def evalTopValues(algo, vectors, plot=False):
if algo == 'distance':
experiment(distanceParams(vectors),
action='eval',
evalTopics=test,
evalPerc=0.95,
plotEval=plot)
elif algo == 'graph':
if vectors == 'corpus':
experiment(graphParams('tf-idf', cosine),
action='eval',
evalTopics=test,
evalPerc=0.95,
plotEval=plot)
elif vectors == 'world':
experiment(graphParams('word2vec', cosine),
action='eval',
evalTopics=test,
evalPerc=0.95,
plotEval=plot)
elif algo == 'gauss':
experiment(densityParams(vectors),
action='eval',
evalTopics=test,
evalPerc=0.95,
plotEval=plot)
def printTopSelected(algo, vectors):
if algo == 'distance':
if vectors == 'corpus':
experiment(distanceParams(vectors),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh=0.744)
elif vectors == 'world':
experiment(distanceParams(vectors),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh=0.7316)
elif algo == 'graph':
if vectors == 'corpus':
experiment(graphParams('tf-idf', cosine),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh='median',
th2per=thres2perc)
elif vectors == 'world':
experiment(graphParams('word2vec', cosine),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh='median',
th2per=thres2perc)
elif algo == 'gauss':
if vectors == 'corpus':
experiment(densityParams(vectors),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh='median')
elif vectors == 'world':
experiment(densityParams(vectors),
action='printTop',
evalTopics=test,
evalPerc=0.95,
evalThresh=0.715)
def runGridDensity(vectors, action='run', topics=dev):
experiment(densityParams(vectors), action=action, topics=topics)
def runGridDistance(vectors, action='run', topics=dev):
experiment(distanceParams(vectors), topics=topics, action=action)
def runGridGraphWorld(action='run', topics=dev):
if action == 'print':
experiment(graphParams('word2vec', cosine),
action='print',
topics=topics)
else:
experiment(graphParams('word2vec', cosine), topics=topics)
experiment(graphParams('word2vec', l1), topics=topics)
experiment(graphParams('word2vec-avg', l1), topics=topics)
experiment(graphParams('word2vec', l2), topics=topics)
experiment(graphParams('word2vec-avg', l2), topics=topics)
experiment(graphParams('glove', cosine), topics=topics)
experiment(graphParams('glove', l1), topics=topics)
experiment(graphParams('glove-avg', l1), topics=topics)
experiment(graphParams('glove', l2), topics=topics)
experiment(graphParams('glove-avg', l2), topics=topics)
def runGridGraphCorpus(action='run', topics=dev):
if action == 'print':
experiment(graphParams('tf-idf', cosine), action=action, topics=topics)
else:
experiment(graphParams('tf-idf', cosine), topics=topics)
experiment(graphParams('tf-idf', l1), topics=topics)
experiment(graphParams('tf-idf', l2), topics=topics)
experiment(graphParams('probability', cosine), topics=topics)
experiment(graphParams('probability', l1), topics=topics)
experiment(graphParams('probability', l2), topics=topics)
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')