def __init__(self, conf):
self.conf = conf
if not conf.contains('ratings') and not conf.contains('social'):
print 'The config file is not in the correct format!'
exit(-1)
if conf.contains('ratings'):
ratingData = FileIO.loadDataSet(conf, conf['ratings'])
self.dao = RatingDAO(conf, ratingData)
if conf.contains('social'):
relationData = FileIO.loadRelationship(conf, conf['social'])
self.sao = SocialDAO(conf, relationData)
def __init__(self,conf,trainingSet=None,testSet=None,relation=list(),fold='[1]'):
super(SocialRecommender, self).__init__(conf,trainingSet,testSet,fold)
self.social = SocialDAO(self.config,relation) #social relations access control
# data clean
cleanList = []
cleanPair = []
for user in self.social.followees:
if not self.data.user.has_key(user):
cleanList.append(user)
for u2 in self.social.followees[user]:
if not self.data.user.has_key(u2):
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followees[u]
for pair in cleanPair:
if self.social.followees.has_key(pair[0]):
del self.social.followees[pair[0]][pair[1]]
cleanList = []
cleanPair = []
for user in self.social.followers:
if not self.data.user.has_key(user):
cleanList.append(user)
for u2 in self.social.followers[user]:
if not self.data.user.has_key(u2):
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followers[u]
for pair in cleanPair:
if self.social.followers.has_key(pair[0]):
del self.social.followers[pair[0]][pair[1]]
def __init__(self,
conf,
trainingSet=None,
testSet=None,
relation=list(),
fold='[1]'):
super(SocialRecommender, self).__init__(conf, trainingSet, testSet,
fold)
self.sao = SocialDAO(self.config,
relation) #social relations access control
def __init__(self, conf, trainingSet, testSet, relation, fold='[1]'):
super(SocialRecommender, self).__init__(conf, trainingSet, testSet,
fold)
self.social = SocialDAO(self.config,
relation) #social relations access control
# data clean
cleanList = []
cleanPair = []
for user in self.social.followees:
if user not in self.data.user:
cleanList.append(user)
for u2 in self.social.followees[user]:
if u2 not in self.data.user:
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followees[u]
for pair in cleanPair:
if pair[0] in self.social.followees:
del self.social.followees[pair[0]][pair[1]]
cleanList = []
cleanPair = []
for user in self.social.followers:
if user not in self.data.user:
cleanList.append(user)
for u2 in self.social.followers[user]:
if u2 not in self.data.user:
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followers[u]
for pair in cleanPair:
if pair[0] in self.social.followers:
del self.social.followers[pair[0]][pair[1]]
idx = []
for n, pair in enumerate(self.social.relation):
if pair[0] not in self.data.user or pair[1] not in self.data.user:
idx.append(n)
for item in reversed(idx):
del self.social.relation[item]
class Display(object):
def __init__(self, conf):
self.conf = conf
if not conf.contains('ratings') and not conf.contains('social'):
print 'The config file is not in the correct format!'
exit(-1)
if conf.contains('ratings'):
ratingData = FileIO.loadDataSet(conf, conf['ratings'])
self.dao = RatingDAO(conf, ratingData)
if conf.contains('social'):
relationData = FileIO.loadRelationship(conf, conf['social'])
self.sao = SocialDAO(conf, relationData)
def draw(self):
print 'draw chart...'
#rating
if self.conf.contains('ratings'):
y = [triple[2] for triple in self.dao.trainingData]
x = self.dao.rScale
if len(x) < 20:
Chart.hist(x, y, len(self.dao.rScale), '#058edc',
'Rating Histogram', 'Rating Scale', 'Count',
'../visual/visualization/images/rh')
y = [len(self.dao.userRated(u)[0]) for u in self.dao.user]
Chart.distribution(y, 'Rating Count Distribution', '',
'Rated items count per user',
'../visual/visualization/images/rcu')
y = [len(self.dao.itemRated(i)[0]) for i in self.dao.item]
Chart.distribution(y, 'Rating Count Distribution', '',
'user Rated count per item',
'../visual/visualization/images/rci')
#social
if self.conf.contains('social'):
x = [len(self.sao.getFollowers(u)) for u in self.sao.user]
y = [len(self.sao.getFollowees(u)) for u in self.sao.user]
Chart.scatter(x, y, 'red', 'Follower&Followee', 'Follower count',
'Followee count',
'../visual/visualization/images/ff')
y = [len(self.sao.getFollowers(u)) for u in self.sao.user]
Chart.distribution(y, 'Followers Distribution', '',
'Followers count per user',
'../visual/visualization/images/fd1')
y = [len(self.sao.getFollowees(u)) for u in self.sao.user]
Chart.distribution(y, 'Followees Distribution', '',
'Followees count per user',
'../visual/visualization/images/fd2')
def render(self):
self.draw()
html ="<html><head><title>Data Analysis</title>\n" \
"<link rel='stylesheet' type='text/css' href='reportStyle.css'/></head>\n" \
"<body><div class='reportTitle'><div class='in'>Data Analysis</div></div>\n" \
"<div class='main'><div class='area1'>\n" \
"<div class='title'><h3>Data Files</h3></div><div class='text'>"
if self.conf.contains('ratings'):
html += "<b>Rating Data</b>: {rating}".format(
rating=abspath(self.conf['ratings']))
if self.conf.contains('social'):
html += "<br><b>Social Data</b>: {social}".format(
social=abspath(self.conf['social']))
html+="</div></div><div style='padding-top:20px'><center>" \
"<img src='images/header2.png'/></center></div>\n"
if self.conf.contains('ratings'):
html += "<div class='area1'><div class='title'><h3>Rating Data</h3></div>\n"
html += "<div class='text'><b>Rating Scale</b>: {scale}</br>".format(
scale=' '.join([str(item) for item in self.dao.rScale]))
html += "<b>User Count</b>: {user}<br><b>Item Count</b>: {item}<br><b>Record Count</b>: {record}<br><b>Global Mean</b>: {mean}</div>\n"\
.format(user = str(len(self.dao.user)),item=str(len(self.dao.item)),record = str(len(self.dao.trainingData)),
mean = str(round(denormalize(self.dao.globalMean,self.dao.rScale[-1],self.dao.rScale[0]),3)))
html += "<center><div class='img'><img src='images/rh.png' width='640px' height='480px'/></div></center>\n"
html += "<center><div class='img'><img src='images/rcu.png' width='640px' height='480px'/></div></center>\n"
html += "<center><div class='img'><img src='images/rci.png' width='640px' height='480px'/></div></center>\n"
html += "</div><div style='padding-top:20px'><center>" \
"<img src='images/header2.png'/></center></div>\n"
if self.conf.contains('social'):
html += "<div class='area1'><div class='title'><h3>Social Data</h3></div>\n"
html += "<div class='text'><b>User Count</b>: {user}<br><b>Relation Count</b>: {relation}<br></div>\n" \
.format(user=str(len(self.sao.user)), relation=str(len(self.sao.relation)))
html += "<center><div class='img'><img src='images/ff.png' width='640px' height='480px'/></div></center>\n"
html += "<center><div class='img'><img src='images/fd1.png' width='640px' height='480px'/></div></center>\n"
html += "<center><div class='img'><img src='images/fd2.png' width='640px' height='480px'/></div></center>\n"
html += "</div><div style='padding-top:20px'><center>" \
"<img src='images/header2.png'/></center></div>\n"
html += "</div></body></html>"
FileIO.writeFile('../visual/visualization/', 'analysis.html', html)
print 'The report has been output to', abspath(
'../visual/visualization/analysis.html')
webbrowser.open(abspath('../visual/visualization/analysis.html'),
new=0,
autoraise=True)
def __init__(self, conf):
super(SocialRecommender, self).__init__(conf)
self.sao = SocialDAO(self.config) #social relations access control
def __init__(self,
conf,
trainingSet,
testSet,
relation,
inform1,
inform2,
fold='[1]'):
super(MetaRecommender, self).__init__(conf, trainingSet, testSet, fold)
self.social = SocialDAO(self.config,
relation) #social relations access control
self.meta = MetaDAO(self.config, inform1, inform2)
# data clean
cleanList = []
cleanPair = []
for user in self.social.followees:
if not self.data.user.has_key(user):
cleanList.append(user)
for u2 in self.social.followees[user]:
if not self.data.user.has_key(u2):
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followees[u]
for pair in cleanPair:
if self.social.followees.has_key(pair[0]):
del self.social.followees[pair[0]][pair[1]]
cleanList = []
cleanPair = []
for user in self.social.followers:
if not self.data.user.has_key(user):
cleanList.append(user)
for u2 in self.social.followers[user]:
if not self.data.user.has_key(u2):
cleanPair.append((user, u2))
for u in cleanList:
del self.social.followers[u]
for pair in cleanPair:
if self.social.followers.has_key(pair[0]):
del self.social.followers[pair[0]][pair[1]]
idx = []
for n, pair in enumerate(self.social.relation):
if pair[0] not in self.data.user or pair[1] not in self.data.user:
idx.append(n)
for item in reversed(idx):
del self.social.relation[item]
cleanList = []
cleanPair = []
for movie in self.meta.actors:
if not self.data.item.has_key(movie):
cleanList.append(movie)
for actor in self.meta.actors[movie]:
if not self.meta.actor.has_key(actor):
cleanPair.append((movie, actor))
for m in cleanList:
del self.meta.actors[m]
for pair in cleanPair:
if self.meta.actors.has_key(pair[0]):
del self.meta.actors[pair[0]][pair[1]]
cleanList = []
cleanPair = []
for actor in self.meta.act:
if not self.meta.actor.has_key(actor):
cleanList.append(actor)
for movie in self.meta.act[actor]:
if not self.meta.item.has_key(movie):
cleanPair.append((actor, movie))
for a in cleanList:
del self.meta.act[a]
for pair in cleanPair:
if self.meta.act.has_key(pair[0]):
del self.meta.act[pair[0]][pair[1]]
idx = []
for n, pair in enumerate(self.meta.inform1):
if pair[0] not in self.meta.item or pair[1] not in self.meta.actor:
idx.append(n)
for item in reversed(idx):
del self.meta.inform1[item]
cleanList = []
cleanPair = []
for movie in self.meta.md:
if not self.data.item.has_key(movie):
cleanList.append(movie)
for dire in self.meta.md[movie]:
if not self.meta.dire.has_key(dire):
cleanPair.append((movie, dire))
for m in cleanList:
del self.meta.md[m]
for pair in cleanPair:
if self.meta.md.has_key(pair[0]):
del self.meta.md[pair[0]][pair[1]]
cleanList = []
cleanPair = []
for dire in self.meta.dm:
if not self.meta.dire.has_key(dire):
cleanList.append(dire)
for movie in self.meta.dm[dire]:
if not self.meta.item.has_key(movie):
cleanPair.append((dire, movie))
for d in cleanList:
del self.meta.dm[d]
for pair in cleanPair:
if self.meta.dm.has_key(pair[0]):
del self.meta.dm[pair[0]][pair[1]]
idx = []
for n, pair in enumerate(self.meta.inform2):
if pair[0] not in self.meta.item or pair[1] not in self.meta.dire:
idx.append(n)
for item in reversed(idx):
del self.meta.inform2[item]