def divide_new_elements(self, messages, unlearn, original=None):
"""Divides a given set of emails to be unlearned into ham and spam lists and unlearns both.
Param: messages contains indices of emails to learn/unlearn
"""
if unlearn:
h.unlearn([self.train_y, self.train_x, self.pol_y, self.pol_x], messages)
else:
h.relearn([self.train_y, self.train_x, self.pol_y, self.pol_x], original, messages)
def divide_new_elements(self, messages, unlearn, original=None):
"""Divides a given set of emails to be unlearned into ham and spam lists and unlearns both.
Param: messages contains indices of emails to learn/unlearn
"""
if unlearn:
h.unlearn([self.train_y, self.train_x, self.pol_y, self.pol_x],
messages)
else:
h.relearn([self.train_y, self.train_x, self.pol_y, self.pol_x],
original, messages)
def learn(self, cluster):
"""Learns a cluster from the ActiveUnlearner."""
if len(cluster.ham) + len(cluster.spam) != cluster.size:
print "\nUpdating cluster ham and spam sets...\n"
cluster.divide()
print "Is relearned train_y same as o_train_y before relearning?", self.train_y == self.o_train_y
print "Is relearned train_x same as o_train_x before relearning? ", self.train_x == self.o_train_x
h.relearn([self.train_y, self.train_x, self.pol_y, self.pol_x],\
[self.o_train_y, self.o_train_x, self.o_pol_y, self.o_pol_x], cluster.cluster_set)
# print "THIS IS NEW TRAIN_Y:", self.train_y
# print "THIS IS NEW O_TRAIN_Y:", self.o_train_y
print "Is relearned train_y same as o_train_y after relearning?", self.train_y == self.o_train_y
print "Is relearned train_x same as o_train_x after relearning?", self.train_x == self.o_train_x
if self.train_x != self.o_train_x: # let's log the indices where it's not the same
diff = []
for i in range(len(self.train_x)):
if self.train_x[i] != self.o_train_x[i]:
diff.append(i)
print "Cluster set indicies: ", cluster.cluster_set
print "Disparities after relearning: ", diff
def learn(self, cluster):
"""Learns a cluster from the ActiveUnlearner."""
if len(cluster.ham) + len(cluster.spam) != cluster.size:
print "\nUpdating cluster ham and spam sets...\n"
cluster.divide()
print "Is relearned train_y same as o_train_y before relearning?", self.train_y == self.o_train_y
print "Is relearned train_x same as o_train_x before relearning? ", self.train_x == self.o_train_x
h.relearn([self.train_y, self.train_x, self.pol_y, self.pol_x],\
[self.o_train_y, self.o_train_x, self.o_pol_y, self.o_pol_x], cluster.cluster_set)
# print "THIS IS NEW TRAIN_Y:", self.train_y
# print "THIS IS NEW O_TRAIN_Y:", self.o_train_y
print "Is relearned train_y same as o_train_y after relearning?", self.train_y == self.o_train_y
print "Is relearned train_x same as o_train_x after relearning?", self.train_x == self.o_train_x
if self.train_x != self.o_train_x: # let's log the indices where it's not the same
diff = []
for i in range(len(self.train_x)):
if self.train_x[i] != self.o_train_x[i]:
diff.append(i)
print "Cluster set indicies: ", cluster.cluster_set
print "Disparities after relearning: ", diff
