def __init__(self,
alpha=0,
decay=1,
ignore_leaves=True,
smoothed=True,
vector='word2vec',
w2vdim=300,
lowercase=True,
tree='tree',
kernel_path=KERNEL_PATH):
Semeval.__init__(self,
vector=vector,
stop=False,
lowercase=lowercase,
punctuation=False,
w2vdim=w2vdim)
self.path = kernel_path
self.tree = tree
self.memoization = {}
self.svm = Model()
self.flat_traindata()
self.treekernel = TreeKernel(alpha=alpha,
decay=decay,
ignore_leaves=ignore_leaves,
smoothed=smoothed,
lowercase=lowercase)
self.train()
del self.additional
def __init__(self,
model='svm',
features='bm25,',
comment_features='bm25,',
stop=True,
vector='word2vec',
lowercase=True,
punctuation=True,
proctrain=True,
path=FEATURES_PATH,
alpha=0.1,
sigma=0.9,
gridsearch='random'):
Semeval.__init__(self,
stop=stop,
vector=vector,
lowercase=lowercase,
punctuation=punctuation)
self.path = path
self.features = features.split(',')
self.comment_features = comment_features.split(',')
self.gridsearch = gridsearch
self.svm = Model()
self.model = model
self.bm25 = SemevalBM25(
stop=stop,
lowercase=lowercase,
punctuation=punctuation,
proctrain=proctrain
) if 'bm25' in self.features + self.comment_features else None
self.cosine = SemevalCosine(
stop=stop,
lowercase=lowercase,
punctuation=punctuation,
proctrain=proctrain
) if 'cosine' in self.features + self.comment_features else None
self.softcosine = SemevalSoftCosine(
stop=stop,
lowercase=lowercase,
punctuation=punctuation,
proctrain=proctrain,
vector=vector
) if 'softcosine' in self.features + self.comment_features else None
self.translation = SemevalTranslation(
alpha=alpha,
sigma=sigma,
stop=stop,
lowercase=lowercase,
punctuation=punctuation,
proctrain=proctrain,
vector=self.vector
) if 'translation' in self.features + self.comment_features else None
self.train()
def __init__(self, model='svm', features='bm25,', comment_features='bm25,', stop=True, vector='word2vec', path=FEATURES_PATH, alpha=0.1, sigma=0.9, gridsearch='random'):
Quora.__init__(self, stop=stop, vector=vector)
self.path = path
self.features = features.split(',')
self.comment_features = comment_features.split(',')
self.gridsearch = gridsearch
self.svm = Model()
self.model = model
self.bm25 = QuoraBM25(stop=stop) if 'bm25' in self.features+self.comment_features else None
self.cosine = QuoraCosine(stop=stop) if 'cosine' in self.features+self.comment_features else None
self.softcosine = QuoraSoftCosine(stop=stop, vector=vector) if 'softcosine' in self.features+self.comment_features else None
self.translation = QuoraTranslations(alpha=alpha, sigma=sigma, stop=stop, vector=self.vector) if 'translation' in self.features+self.comment_features else None
self.train()
def __init__(self, stop={}, lowercase={}, punctuation={}, vector={}, scale=True, alpha=0.9, sigma=0.1):
self.stop = stop
self.lowercase = lowercase
self.punctuation = punctuation
self.scale = scale
self.vector = vector
self.alpha = alpha
self.sigma = sigma
self.questions, self.ranking = self.load()
self.ensemble = Model()
self.train()
ranking = self.test()
p.dump(ranking, open(os.path.join(SEMI_PATH, 'reranking'), 'wb'))
def __init__(self,
stop={},
lowercase={},
punctuation={},
vector={},
scale=True,
w2vdim=300,
kernel_path='',
alpha=0.8,
sigma=0.2):
self.stop = stop
self.lowercase = lowercase
self.punctuation = punctuation
self.scale = scale
self.vector = vector
self.alpha = alpha
self.sigma = sigma
self.kernel_path = kernel_path
self.w2vdim = w2vdim
self.theta = 0.9
self.ensemble = Model()
self.train()