def costFunction(self, params):
"""Returns Error as a function of Lambda Params.
Args:
params: List of doubles, represents different lambda values,
lambda_phi is the first param; lambda_pi's are the next
6 parameters
Returns:
Cost - Integer representing number of misclassified results in the
training set.
"""
# Set Lambda Phi & Pi based upon params
# 1st Param - Lambda Phi
# 2nd Param - Lambda Pi
self.lambda_phi = params[0]
self.lambda_pi = map(lambda x: x / sum(params[1:]), params[1:])
# Train the classifier
self.train()
# Calculate Error as
# Number of Training Errors
error = 0
for (ques, ans) in self.training_set:
res = self.get_classification(ques)
pred_ans = ut.clean(ut.key_max_val_dict(res))
if pred_ans != ans:
error += 1
return error
def get_reply (self, string): self.trn = self.trn_1 self.threshold = self.threshold_1 self.tmap = self.training_map_1 #print self.lambda_pi_1 # Get the classification for 'string' res = self.trn.get_classification(string) # Get the value, reply of highest ranked response value = res[ut.key_max_val_dict(res)] reply = ut.key_max_val_dict(res) response_prior = '' # Check if it crosses minimum threshold if value >= self.threshold: # Check if it has been repeated in history # and respond appropriately """ if reply in self.history: response_prior = 'I\'ll repeat myself, ' #reply = 'I\'ll repeat myself, '+ choice(self.tmap.get(reply, [reply])) if response_prior + reply in self.history: response_prior = 'I am tired of this conversation, but ' + response_prior if response_prior + reply in self.history: response_prior = '' """ # Put the string in history self.history.append (response_prior + reply) reply = response_prior + choice(self.tmap.get(reply, [reply])) return reply # In case responses are below the threshold else: reply = ' I don\'t know anything about this' if reply in self.history: reply = 'Ask me something else' self.history.append(reply) return reply
def get_reply(self, string):
self.trn = self.trn_1
self.threshold = self.threshold_1
self.tmap = self.training_map_1
#print self.lambda_pi_1
# Get the classification for 'string'
res = self.trn.get_classification(string)
# Get the value, reply of highest ranked response
value = res[ut.key_max_val_dict(res)]
reply = ut.key_max_val_dict(res)
response_prior = ''
# Check if it crosses minimum threshold
if value >= self.threshold:
# Check if it has been repeated in history
# and respond appropriately
"""
if reply in self.history:
response_prior = 'I\'ll repeat myself, '
#reply = 'I\'ll repeat myself, '+ choice(self.tmap.get(reply, [reply]))
if response_prior + reply in self.history:
response_prior = 'I am tired of this conversation, but ' + response_prior
if response_prior + reply in self.history:
response_prior = ''
"""
# Put the string in history
self.history.append(response_prior + reply)
reply = response_prior + choice(self.tmap.get(reply, [reply]))
return reply
# In case responses are below the threshold
else:
reply = ' I don\'t know anything about this'
if reply in self.history:
reply = 'Ask me something else'
self.history.append(reply)
return reply
def costFunction (self, params): self.lambda_phi = params[0] self.lambda_pi = map(lambda x: x/sum(params[1:]), params[1:] ) #print self.lambda_pi # Train the classifier self.train() # Calculate Error error = 0.0 for (ques,ans) in self.training_set: res = self.get_classification(ques) pred_ans = ut.clean(ut.key_max_val_dict (res)) # Add one for every misclassified result if(pred_ans != ans): # print pred_ans, ans # raw_input() error += 1 return error
def costFunction(self, params):
self.lambda_pi = map(lambda x: x / sum(params), params)
#print self.lambda_pi
# Train the classifier
self.train()
# Calculate Error
error = 0.0
for (ques, ans) in self.training_set:
res = self.get_classification(ques)
pred_ans = ut.clean(ut.key_max_val_dict(res))
# Add one for every misclassified result
if (pred_ans != ans):
# print pred_ans, ans
# raw_input()
error += 1
return error
def get_reply(self, string, annotation):
"""Performs classification on the given text, based upon annotations, history and relevance and
gives an appropriate response.
Args:
string : The string to be classified
annotation : Chooses from two different datasets based upon this value
Returns:
A Final Response String to be supplied to the user.
"""
# If Annotation is 1; Then limit conversation length to 25 interactions;
# Specific to this situation
self.conversation_count += 1
if self.conversation_count > 25 and int(annotation) == 1:
return "I am irritated now, Talk to me Later"
if int(annotation) == 1:
self.trn = self.trn_1
self.threshold = self.threshold_1
self.tmap = self.training_map_1
else:
self.trn = self.trn_2
self.threshold = self.threshold_2
self.tmap = self.training_map_2
if self.trn is not self.trn_1:
print annotation
# Get the classification for 'string'
res = self.trn.get_classification(string)
# Get the value, reply of highest ranked response
value = res[ut.key_max_val_dict(res)]
reply = ut.key_max_val_dict(res)
response_prior = ""
# Check if it crosses minimum threshold
if value >= self.threshold:
# Check if it has been repeated in history
# and respond appropriately
if reply in self.history:
response_prior = "I'll repeat myself, "
if response_prior + reply in self.history:
response_prior = "I am tired of this conversation, but " + response_prior
if response_prior + reply in self.history:
response_prior = ""
# Put the string in history
self.history.append(response_prior + reply)
reply = response_prior + choice(self.tmap.get(reply, [reply]))
return reply
# In case responses are below the threshold
else:
reply = " I don't know anything about this"
if reply in self.history:
reply = "Ask me something else or leave me alone"
self.history.append(reply)
return reply