def prepare_training_data(self):
self._dump2pickle('tmp/message.pickle')
self._preprocess()
self._tag_mapping()
from analysis.select_samples import select_samples
from analysis.select_samples import compute_order
from analysis.select_samples import save_samples
message = Serialize.loads(open('tmp/workspace.pickle').read())
samples = select_samples(message)
order = compute_order(samples)
save_samples(samples, order, 'tmp/samples.pickle')
return "done"
def prepare_training_data(self):
self._dump2pickle('tmp/message.pickle')
self._preprocess()
self._tag_mapping()
from analysis.select_samples import select_samples
from analysis.select_samples import compute_order
from analysis.select_samples import save_samples
message = Serialize.loads(open('tmp/workspace.pickle').read())
samples = select_samples(message)
order = compute_order(samples)
save_samples(samples, order, 'tmp/samples.pickle')
return "done"
def train(self, step = 100000):
from analysis.autoweight import AutoWeight
from analysis.autoweight import load_weights
from analysis.autoweight import save_weights
from analysis.autoweight import LearnerSigmoid
data = Serialize.loads(open('tmp/samples.pickle').read())
samples = data['samples']
order = data['order']
iweight = load_weights()
aw = AutoWeight(samples, order, iweight, LearnerSigmoid())
aw.sgd(step)
save_weights(aw)
self.score.load_weight()
return "done"
def train(self, step=100000):
from analysis.autoweight import AutoWeight
from analysis.autoweight import load_weights
from analysis.autoweight import save_weights
from analysis.autoweight import LearnerSigmoid
data = Serialize.loads(open('tmp/samples.pickle').read())
samples = data['samples']
order = data['order']
iweight = load_weights()
aw = AutoWeight(samples, order, iweight, LearnerSigmoid())
aw.sgd(step)
save_weights(aw)
self.score.load_weight()
return "done"
def _preprocess(self):
import time
# load
begin = time.time()
message = Serialize.loads(open('tmp/message.pickle').read())
end = time.time()
print "Load finish. Time elapsed: %.3f" % (end - begin)
# Preprocessing
# tag2msg and msg2tag dict
tl = message['tag_list']
td = {}
td_r = {}
for (msg_id, tag_id) in tl:
if not msg_id in td:
td[msg_id] = {}
td[msg_id][tag_id] = 1
if not tag_id in td_r:
td_r[tag_id] = {}
td_r[tag_id][msg_id] = 1
message['dict_msg2tag'] = td
message['dict_tag2msg'] = td_r
# 1. add tags attributes to msg
# 2. make msg dict
# 3. make seen list
ml = message['message_list']
md = {}
seen_list = []
for m in ml:
if m.flag == "seen":
seen_list.append(m)
if m.msg_id in td:
m.tags = td[m.msg_id]
else:
m.tags = {}
md[m.msg_id] = m
message['dict_msg'] = md
message['seen_list'] = seen_list
# save
begin = time.time()
open('tmp/workspace.pickle', 'w').write(Serialize.dumps(message))
end = time.time()
print "Save finish. Time elapsed: %.3f" % (end - begin)
def _preprocess(self):
import time
# load
begin = time.time()
message = Serialize.loads(open('tmp/message.pickle').read())
end = time.time()
print "Load finish. Time elapsed: %.3f" % (end - begin)
# Preprocessing
# tag2msg and msg2tag dict
tl = message['tag_list']
td = {}
td_r = {}
for (msg_id, tag_id) in tl:
if not msg_id in td:
td[msg_id] = {}
td[msg_id][tag_id] = 1
if not tag_id in td_r:
td_r[tag_id] = {}
td_r[tag_id][msg_id] = 1
message['dict_msg2tag'] = td
message['dict_tag2msg'] = td_r
# 1. add tags attributes to msg
# 2. make msg dict
# 3. make seen list
ml = message['message_list']
md = {}
seen_list = []
for m in ml:
if m.flag == "seen":
seen_list.append(m)
if m.msg_id in td:
m.tags = td[m.msg_id]
else:
m.tags = {}
md[m.msg_id] = m
message['dict_msg'] = md
message['seen_list'] = seen_list
# save
begin = time.time()
open('tmp/workspace.pickle', 'w').write(Serialize.dumps(message))
end = time.time()
print "Save finish. Time elapsed: %.3f" % (end - begin)
def _str2pyobj(self, message):
return Serialize.loads(base64.decodestring(message))
def _str2pyobj(self, message):
return Serialize.loads(base64.decodestring(message))