def __init__(self,
database_path,
global_description='',
positive_description='',
negative_description=''):
"""Accepts a database where a token would be
tested positive and negative."""
self.database = NaiveBayesDB(database_path,
global_description=global_description,
positive_description=positive_description,
negative_description=negative_description)
self.tokens = None
def __init__(self, database_path,
global_description='',
positive_description='',
negative_description=''):
"""Accepts a database where a token would be
tested positive and negative."""
self.database = NaiveBayesDB(database_path,
global_description=global_description,
positive_description=positive_description,
negative_description=negative_description)
self.tokens = None
class NaiveBayesClassifier(object):
"""High level naive Bayes classifier.
NaiveBayesDB will create the database if database_path
does not exist. Use descriptions when creating a new database."""
def __init__(self, database_path,
global_description='',
positive_description='',
negative_description=''):
"""Accepts a database where a token would be
tested positive and negative."""
self.database = NaiveBayesDB(database_path,
global_description=global_description,
positive_description=positive_description,
negative_description=negative_description)
self.tokens = None
def register_tokens(self, tokens):
self.tokens = tokens
return None
##############################
# database operations call NaiveBayesDB functions to abstract db interactions
# from this class
def train_positive(self):
"""For each token in self.tokens, add token/counter to positive_classification,
and/or increment positive_counter in database"""
self.database.train_positive(self.tokens)
return None
def untrain_positive(self):
"""For each token in tokens, decrement token counter in positive_classification.
If token counter == 0, remove token from table"""
self.database.untrain_positive(self, tokens)
return None
def train_negative(self):
"""For each token in tokens, add token/counter to negative_classification,
and/or increment negative_counter in database."""
self.database.train_negative(self.tokens)
return None
def untrain_negative(self):
"""For each token in tokens, decrement token counter in negative_classification.
If token counter == 0, remove token from table"""
self.database.untrain_negative(self, tokens)
##############################
# bayes formula operations
def calculate_probabilities(self):
"""Stores the Token.token_string values, as compared positive and negative
in the database, within the Token.positive_value and Token.negative_value,
respectively."""
total_positive = self.database.total_for_polarity(polarity='positive')
total_negative = self.database.total_for_polarity(polarity='negative')
for token in self.tokens:
positive_count = self.database.counter_for_token(token.token_string,
polarity='positive')
negative_count = self.database.counter_for_token(token.token_string,
polarity='negative')
numerator = (float(positive_count)/total_positive)
denominator = ((float(positive_count)/total_positive) + (float(negative_count)/total_negative))
token.positive_value = numerator/denominator
numerator = (float(negative_count)/total_negative)
denominator = ((float(negative_count)/total_negative) + (float(positive_count)/total_positive))
token.negative_value = numerator/denominator
return True
def sum_positive(self):
"""p(S) = (p1 * p2 ... pn) /
( (p1 * p2 ... * pn) + ( (1 - p1) * (1 - p2) ... * (1 - pn) ) )"""
numerator = sum([token.positive_value for token in self.tokens])
denominator = numerator + sum([1-token.positive_value for token in self.tokens])
return float(numerator)/denominator
def sum_negative(self):
"""p(S) = (p1 * p2 ... pn) /
( (p1 * p2 ... * pn) + ( (1 - p1) * (1 - p2) ... * (1 - pn) ) )"""
numerator = sum([token.negative_value for token in self.tokens])
denominator = numerator + sum([1-token.negative_value for token in self.tokens])
return float(numerator)/denominator
class NaiveBayesClassifier(object):
"""High level naive Bayes classifier.
NaiveBayesDB will create the database if database_path
does not exist. Use descriptions when creating a new database."""
def __init__(self,
database_path,
global_description='',
positive_description='',
negative_description=''):
"""Accepts a database where a token would be
tested positive and negative."""
self.database = NaiveBayesDB(database_path,
global_description=global_description,
positive_description=positive_description,
negative_description=negative_description)
self.tokens = None
def register_tokens(self, tokens):
self.tokens = tokens
return None
##############################
# database operations call NaiveBayesDB functions to abstract db interactions
# from this class
def train_positive(self):
"""For each token in self.tokens, add token/counter to positive_classification,
and/or increment positive_counter in database"""
self.database.train_positive(self.tokens)
return None
def untrain_positive(self):
"""For each token in tokens, decrement token counter in positive_classification.
If token counter == 0, remove token from table"""
self.database.untrain_positive(self, tokens)
return None
def train_negative(self):
"""For each token in tokens, add token/counter to negative_classification,
and/or increment negative_counter in database."""
self.database.train_negative(self.tokens)
return None
def untrain_negative(self):
"""For each token in tokens, decrement token counter in negative_classification.
If token counter == 0, remove token from table"""
self.database.untrain_negative(self, tokens)
##############################
# bayes formula operations
def calculate_probabilities(self):
"""Stores the Token.token_string values, as compared positive and negative
in the database, within the Token.positive_value and Token.negative_value,
respectively."""
total_positive = self.database.total_for_polarity(polarity='positive')
total_negative = self.database.total_for_polarity(polarity='negative')
for token in self.tokens:
positive_count = self.database.counter_for_token(
token.token_string, polarity='positive')
negative_count = self.database.counter_for_token(
token.token_string, polarity='negative')
numerator = (float(positive_count) / total_positive)
denominator = ((float(positive_count) / total_positive) +
(float(negative_count) / total_negative))
token.positive_value = numerator / denominator
numerator = (float(negative_count) / total_negative)
denominator = ((float(negative_count) / total_negative) +
(float(positive_count) / total_positive))
token.negative_value = numerator / denominator
return True
def sum_positive(self):
"""p(S) = (p1 * p2 ... pn) /
( (p1 * p2 ... * pn) + ( (1 - p1) * (1 - p2) ... * (1 - pn) ) )"""
numerator = sum([token.positive_value for token in self.tokens])
denominator = numerator + sum(
[1 - token.positive_value for token in self.tokens])
return float(numerator) / denominator
def sum_negative(self):
"""p(S) = (p1 * p2 ... pn) /
( (p1 * p2 ... * pn) + ( (1 - p1) * (1 - p2) ... * (1 - pn) ) )"""
numerator = sum([token.negative_value for token in self.tokens])
denominator = numerator + sum(
[1 - token.negative_value for token in self.tokens])
return float(numerator) / denominator