print("\n > running in FINAL mode!\n")
#training, testing = loader.load_final()
data_folder = [
"../dataset/brexit/BrexitOpposite.txt",
"../dataset/brexit/BrexitNeutral.txt",
"../dataset/brexit/BrexitSupport.txt"
]
training, testing = loader.load_stance_brexit_5cross(
wholeFile, cross_num)
else:
training, validation, testing = loader.load_train_val_test()
if SEMEVAL_GOLD:
print("\n > running in Post-Mortem mode!\n")
gold_data = SemEvalDataLoader().get_gold(task=TASK)
gX = [obs[1] for obs in gold_data]
gy = [obs[0] for obs in gold_data]
gold = prepare_dataset(gX, gy, loader.pipeline, loader.y_one_hot)
validation = testing
testing = gold
FINAL = False
############################################################################
# NN MODEL
# ------------
# Uncomment one of the following model definitions, in order to define a model
############################################################################
print("Building NN Model...")
# nn_model = build_attention_RNN(embeddings, classes=3, max_length=max_length,
def __init__(self,
word_indices,
text_lengths,
subtask="A",
silver=False,
**kwargs):
self.word_indices = word_indices
filter_classes = kwargs.get("filter_classes", None)
self.y_one_hot = kwargs.get("y_one_hot", True)
self.pipeline = Pipeline([
('preprocess',
CustomPreProcessor(
TextPreProcessor(
backoff=[
'url', 'email', 'percent', 'money', 'phone', 'user',
'time', 'url', 'date', 'number'
],
include_tags={
"hashtag", "allcaps", "elongated", "repeated",
'emphasis', 'censored'
},
fix_html=True,
segmenter="twitter",
corrector="twitter",
unpack_hashtags=True,
unpack_contractions=True,
spell_correct_elong=False,
tokenizer=SocialTokenizer(lowercase=True).tokenize,
dicts=[emoticons]))),
('ext',
EmbeddingsExtractor(word_indices=word_indices,
max_lengths=text_lengths,
add_tokens=(False,
True) if subtask != "A" else True,
unk_policy="random"))
])
# loading data
print("Loading data...")
dataset = SemEvalDataLoader(verbose=False).get_data(task=subtask,
years=None,
datasets=None,
only_semeval=True)
random.Random(42).shuffle(dataset)
if filter_classes:
dataset = [d for d in dataset if d[0] in filter_classes]
self.X = [obs[1] for obs in dataset]
self.y = [obs[0] for obs in dataset]
print("total observations:", len(self.y))
print("-------------------\ntraining set stats\n-------------------")
print_dataset_statistics(self.y)
print("-------------------")
if silver:
print("Loading silver data...")
dataset = SemEvalDataLoader().get_silver()
self.silver_X = [obs[1] for obs in dataset]
self.silver_y = [obs[0] for obs in dataset]
print("total observations:", len(self.silver_y))
from utilities.data_loader import get_embeddings
from utilities.sklearn import eval_clf, nbow_model
numpy.random.seed(1337) # for reproducibility
def tok(text):
return text
WV_CORPUS = "datastories.twitter"
WV_DIM = 300
embeddings, word_indices = get_embeddings(corpus=WV_CORPUS, dim=WV_DIM)
train_set = SemEvalDataLoader(verbose=False).get_data(task="A",
years=None,
datasets=None,
only_semeval=True)
X = [obs[1] for obs in train_set]
y = [obs[0] for obs in train_set]
test_data = SemEvalDataLoader(verbose=False).get_gold(task="A")
X_test = [obs[1] for obs in test_data]
y_test = [obs[0] for obs in test_data]
print("-----------------------------")
print("LinearSVC")
nbow = nbow_model("clf", embeddings, word_indices)
nbow.fit(X, y)
results = eval_clf(nbow.predict(X_test), y_test)
for res, val in results.items():
print("{}: {:.3f}".format(res, val))