def __init__(self, dataset, compute_ranks, compute_mono_rank, build_indexes, max_per_rank=100, index_url='http://127.0.0.1:8080', debug=False):
self.__debug = debug
self.__index_url = index_url
self.__eggomatic = EggOMatic(dataset, compute_ranks, compute_mono_rank, build_indexes)
self.__html_helper = EggHelper(self.__debug, max_per_rank)
self.__html_helper.set_index_url(index_url)
self.__max_per_rank = max_per_rank
class EggOMaticExperiments:
__exp_OSim = True
__exp_KSim = True
__offline_list = ['offline1'] # or 'gold2'
__online_list = ['online1']
__top_tags = 20
__begin_top_many_users = 1
__end_top_many_users = 10
# exponential step 2, 4 ,8 16
__step_top_many_users = 2
def set_top_tags(self, top_tags=20):
self.__top_tags = top_tags
def set_forb_tags(self, forb_tags):
self.__forb_tags = forb_tags
def set_offline_list(self, type=['offline1']):
self.__offline_list = type
def set_online_list(self, list):
self.__online_list = list
def set_error_list(self, list=['osim','ksim']):
self.__error_list = list
def __init__(self, dataset, compute_ranks=True, compute_mono_rank=True, build_index=True, max_per_rank=10000000000, top_tags=None):
self.__dataset = dataset
self.__eggomatic = EggOMatic(dataset, compute_ranks, compute_mono_rank, build_index, max_per_rank, top_tags)
def __calc_error_save(self, offline_type, online_type, type_rank, error_type):
f = open(self.__dataset + '.' + offline_type + '_VS_' + online_type + '.' + error_type, 'a')
print 'writing ' + self.__dataset + '.' + offline_type + '_VS_' + online_type + '.' + error_type
offline_rank = type_rank[offline_type]
online_rank = type_rank[online_type]
for top_many_users in [self.__step_top_many_users**i for i in range(self.__begin_top_many_users,self.__end_top_many_users)]:
float_error, info_val = eval('rank_dist_%s(offline_rank, online_rank, top_many_users)' % error_type)
if info_val >= 0:
f.write('%d %d %f \n' % (top_many_users, len(offline_rank), float_error))
def __empty_result_files(self):
for offline_type in self.__offline_list:
for online_type in self.__online_list:
for error_type in self.__error_list:
f = open(self.__dataset + '.' + offline_type + '_VS_' + online_type + '.' + error_type, 'w')
f.close()
def run(self):
filename = self.__dataset + '.tagged_graph'
tags_filename = filename + '.ranks/all.tags'
f = open(tags_filename)
lines = []
for i in range(self.__top_tags + 1000):
lines.append(f.readline())
f.close()
top_tags = map(lambda line: line.strip().split(' ')[0], lines)
top_tags = filter(lambda x:not x in self.__forb_tags, top_tags)[0:self.__top_tags]
#top_tags = tags.get_top_tags(self.__top_tags)
#top_tags = map(lambda x: x[0], top_tags)
self.__ranker = RankerByTags()
self.__ranker.load(filename)
# tag_sets = []
# for i in range(len(top_tags)):
# tag1 = top_tags[i]
# for tag2 in top_tags[:i-1]:
# tag_sets.append((tag1, tag2))
# empty result files
self.__empty_result_files()
count = 0
nro_exps = len(top_tags)*(len(top_tags)-1) / 2
len_top_tags = len(top_tags)
for i in range(1,len(top_tags)):
tag1 = top_tags[i]
for tag2 in top_tags[:i]:
ranks = self.__eggomatic.rank_by_tags([tag1, tag2])
type_rank = {}
type_rank['offline1'] = ranks[0]
type_rank['mono'] = ranks[1]
type_rank['online1'] = ranks[2]
type_rank['online2'] = ranks[3]
type_rank['online3'] = ranks[4]
type_rank['online4'] = ranks[5]
type_rank['offline2'] = ranks[6]
type_rank['online6'] = ranks[7]
if (tag1=='portugal' and tag2=='music') or (tag2=='portugal' and tag1=='music'):
f = open('ranks.txt','w')
for trank, rank in type_rank.iteritems():
f.write('----------------------------------------\n')
f.write('%s\n' % trank)
for t in rank:
f.write('%s\n' % str(t))
f.close()
for offline_type in self.__offline_list:
for online_type in self.__online_list:
for error_type in self.__error_list:
print 'for tags: %s %s' % (tag1,tag2)
self.__calc_error_save(offline_type, online_type, type_rank, error_type)
count += 1
print 'NUMBER OF EXPERIMENTS/PAIRS COMPLETED: %d of %d' % (count, nro_exps)
print 'EXPERIMENTS COMPLETED FOR THE TOP %d of %d TAGS (all pairs)' % (i,len_top_tags)
def __init__(self, dataset, compute_ranks=True, compute_mono_rank=True, build_index=True, max_per_rank=10000000000, top_tags=None):
self.__dataset = dataset
self.__eggomatic = EggOMatic(dataset, compute_ranks, compute_mono_rank, build_index, max_per_rank, top_tags)
class EggOMaticWeb(object):
__error = 'ERROR: bad URL, clean your nose -> %s'
def __init__(self, dataset, compute_ranks, compute_mono_rank, build_indexes, max_per_rank=100, index_url='http://127.0.0.1:8080', debug=False):
self.__debug = debug
self.__index_url = index_url
self.__eggomatic = EggOMatic(dataset, compute_ranks, compute_mono_rank, build_indexes)
self.__html_helper = EggHelper(self.__debug, max_per_rank)
self.__html_helper.set_index_url(index_url)
self.__max_per_rank = max_per_rank
def set_index_url(self, index_url):
self.__html_helper.set_index_url(index_url)
def set_debug(self, flag):
self.__debug = flag
def index(self):
return self.__html_helper.header() + self.__html_helper.end()
index.exposed = True
def query(self, tags, clustering='0'):
clustering = clustering == '1'
tags = tags.strip().lower()
match = re.search('[a-z0-9\- ]+', tags)
if len(tags) > 0 and (not match or len(match.group()) != len(tags)):
return self.__html_helper.bad_input(tags)
tags = map(lambda x:x.strip(),tags.split(' '))
cluster, rank = [], []
print 'TAGS: %s' % tags
if not tags or len(tags) == 0 or (len(tags)==1 and tags[0]==''):
rank = self.__eggomatic.rank_by_tag('')
elif len(tags) > 1:
for tag in tags:
if not self.__eggomatic.good_tag(tag):
return self.__html_helper.bad_tag(tag)
ranks = self.__eggomatic.rank_by_tags(tags)
for tag in tags:
if self.__eggomatic.good_tag(tag):
cluster.append(tag)
else: # len(tags) == 1:
if not self.__eggomatic.good_tag(tags[0]):
return self.__html_helper.bad_tag(tags[0])
elif self.__eggomatic.has_bigger_cluster(tags[0]):
cluster_number = 0
if clustering:
if not self.__eggomatic.has_many_clusters(tags[0]):
cluster = self.__eggomatic.clusters(tags[0])[0]
else:
clusters = self.__eggomatic.clusters(tags[0])
return self.__html_helper.choose_cluster_page(clusters, self.__eggomatic.total_users(), self.__eggomatic.total_tags())
else:
cluster = [tags[0]]
ranks = self.__eggomatic.rank_by_tag(tags[0], clustering, cluster_number)
else:
cluster = [tags[0]]
ranks = self.__eggomatic.rank_by_tag(tags[0])
# TODO ask if has more than one cluster.
# else:
# return self.__html_helper.header() + self.__html_helper.end()
cluster = list(cluster)
cluster.sort()
return self.__html_helper.complete_page(ranks, cluster, self.__eggomatic.total_users(), self.__eggomatic.total_tags())
query.exposed = True
def query_user(self, user, clustering='0'):
clustering = clustering == '1'
user = user.strip() #.lower()
match = re.search('[A-Za-z0-9\-]+', user)
if len(user) == 0:
return self.index()
if (not match or len(match.group()) != len(user)):
return self.__html_helper.bad_input_user(user)
if not self.__eggomatic.good_user(user):
return self.__html_helper.bad_user(user)
if clustering:
rank = self.__eggomatic.user_ranks_clustering(user)
else:
rank = self.__eggomatic.user_ranks(user)
return self.__html_helper.complete_page_user(rank, [user], self.__eggomatic.total_users(), self.__eggomatic.total_tags(), clustering)
query_user.exposed = True
def default(self, param):
print 'default'
param = param.lower()
match = re.search('[a-z\-_]+', param)
if not match:
return self.__error % str(param)
param = param.lower()
tags = param.split('_')
if len(tags) == 1:
rank = self.__eggomatic.rank_by_tag(tags[0])
default.exposed = True