def bestWordCohOnTestcro(subsample=None):
topics = subsampleList(croelectTopics,
subsample) if subsample else croelectTopics
experiment(palmettoBestCrowiki(),
action='run',
topics=topics,
posClass=['theme', 'theme_noise'])
experiment(palmettoBestCrowiki(),
action='signif',
topics=topics,
posClass=['theme', 'theme_noise'])
def bestDocCohOnTestcro(subsample=None):
topics = subsampleList(croelectTopics,
subsample) if subsample else croelectTopics
experiment(bestParamsDocCroelect(),
action='run',
topics=topics,
posClass=['theme', 'theme_noise'])
experiment(bestParamsDocCroelect(),
action='signif',
topics=topics,
posClass=['theme', 'theme_noise'])
def evaluateParamsOnDev(action='run', subsample=None, type=None):
'''
Evaluate measure parameters on the development set, ie
construct the measures from parameter sets and evaluate measures.
:param action: 'run' to evaluate, 'print' to print results
:param subsample: if not None, size of dev subsample to run measures on, for testing
:return:
'''
if subsample:
from doc_topic_coh.coherence.measure_evaluation.utils import subsampleList
evalTopics = subsampleList(dev, subsample)
else:
evalTopics = dev
if not type or type.startswith('graph'):
runGridGraphCorpus(action, evalTopics)
runGridGraphWorld(action, evalTopics)
if not type or type.startswith('distance'):
runGridDistance('corpus', action, evalTopics)
runGridDistance('world', action, evalTopics)
if not type or type.startswith('density'):
runGridDensity('corpus', action, evalTopics)
runGridDensity('world', action, evalTopics)
def bestDocCohOnTest(subsample=None):
topics = subsampleList(test, subsample) if subsample else test
experiment(bestParamsDoc(), topics=topics, action='run')
experiment(bestParamsDoc(), topics=topics, action='signif')
def wordCohOnTest(subsample=None):
topics = subsampleList(test, subsample) if subsample else test
experiment(palmettoWordCoherence(True), topics=topics, action='run')
experiment(palmettoWordCoherence(True), topics=topics, action='signif')