def add_seeds(self):
'''adds seeds to db.coll with reliability score of 1.0'''
self.logger.debug('add_seeds: %d %s' % (len(self.seeds), self.seeds))
for s in self.seeds:
self.logger.debug('seed: %s' % s)
args = s.split('\t')
doc = {'arg%d' % n: v for n, v in enumerate(args, 1)}
doc['it'] = 0
doc['score'] = 1.0
mongodb.cache(self.db, self.boot_i, doc)
def add_seeds(self):
'''adds seeds to db.coll with reliability score of 1.0'''
self.logger.debug('add_seeds: %d %s' %
(len(self.seeds), self.seeds))
for s in self.seeds:
self.logger.debug('seed: %s' % s)
args = s.split('\t')
doc = {'arg%d'%n:v
for n,v in enumerate(args, 1)}
doc['it'] = 0
doc['score'] = 1.0
mongodb.cache(self.db, self.boot_i, doc)
def iterate_p(self, mutexes=[]):
'''perform an iteration of bootstrapping saving n patterns with the
highest reliability score'''
if not getattr(self, 'connection', None):
self.init_connection()
self.logger.info(' ### BOOTSTRAPPING PATTERN ITERATION: %d ###' %
self.it)
# read promoted instances of last bootstrpping iteration
self.logger.info('getting promoted instances...' '')
I = self.get_I(self.it - 1)
self.logger.info('I: %d' % len(I))
self.logger.info('getting promoted instances: done.' '')
# find matching patterns
self.logger.info('getting matching patterns...')
P_ = self.I2P(I)
P = self.mutex_filter_p(I, P_, mutexes)
self.logger.info('getting matching patterns: done.')
# rank patterns by reliability score
self.logger.info('ranking patterns ...')
rs = self.scorer.rank_patterns(I, P, self.it)
self.logger.info('ranking patterns: done.')
# save top n to <matrix>_boot_p
self.logger.info('saving top %d patterns...' % self.n)
for r in rs[:self.n]:
self.logger.info('r: %s' % r)
mongodb.cache(self.db, self.boot_p, r)
self.logger.info('saving top %d patterns: done.' % self.n)
self.logger.info('ensuring indices ...')
# index for iteration number
self.db[self.boot_p].ensure_index([
('it', pymongo.DESCENDING),
])
# index for <REL>
self.db[self.boot_p].ensure_index([
('rel', pymongo.ASCENDING),
])
self.logger.info('ensuring indices: done.')
def iterate_i(self, mutexes=[]):
'''perform an iteration of bootstrapping saving n instances with the
highest reliability score'''
if not getattr(self, 'connection', None):
self.init_connection()
self.logger.info(' ### BOOTSTRAPPING INSTANCE ITERATION: %d ###' %
self.it)
# read promoted patterns of last bootstrpping iteration
self.logger.info('getting promoted patterns...' '')
P = self.get_P(self.it)
self.logger.info('P: %d' % len(P))
self.logger.info('getting promoted patterns: done.' '')
# find matching instances
self.logger.info('getting matching instances...')
I_ = self.P2I(P)
I = self.mutex_filter_i(I_, P, mutexes)
self.logger.info('getting matching instances: done.')
# rank instances by reliability score
self.logger.info('ranking instances ...')
rs = self.scorer.rank_instances(I, P, self.it)
self.logger.info('ranking instances: done.')
# save top n to <matrix>_boot_p
self.logger.info('saving top %d instances...' % self.n)
for r in rs[:self.n]:
self.logger.info('r: %s' % r)
mongodb.cache(self.db, self.boot_i, r)
self.logger.info('saving top %d instances: done.' % self.n)
self.logger.info('ensuring indices ...')
# index for iteration number
self.db[self.boot_i].ensure_index([
('it', pymongo.DESCENDING),
])
# index for <ARGJ,...,ARGN>
self.db[self.boot_i].ensure_index([(arg, pymongo.ASCENDING)
for arg in self.args])
self.logger.info('ensuring indices: done.')
def iterate_i(self, mutexes=[]):
'''perform an iteration of bootstrapping saving n instances with the
highest reliability score'''
if not getattr(self, 'connection', None):
self.init_connection()
self.logger.info(' ### BOOTSTRAPPING INSTANCE ITERATION: %d ###' %
self.it)
# read promoted patterns of last bootstrpping iteration
self.logger.info('getting promoted patterns...''')
P = self.get_P(self.it)
self.logger.info('P: %d' % len(P))
self.logger.info('getting promoted patterns: done.''')
# find matching instances
self.logger.info('getting matching instances...')
I_ = self.P2I(P)
I = self.mutex_filter_i(I_, P, mutexes)
self.logger.info('getting matching instances: done.')
# rank instances by reliability score
self.logger.info('ranking instances ...')
rs = self.scorer.rank_instances(I, P, self.it)
self.logger.info('ranking instances: done.')
# save top n to <matrix>_boot_p
self.logger.info('saving top %d instances...' % self.n)
for r in rs[:self.n]:
self.logger.info('r: %s' % r)
mongodb.cache(self.db, self.boot_i, r)
self.logger.info('saving top %d instances: done.' % self.n)
self.logger.info('ensuring indices ...')
# index for iteration number
self.db[self.boot_i].ensure_index( [('it', pymongo.DESCENDING), ] )
# index for <ARGJ,...,ARGN>
self.db[self.boot_i].ensure_index(
[(arg, pymongo.ASCENDING)
for arg in self.args]
)
self.logger.info('ensuring indices: done.')
def iterate_p(self, mutexes=[]):
'''perform an iteration of bootstrapping saving n patterns with the
highest reliability score'''
if not getattr(self, 'connection', None):
self.init_connection()
self.logger.info(' ### BOOTSTRAPPING PATTERN ITERATION: %d ###' %
self.it)
# read promoted instances of last bootstrpping iteration
self.logger.info('getting promoted instances...''')
I = self.get_I(self.it-1)
self.logger.info('I: %d' % len(I))
self.logger.info('getting promoted instances: done.''')
# find matching patterns
self.logger.info('getting matching patterns...')
P_ = self.I2P(I)
P = self.mutex_filter_p(I, P_, mutexes)
self.logger.info('getting matching patterns: done.')
# rank patterns by reliability score
self.logger.info('ranking patterns ...')
rs = self.scorer.rank_patterns(I, P, self.it)
self.logger.info('ranking patterns: done.')
# save top n to <matrix>_boot_p
self.logger.info('saving top %d patterns...' % self.n)
for r in rs[:self.n]:
self.logger.info('r: %s' % r)
mongodb.cache(self.db, self.boot_p, r)
self.logger.info('saving top %d patterns: done.' % self.n)
self.logger.info('ensuring indices ...')
# index for iteration number
self.db[self.boot_p].ensure_index( [('it', pymongo.DESCENDING), ] )
# index for <REL>
self.db[self.boot_p].ensure_index( [('rel', pymongo.ASCENDING), ] )
self.logger.info('ensuring indices: done.')
