribution'])):
avg = 1.0
try:
avg *= sum(partial['cluster'][str(i)])/len(partial['cluster'][str(i)])
except ZeroDivisionError:
avg = 0.0
score.append(avg*contribution['contribution'][i])
if len(score) > 0:
document = {}
document['_id'] = fid
document['name'] = friend['name']
document['score'] = score
fullScoreCollection.insert(document)
print friend['name'], score
def getPartialInfluenceScore():
clusterInfoCollection = getPagesClusterInfoCollection()
clusterInfluencerCollection = getClusterInfluencerScoreCollection()
friendsCollection = getFriendsCollection()
clusterInfluencerCollection.drop()
clusterNumber = len(clusterInfoCollection.distinct('cluster'))
for friend in friendsCollection.find():
_id = friend['id']
document = {}
document['_id'] = _id
document['cluster'] = {}
for i in range(clusterNumber):
document['cluster'][str(i)] = []
clusterInfluencerCollection.insert(document)
pagesCursor = clusterInfoCollection.find({"count": {"$gt": 3}})
epoch = datetime.datetime.utcfromtimestamp(0)
dt = 7 * 24 * 60 * 60
scores = []
done = 0
for page in pagesCursor:
users = page['people']
try:
users.sort(key=lambda x: x['created_time'])
cluster = page['cluster']
liketime = []
for user in users:
liketime.append((user['created_time'] - epoch).total_seconds())
back = 0
done += 1
print done
for user in users:
userId = user['id']
timeahead = (user['created_time'] - epoch).total_seconds() + dt
timeback = (user['created_time'] - epoch).total_seconds() - dt
ahead = bisect.bisect_right(liketime, timeahead)
score = ahead - back
back += 1
# print userId, cluster, score
clusterInfluencerCollection.update(
{'_id': userId},
{'$push': {
'cluster.' + str(cluster): score
}},
upsert=False)
except:
print "hmmm"
def getPartialInfluenceScore():
clusterInfoCollection = getPagesClusterInfoCollection()
clusterInfluencerCollection = getClusterInfluencerScoreCollection()
friendsCollection = getFriendsCollection()
clusterInfluencerCollection.drop()
clusterNumber = len(clusterInfoCollection.distinct('cluster'))
for friend in friendsCollection.find():
_id = friend['id']
document = {}
document['_id'] = _id
document['cluster'] = {}
for i in range(clusterNumber):
document['cluster'][str(i)] = []
clusterInfluencerCollection.insert(document)
pagesCursor = clusterInfoCollection.find({"count": {"$gt": 3}})
epoch = datetime.datetime.utcfromtimestamp(0)
dt = 7 * 24 * 60 * 60
scores = []
done = 0
for page in pagesCursor:
users = page['people']
try:
users.sort(key=lambda x: x['created_time'])
cluster = page['cluster']
liketime = []
for user in users:
liketime.append((user['created_time'] - epoch).total_seconds())
back = 0
done += 1
print done
for user in users:
userId = user['id']
timeahead = (user['created_time'] - epoch).total_seconds() + dt
timeback = (user['created_time'] - epoch).total_seconds() - dt
ahead = bisect.bisect_right(liketime, timeahead)
score = ahead - back
back += 1
# print userId, cluster, score
clusterInfluencerCollection.update({'_id': userId}, {'$push': {'cluster.' + str(cluster): score}},
upsert=False)
except:
print "hmmm"
import requests
from database import getLikesCollection, getFriendsCollection
from utilities import url, access_token
from Queue import Queue
import threading
import json
idQueue = Queue()
likesCollection = getLikesCollection()
friendsCollection = getFriendsCollection()
for friend in friendsCollection.find():
idQueue.put(friend['id'])
class getLikes(threading.Thread):
def __init__(self):
threading.Thread.__init__(self)
self.queue = idQueue
def run(self):
while True:
try:
fbid = self.queue.get()
rurl = url + '/v2.3/' + fbid
response = requests.get(rurl,
params={
'access_token': access_token,
'fields': 'likes'
})
import requests
from database import getFriendsCollection
from utilities import url, access_token
friendscollection = getFriendsCollection()
def getFriends():
global url, access_token,count
rurl = url + '/v2.3/me'
response = requests.get(rurl, params = {'access_token' : access_token, 'fields' : 'friends'})
while True:
data = response.json()
if not 'friends'in data:
data['friends'] = data
insertFriends(data['friends']['data'])
if 'next' in data['friends']['paging']:
rurl = data['friends']['paging']['next']
else:
break
response = requests.get(rurl)
def insertFriends(data):
global friendscollection,count
for person in data:
if not friendscollection.find_one({"id" : person['id']}):
count = count + 1
friendscollection.insert(person)
print person['name'], person['id']
