class PhraseClassifier(object):
""" Classify a phrase based on a pre-existing model """
def __init__(self, directory, label):
metafile = os.path.join(directory, label + "-meta.json")
with open(metafile, 'r') as v:
all_vars = json.load(v)
self.vocab = generateVocabVectors(all_vars["vocab"])
self.max_vector_len = all_vars["max_vector_len"]
self.max_phrase_len = all_vars["max_phrase_len"]
self.net_sizes = all_vars["sizes"]
self.targets = all_vars["targets"]
model_filename = os.path.join(directory, label + ".model")
state_filename = os.path.join(directory, label + ".state")
self.classifier = Classifier(self.net_sizes, model_filename, state_filename)
def classify(self, phrase, cut_to_len=True):
""" Classify a phrase based on the loaded model. If cut_to_len is True, cut to
desired length."""
if (len(phrase) > self.max_phrase_len):
if not cut_to_len:
raise Exception("Phrase too long.")
phrase = phrase[0:self.max_phrase_len]
numbers = self.classifier.classify(stringToVector(phrase, self.vocab, self.max_vector_len))
return zip(self.targets, numbers)
def __init__(self, directory, label):
metafile = os.path.join(directory, label + "-meta.json")
with open(metafile, 'r') as v:
all_vars = json.load(v)
self.vocab = generateVocabVectors(all_vars["vocab"])
self.max_vector_len = all_vars["max_vector_len"]
self.max_phrase_len = all_vars["max_phrase_len"]
self.net_sizes = all_vars["sizes"]
self.targets = all_vars["targets"]
model_filename = os.path.join(directory, label + ".model")
state_filename = os.path.join(directory, label + ".state")
self.classifier = Classifier(self.net_sizes, model_filename, state_filename)