def get_spacy_model(spacy_model_name: str, pos_tags: bool, parse: bool, ner: bool) -> SpacyModelType:
"""
In order to avoid loading spacy models a whole bunch of times, we'll save references to them,
keyed by the options we used to create the spacy model, so any particular configuration only
gets loaded once.
"""
options = (spacy_model_name, pos_tags, parse, ner)
if options not in LOADED_SPACY_MODELS:
disable = ['vectors', 'textcat']
if not pos_tags:
disable.append('tagger')
if not parse:
disable.append('parser')
if not ner:
disable.append('ner')
try:
spacy_model = spacy.load(spacy_model_name, disable=disable)
except OSError:
logger.warning(f"Spacy models '{spacy_model_name}' not found. Downloading and installing.")
spacy_download(spacy_model_name)
spacy_model = spacy.load(spacy_model_name, disable=disable)
LOADED_SPACY_MODELS[options] = spacy_model
return LOADED_SPACY_MODELS[options]
def main(model=None, new_model_name='animal', output_dir=None, n_iter=20):
"""Set up the pipeline and entity recognizer, and train the new entity."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
# Add entity recognizer to model if it's not in the pipeline
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'ner' not in nlp.pipe_names:
ner = nlp.create_pipe('ner')
nlp.add_pipe(ner)
# otherwise, get it, so we can add labels to it
else:
ner = nlp.get_pipe('ner')
ner.add_label(LABEL) # add new entity label to entity recognizer
if model is None:
optimizer = nlp.begin_training()
else:
# Note that 'begin_training' initializes the models, so it'll zero out
# existing entity types.
optimizer = nlp.entity.create_optimizer()
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'ner']
with nlp.disable_pipes(*other_pipes): # only train NER
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for text, annotations in TRAIN_DATA:
nlp.update([text], [annotations], sgd=optimizer, drop=0.35,
losses=losses)
print(losses)
# test the trained model
test_text = 'Do you like horses?'
doc = nlp(test_text)
print("Entities in '%s'" % test_text)
for ent in doc.ents:
print(ent.label_, ent.text)
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.meta['name'] = new_model_name # rename model
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc2 = nlp2(test_text)
for ent in doc2.ents:
print(ent.label_, ent.text)
def get_nlp(lang="en"):
"""Load spaCy model for a given language, determined by `models' dict or by MODEL_ENV_VAR"""
instance = nlp.get(lang)
if instance is None:
import spacy
model = models.get(lang)
if not model:
models[lang] = model = os.environ.get("_".join((MODEL_ENV_VAR, lang.upper()))) or \
os.environ.get(MODEL_ENV_VAR) or DEFAULT_MODEL.get(lang, "xx")
started = time.time()
with external_write_mode():
print("Loading spaCy model '%s'... " % model, end="", flush=True)
try:
nlp[lang] = instance = spacy.load(model)
except OSError:
spacy.cli.download(model)
try:
nlp[lang] = instance = spacy.load(model)
except OSError as e:
raise OSError("Failed to get spaCy model. Download it manually using "
"`python -m spacy download %s`." % model) from e
tokenizer[lang] = instance.tokenizer
instance.tokenizer = lambda words: spacy.tokens.Doc(instance.vocab, words=words)
print("Done (%.3fs)." % (time.time() - started))
return instance
def _load_spacy(self, evaluation, options):
language_code = None
language_name = self.get_option(options, 'Language', evaluation)
if language_name is None:
language_name = String('Undefined')
if isinstance(language_name, String):
language_code = _SpacyBuiltin._language_codes.get(language_name.get_string_value())
if not language_code:
evaluation.message(self.get_name(), 'lang', language_name, strip_context(self.get_name()))
return None
instance = _SpacyBuiltin._spacy_instances.get(language_code)
if instance:
return instance
try:
if 'SPACY_DATA' in os.environ:
instance = spacy.load(language_code, via=os.environ['SPACY_DATA'])
else:
instance = spacy.load(language_code)
_SpacyBuiltin._spacy_instances[language_code] = instance
return instance
except RuntimeError as e:
evaluation.message(self.get_name(), 'runtime', str(e))
return None
def main(model=None, output_dir=None, n_iter=15):
"""Load the model, set up the pipeline and train the parser."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank("en") # create blank Language class
print("Created blank 'en' model")
# add the parser to the pipeline if it doesn't exist
# nlp.create_pipe works for built-ins that are registered with spaCy
if "parser" not in nlp.pipe_names:
parser = nlp.create_pipe("parser")
nlp.add_pipe(parser, first=True)
# otherwise, get it, so we can add labels to it
else:
parser = nlp.get_pipe("parser")
# add labels to the parser
for _, annotations in TRAIN_DATA:
for dep in annotations.get("deps", []):
parser.add_label(dep)
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "parser"]
with nlp.disable_pipes(*other_pipes): # only train parser
optimizer = nlp.begin_training()
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
# batch up the examples using spaCy's minibatch
batches = minibatch(TRAIN_DATA, size=compounding(4.0, 32.0, 1.001))
for batch in batches:
texts, annotations = zip(*batch)
nlp.update(texts, annotations, sgd=optimizer, losses=losses)
print("Losses", losses)
# test the trained model
test_text = "I like securities."
doc = nlp(test_text)
print("Dependencies", [(t.text, t.dep_, t.head.text) for t in doc])
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc = nlp2(test_text)
print("Dependencies", [(t.text, t.dep_, t.head.text) for t in doc])
def __init__(self, model='en', disable=None):
if disable is None:
disable = []
try:
self._parser = spacy.load(model, disable=disable)
except OSError:
url = 'https://spacy.io/models'
if license_prompt('Spacy {} model'.format(model), url) is False:
sys.exit(0)
spacy_download(model)
self._parser = spacy.load(model, disable=disable)
def get_ents():
data = flask.request.args.get('fragment')
is_custom = flask.request.args.get('custom')
if is_custom is not None:
nlp = spacy.load(Path('./gina_haspel'))
else:
nlp = spacy.load('en')
doc = nlp(data)
print(doc)
tuples = [(str(x), x.label_)
for x
in doc.ents]
return flask.jsonify(dict(tuples))
def get_nlp():
if nlp.instance is None:
import spacy
model_name = os.environ.get("SPACY_MODEL", "en")
nlp.instance = spacy.load(model_name)
if nlp.instance.tagger is None: # Model not really loaded
spacy.cli.download(model_name)
nlp.instance = spacy.load(model_name)
assert nlp.instance.tagger, "Failed to get spaCy model. " \
"Download it manually using `python -m spacy download %s`." % model_name
nlp.tokenizer = nlp.instance.tokenizer
nlp.instance.tokenizer = nlp.tokenizer.tokens_from_list
return nlp.instance
def _spacy_en():
yield None
try:
spacyen = spacy.load('en_default', via=data_path)
except RuntimeError as e:
if e.message == "Model not installed. Please run 'python -m spacy.en.download' to install latest compatible model.":
print("Need to download Spacy data. Starting download now")
sputnik.install('spacy', spacy.about.__version__,
'en_default', data_path=data_path)
spacyen = spacy.load('en_default', via=data_path)
else:
raise
while True:
yield spacyen
def main(model=None, output_dir=None, n_iter=15):
"""Load the model, set up the pipeline and train the parser."""
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank("en") # create blank Language class
print("Created blank 'en' model")
# We'll use the built-in dependency parser class, but we want to create a
# fresh instance – just in case.
if "parser" in nlp.pipe_names:
nlp.remove_pipe("parser")
parser = nlp.create_pipe("parser")
nlp.add_pipe(parser, first=True)
for text, annotations in TRAIN_DATA:
for dep in annotations.get("deps", []):
parser.add_label(dep)
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "parser"]
with nlp.disable_pipes(*other_pipes): # only train parser
optimizer = nlp.begin_training()
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
# batch up the examples using spaCy's minibatch
batches = minibatch(TRAIN_DATA, size=compounding(4.0, 32.0, 1.001))
for batch in batches:
texts, annotations = zip(*batch)
nlp.update(texts, annotations, sgd=optimizer, losses=losses)
print("Losses", losses)
# test the trained model
test_model(nlp)
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
# test the saved model
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
test_model(nlp2)
def _load_nlp(self):
if self.nlp is None:
print("Loading SpacySentenceWordCache")
nlp = spacy.load('en_core_web_lg')
nlp.add_pipe(nlp.create_pipe('sentencizer'))
self.nlp = nlp
return self.nlp
def main(lang, in_dir, out_loc, negative=5, n_workers=4, window=5, size=128, min_count=10, nr_iter=2):
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
model = Word2Vec(
size=size,
window=window,
min_count=min_count,
workers=n_workers,
sample=1e-5,
negative=negative
)
nlp = spacy.load(lang, parser=False, tagger=False, entity=False)
corpus = Corpus(in_dir)
total_words = 0
total_sents = 0
for text_no, text_loc in enumerate(iter_dir(corpus.directory)):
with io.open(text_loc, 'r', encoding='utf8') as file_:
text = file_.read()
total_sents += text.count('\n')
doc = nlp(text)
total_words += corpus.count_doc(doc)
logger.info("PROGRESS: at batch #%i, processed %i words, keeping %i word types",
text_no, total_words, len(corpus.strings))
model.corpus_count = total_sents
model.raw_vocab = defaultdict(int)
for orth, freq in corpus.counts:
if freq >= min_count:
model.raw_vocab[nlp.vocab.strings[orth]] = freq
model.scale_vocab()
model.finalize_vocab()
model.iter = nr_iter
model.train(corpus)
model.save(out_loc)
def train(train_loc, dev_loc, shape, settings):
train_texts1, train_texts2, train_labels = read_snli(train_loc)
dev_texts1, dev_texts2, dev_labels = read_snli(dev_loc)
print("Loading spaCy")
nlp = spacy.load('en')
assert nlp.path is not None
print("Compiling network")
model = build_model(get_embeddings(nlp.vocab), shape, settings)
print("Processing texts...")
Xs = []
for texts in (train_texts1, train_texts2, dev_texts1, dev_texts2):
Xs.append(get_word_ids(list(nlp.pipe(texts, n_threads=20, batch_size=20000)),
max_length=shape[0],
rnn_encode=settings['gru_encode'],
tree_truncate=settings['tree_truncate']))
train_X1, train_X2, dev_X1, dev_X2 = Xs
print(settings)
model.fit(
[train_X1, train_X2],
train_labels,
validation_data=([dev_X1, dev_X2], dev_labels),
nb_epoch=settings['nr_epoch'],
batch_size=settings['batch_size'])
if not (nlp.path / 'similarity').exists():
(nlp.path / 'similarity').mkdir()
print("Saving to", nlp.path / 'similarity')
weights = model.get_weights()
with (nlp.path / 'similarity' / 'model').open('wb') as file_:
pickle.dump(weights[1:], file_)
with (nlp.path / 'similarity' / 'config.json').open('wb') as file_:
file_.write(model.to_json())
def train(train_texts, train_labels, dev_texts, dev_labels,
lstm_shape, lstm_settings, lstm_optimizer, batch_size=100, nb_epoch=5,
by_sentence=True):
nlp = spacy.load('en', entity=False)
if 'nr_vector' not in lstm_shape:
lstm_shape['nr_vector'] = max(lex.rank+1 for lex in vocab if lex.has_vector)
print("Make model")
model = Classifier(SentimentModel(nlp, lstm_shape, **lstm_settings))
print("Parsing texts...")
if by_sentence:
train_data = SentenceDataset(nlp, train_texts, train_labels, lstm_shape['max_length'])
dev_data = SentenceDataset(nlp, dev_texts, dev_labels, lstm_shape['max_length'])
else:
train_data = DocDataset(nlp, train_texts, train_labels)
dev_data = DocDataset(nlp, dev_texts, dev_labels)
train_iter = SerialIterator(train_data, batch_size=batch_size,
shuffle=True, repeat=True)
dev_iter = SerialIterator(dev_data, batch_size=batch_size,
shuffle=False, repeat=False)
optimizer = chainer.optimizers.Adam()
optimizer.setup(model)
updater = chainer.training.StandardUpdater(train_iter, optimizer, device=0)
trainer = chainer.training.Trainer(updater, (20, 'epoch'), out='result')
trainer.extend(extensions.Evaluator(dev_iter, model, device=0))
trainer.extend(extensions.LogReport())
trainer.extend(extensions.PrintReport([
'epoch', 'main/accuracy', 'validation/main/accuracy']))
trainer.extend(extensions.ProgressBar())
trainer.run()
def test_access_parse_for_merged():
nlp = spacy.load('en_core_web_sm')
t_t = nlp.tokenizer("Highly rated - I'll definitely")
nlp.tagger(t_t)
nlp.parser(t_t)
nlp.parser(t_t)
ss(t_t)
def __init__(self, model=None, output_dir=None, n_iter=50, threshold_success=None, successive_successes=3, input_data = "TRAINING_DATA.json", param='ner'):
"""Load the model, set up the pipeline and train the entity recognizer."""
if model is not None:
self.nlp = nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
self.nlp = nlp = spacy.load('de', parser=False) # load default `de` model
print("Loaded existing 'de' model")
self.output_dir = output_dir
self.n_iter = n_iter
self.threshold_success = threshold_success
self.successive_successes = successive_successes
self.TRAIN_DATA = json.load(open(input_data))
self.train(param)
self.test()
def __init__(self,lang='en'):
try:
import spacy
except:
raise Exception("spacy not installed. Use `pip install spacy`.")
super(SpaCy, self).__init__(name="spaCy")
self.model = spacy.load('en')
def main(argv):
if not argv or len(argv) < 2:
raise TypeError("not enough arguments. two are required")
# Load Spacy's English tokenizer model
print "Loading Spacy's English model"
nlp = spacy.load('en')
# Download two news articles from internet, supplied as urls in command line arguments
utext1 = getOnlyText(argv[0])
utext2 = getOnlyText(argv[1])
# Use Spacy to tokenize documents, then remove stop words and non-alpha
print "Parsing files"
doc1 = filterText(nlp(utext1), nlp)
doc2 = filterText(nlp(utext2), nlp)
# Similarity is estimated using the cosine metric, between Span.vector and other.vector.
# By default, Span.vector is computed by averaging the vectors of its tokens. [spacy.io]
print "Document Vectors Similarity Score:", doc1.similarity(doc2)
# Build Bag of Words with Spacy
wordBag1 = doc1.count_by(LOWER)
wordBag2 = doc2.count_by(LOWER)
# Combine Bag of Words dicts in vector format, calculate cosine similarity of resulting vectors
vect = DictVectorizer(sparse=False)
wordbagVectors = vect.fit_transform([wordBag2, wordBag1])
score = 1 - spatial.distance.cosine(wordbagVectors[0], wordbagVectors[1])
print "Bag of Words Cosine Similarity Score:", score
return score
def create_nlp_instance():
import spacy
from spacymoji import Emoji
nlp = spacy.load('en')
emoji_pipe = Emoji(nlp)
nlp.add_pipe(emoji_pipe, first=True)
# Merge hashtag tokens which were split by spacy
def hashtag_pipe(doc):
merged_hashtag = False
while True:
for token_index, token in enumerate(doc):
if token.text == '#':
if token.head is not None:
start_index = token.idx
end_index = start_index + len(token.head.text) + 1
if doc.merge(start_index, end_index) is not None:
merged_hashtag = True
break
if not merged_hashtag:
break
merged_hashtag = False
return doc
nlp.add_pipe(hashtag_pipe)
return nlp
def main():
train, dev = datasets.imdb()
train_X, train_y = zip(*train)
dev_X, dev_y = zip(*dev)
model = LinearModel(2)
train_y = to_categorical(train_y, nb_classes=2)
dev_y = to_categorical(dev_y, nb_classes=2)
nlp = spacy.load("en")
train_X = [
model.ops.asarray([tok.orth for tok in doc], dtype="uint64")
for doc in nlp.pipe(train_X)
]
dev_X = [
model.ops.asarray([tok.orth for tok in doc], dtype="uint64")
for doc in nlp.pipe(dev_X)
]
dev_X = preprocess(model.ops, dev_X)
with model.begin_training(train_X, train_y, L2=1e-6) as (trainer, optimizer):
trainer.dropout = 0.0
trainer.batch_size = 512
trainer.nb_epoch = 3
trainer.each_epoch.append(lambda: print(model.evaluate(dev_X, dev_y)))
for X, y in trainer.iterate(train_X, train_y):
keys_vals_lens = preprocess(model.ops, X)
scores, backprop = model.begin_update(keys_vals_lens, drop=trainer.dropout)
backprop(scores - y, optimizer)
with model.use_params(optimizer.averages):
print(model.evaluate(dev_X, dev_y))
def __init__(self, nlp=None, greedyness=0.5, max_dist=50, max_dist_match=500, conll=None, use_no_coref_list=True, debug=False):
self.greedyness = greedyness
self.max_dist = max_dist
self.max_dist_match = max_dist_match
self.debug = debug
if nlp is None:
print("Loading spacy model")
try:
spacy.info('en_core_web_sm')
model = 'en_core_web_sm'
except IOError:
print("No spacy 2 model detected, using spacy1 'en' model")
model = 'en'
nlp = spacy.load(model)
model_path = os.path.join(PACKAGE_DIRECTORY, "weights/conll/" if conll is not None else "weights/")
embed_model_path = os.path.join(PACKAGE_DIRECTORY, "weights/")
print("loading model from", model_path)
self.data = Data(nlp, model_path=embed_model_path, conll=conll, use_no_coref_list=use_no_coref_list, consider_speakers=conll)
self.coref_model = Model(model_path)
self.clusters = {}
self.mention_to_cluster = []
self.mentions_single_scores = {}
self.mentions_single_features = {}
self.mentions_pairs_scores = {}
self.mentions_pairs_features = {}
def train(train_loc, dev_loc, shape, settings):
train_texts1, train_texts2, train_labels = read_snli(train_loc)
dev_texts1, dev_texts2, dev_labels = read_snli(dev_loc)
print("Loading spaCy")
nlp = spacy.load("en_vectors_web_lg")
assert nlp.path is not None
print("Processing texts...")
train_X = create_dataset(nlp, train_texts1, train_texts2, 100, shape[0])
dev_X = create_dataset(nlp, dev_texts1, dev_texts2, 100, shape[0])
print("Compiling network")
model = build_model(get_embeddings(nlp.vocab), shape, settings)
print(settings)
model.fit(
train_X,
train_labels,
validation_data=(dev_X, dev_labels),
epochs=settings["nr_epoch"],
batch_size=settings["batch_size"],
)
if not (nlp.path / "similarity").exists():
(nlp.path / "similarity").mkdir()
print("Saving to", nlp.path / "similarity")
weights = model.get_weights()
# remove the embedding matrix. We can reconstruct it.
del weights[1]
with (nlp.path / "similarity" / "model").open("wb") as file_:
pickle.dump(weights, file_)
with (nlp.path / "similarity" / "config.json").open("w") as file_:
file_.write(model.to_json())
def __init__(self, vdict_path, adict_path, \
batchsize=128, max_length=15, n_ans_vocabulary=1000, mode='train', data_shape=(2048)):
self.batchsize = batchsize
self.d_vocabulary = None
self.batch_index = None
self.batch_len = None
self.rev_adict = None
self.max_length = max_length
self.n_ans_vocabulary = n_ans_vocabulary
self.mode = mode
self.data_shape = data_shape
assert self.mode == 'test'
# load vocabulary
with open(vdict_path,'r') as f:
vdict = json.load(f)
with open(adict_path,'r') as f:
adict = json.load(f)
self.n_vocabulary, self.vdict = len(vdict), vdict
self.n_ans_vocabulary, self.adict = len(adict), adict
self.nlp = spacy.load('en', vectors='en_glove_cc_300_1m_vectors')
self.glove_dict = {} # word -> glove vector
def tokenizeText(sample,parser=spacy.load('en')):
# get the tokens using spaCy
tokens = parser(sample)
# lemmatize
lemmas = []
for tok in tokens:
lemmas.append(tok.lemma_.lower().strip()
if tok.lemma_ != "-PRON-" else tok.lower_)
tokens = lemmas
# stoplist the tokens
tokens = [tok for tok in tokens if tok not in STOPLIST]
# stoplist symbols
tokens = [tok for tok in tokens if tok not in SYMBOLS]
# remove large strings of whitespace
while "" in tokens:
tokens.remove("")
while " " in tokens:
tokens.remove(" ")
while "\n" in tokens:
tokens.remove("\n")
while "\n\n" in tokens:
tokens.remove("\n\n")
return tokens
def main(model_dir=None):
nlp = spacy.load('en', parser=False, entity=False, add_vectors=False)
# v1.1.2 onwards
if nlp.tagger is None:
print('---- WARNING ----')
print('Data directory not found')
print('please run: `python -m spacy.en.download --force all` for better performance')
print('Using feature templates for tagging')
print('-----------------')
nlp.tagger = Tagger(nlp.vocab, features=Tagger.feature_templates)
train_data = [
(
'Who is Shaka Khan?',
[(len('Who is '), len('Who is Shaka Khan'), 'PERSON')]
),
(
'I like London and Berlin.',
[(len('I like '), len('I like London'), 'LOC'),
(len('I like London and '), len('I like London and Berlin'), 'LOC')]
)
]
ner = train_ner(nlp, train_data, ['PERSON', 'LOC'])
doc = nlp.make_doc('Who is Shaka Khan?')
nlp.tagger(doc)
ner(doc)
for word in doc:
print(word.text, word.orth, word.lower, word.tag_, word.ent_type_, word.ent_iob)
if model_dir is not None:
save_model(ner, model_dir)
def test_not_lemmatize_base_forms():
nlp = spacy.load('en', parser=False)
doc = nlp(u"Don't feed the dog")
feed = doc[2]
feed.tag_ = u'VB'
assert feed.text == u'feed'
assert feed.lemma_ == u'feed'
def load_spacy(name, **kwargs):
"""
Load a language-specific spaCy pipeline (collection of data, models, and
resources) for tokenizing, tagging, parsing, etc. text; the most recent
package loaded is cached.
Args:
name (str): standard 2-letter language abbreviation for a language;
currently, spaCy supports English ('en') and German ('de')
**kwargs: keyword arguments passed to :func:`spacy.load`; see the
`spaCy docs <https://spacy.io/docs#english>`_ for details
* via (str): non-default directory from which to load package data
* vocab
* tokenizer
* parser
* tagger
* entity
* matcher
* serializer
* vectors
Returns:
:class:`spacy.<lang>.<Language>`
Raises:
RuntimeError: if package can't be loaded
"""
logger.info('Loading "%s" language spaCy pipeline', name)
return spacy.load(name, **kwargs)
def __init__(self, lemmatize_it=True, stem_it=True, normalize_it=True):
self.lemmatize_it = lemmatize_it
self.stem_it = stem_it
self.normalize_it = normalize_it
self.parser = spacy.load('en')
self.stemmer = gensim.parsing.PorterStemmer()
self.stops = set(nltk.corpus.stopwords.words('english'))
def load_model(model_dir):
model_dir = pathlib.Path(model_dir)
nlp = spacy.load('en', parser=False, entity=False, add_vectors=False)
with (model_dir / 'vocab' / 'strings.json').open('r', encoding='utf8') as file_:
nlp.vocab.strings.load(file_)
nlp.vocab.load_lexemes(model_dir / 'vocab' / 'lexemes.bin')
ner = EntityRecognizer.load(model_dir, nlp.vocab, require=True)
return (nlp, ner)
def __init__(self, encoder_path, bpe_path):
self.nlp = spacy.load('en', disable=['parser', 'tagger', 'ner', 'textcat'])
self.encoder = json.load(open(encoder_path))
self.decoder = {v:k for k,v in self.encoder.items()}
merges = open(bpe_path).read().split('\n')[1:-1]
merges = [tuple(merge.split()) for merge in merges]
self.bpe_ranks = dict(zip(merges, range(len(merges))))
self.cache = {}
print("Top 10 keywords and their Score")
# In[17]:
for i, j, n in zip(scored_key, scored_value, range(10)):
print(i, " \t:\t", j)
# ### Topic Mining from The Keywords and Emotions/Sentiments Associated with it
# Using <b>spaCy</b> for the same
# In[18]:
print("IMPORTING DICTIONARY AND OTHEL NLP TOOLS ")
import spacy
nlp = spacy.load('en_core_web_md')
# <b>Retrieving the generic Domain Specific Buckets</b>
#
# Basic Domain knowledge goes into our model here , identifiying key categories to look into
# In[16]:
# topic_list={'OVERALL':['RECOMMEND', 'EXPERIANCE','VALUE']}
# topic_list.update({'METRO':['METRO', 'METRO RAIL','TRAIN','RAIL','METRO STATION']})
# topic_list.update({'SERVICES':['CUSTOMER CARE', 'COMPLINTS','SECURITY','SMART CARD','PAYMENT','METRO TICKET']})
# topic_list.update({'BUS':['BUS TICKET', 'CONDUCTOR','BUS','PAYMENT','TICKET']})
# topic_list.update({'OTHERS':['OTHERS']})
# topic_list.update({'OPERATIONAL':['SCHEDULE', 'TIMINGS','RELIABILITY']})
# topic_list.update({'LOGISTICS':['BUS STOPS', 'BUS STATIONS','METRO STATIONS','BUS SEATS','SEATS','TICKETING MACHIENE']})
from NER import *
import time
import telebot
import spacy
import os
import json
output_dir = os.getcwd()
MODEL = spacy.load(output_dir)
TOKEN = "881147208:AAGDY-ZvgqonfxS12Dn3GDPubCl4jiJtPJA"
bot = telebot.TeleBot(token=TOKEN)
FLAG = False
FLAG2 = True
diseasess = []
potential_diseas = []
things_to_ask = []
greeting_words = ["hi", "hello", "hey", "helloo", "hellooo",
"g morining", "gmorning", "good morning", "morning",
"good day", "good afternoon", "good evening", "greetings",
"greeting", "good to see you", "its good seeing you",
"how are you", "how're you", "how are you doing",
"how ya doin'", "how ya doin", "how is everything",
"how is everything going", "how's everything going",
"how is you", "how's you", "how are things",
"how're things",
"how is it going", "how's it going", "how's it goin'",
"how's it goin", "how is life been treating you",
import json
import spacy
import numpy as np
nlp = spacy.load("en_vectors_web_lg")
def get_vectors(wordlist):
tokens = nlp(" ".join(str(s) for s in wordlist))
pageVec = np.array([])
for token in tokens:
wordVec = token.vector
wordVec = wordVec.reshape(1, 300)
if pageVec.size == 0:
pageVec = wordVec
else:
pageVec = np.concatenate([pageVec, wordVec], axis=0)
return pageVec
def get_weighted_vectors(wordlist, scores):
tokens = nlp(" ".join(str(s) for s in wordlist))
pageVec = np.array([])
for token, score in zip(tokens, scores):
wordVec = np.multiply(token.vector,np.asarray(score))
wordVec = wordVec.reshape(1,300)
if pageVec.size == 0:
pageVec = wordVec
else:
pageVec = np.concatenate([pageVec,wordVec],axis = 0)
return pageVec
def train_spacy():
TRAIN_DATA = trim_entity_spans(
convert_dataturks_to_spacy("traindata.json"))
# nlp = spacy.blank('en') # create blank Language class at start
nlp = spacy.load("./model") #load the model to add up to it
# create the built-in pipeline components and add them to the pipeline
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'ner' not in nlp.pipe_names:
ner = nlp.create_pipe('ner')
nlp.add_pipe(ner, last=True)
else:
ner = nlp.get_pipe('ner')
# add labels
for _, annotations in TRAIN_DATA:
for ent in annotations.get('entities'):
ner.add_label(ent[2])
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'ner']
with nlp.disable_pipes(*other_pipes): # only train NER
optimizer = nlp.begin_training()
for itn in range(10):
print("Starting iteration " + str(itn))
random.shuffle(TRAIN_DATA)
losses = {}
for text, annotations in TRAIN_DATA:
nlp.update(
[text], # batch of texts
[annotations], # batch of annotations
drop=0.2, # dropout - make it harder to memorise data
sgd=optimizer, # callable to update weights
losses=losses)
print(losses)
# test the model and evaluate it
examples = convert_dataturks_to_spacy("testdata.json")
c = 0
for text, annot in examples:
# f = open("resumes"+str(c)+".txt", "w")
doc_to_test = nlp(text)
# d = {}
# for ent in doc_to_test.ents:
# d[ent.label_] = []
# for ent in doc_to_test.ents:
# d[ent.label_].append(ent.text)
#
# for i in set(d.keys()):
#
# f.write("\n\n")
# f.write(i + ":" + "\n")
# for j in set(d[i]):
# f.write(j.replace('\n', '')+"\n")
d = {}
for ent in doc_to_test.ents:
d[ent.label_] = [0, 0, 0, 0, 0, 0]
for ent in doc_to_test.ents:
doc_gold_text = nlp.make_doc(text)
gold = GoldParse(doc_gold_text, entities=annot.get("entities"))
y_true = [
ent.label_ if ent.label_ in x else 'Not ' + ent.label_
for x in gold.ner
]
y_pred = [
x.ent_type_ if x.ent_type_ == ent.label_ else 'Not ' +
ent.label_ for x in doc_to_test
]
if (d[ent.label_][0] == 0):
print("For Entity " + ent.label_ + "\n")
print(classification_report(y_true, y_pred) + "\n")
(p, r, f,
s) = precision_recall_fscore_support(y_true,
y_pred,
average='weighted')
a = accuracy_score(y_true, y_pred)
d[ent.label_][0] = 1
d[ent.label_][1] += p
d[ent.label_][2] += r
d[ent.label_][3] += f
d[ent.label_][4] += a
d[ent.label_][5] += 1
c += 1
for i in d:
print("\n For Entity " + i + "\n")
print("Accuracy : " + str((d[i][4] / d[i][5]) * 100) + "%")
print("Precision : " + str(d[i][1] / d[i][5]))
print("Recall : " + str(d[i][2] / d[i][5]))
print("F-score : " + str(d[i][3] / d[i][5]))
nlp.to_disk("model")
import numpy as np
from sklearn.datasets import fetch_20newsgroups # import the dataset
import spacy # import spacy for nlp preprocessing
from gensim import corpora, models # import classes for creating bag of words & tf-idf
import re # import regex module
import pandas as pd
from pprint import pprint
import pyLDAvis # import this package for plotting topic model
import pyLDAvis.gensim # import this class to plot based on gensim
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings("ignore",category=DeprecationWarning) # turn off warnings
dataset = fetch_20newsgroups(subset='train', shuffle=True) # load the train data including the labels
nlp = spacy.load('en_core_web_sm') # load the spacy 'en_core_web_sm' model
data = dataset.data # extract only the data
data = [re.sub('\S*@\S*\s?', '', sent) for sent in data] # eliminate e-mail strings from the dataset using the sub method of regex module
texts = [] # create an empty list to store the preprocessed data
allowed_pos = ['ADJ', 'ADV', 'NOUN', 'PROPN', 'VERB'] # only the words having the parts of speech will be included in the dataset
for document in data: # looping over the documents one by one
text = []
doc = nlp(document) # return a spacy document object for processing
for w in doc: # looping over single tokens of the document object
if (not w.is_stop) and (not w.is_punct) and (not w.like_num) and (w.pos_ in allowed_pos): # eliminating stop words, punctuations, numbers and the words whose POS is not included on the list described above using features provided by spacy
text.append(w.lemma_) # take only the lemma of the word
texts.append(text) # append the documents one by one to the 'texts' list created earlier
dictionary = corpora.Dictionary(texts) # create a bag of words model class using the Dictionary class of gensim
corpus = [ dictionary.doc2bow(text) for text in texts] # create the bag of words corpus off the Dictionary class
#turk-eng:
#sentence = "Allaha ve Peygamberine kim inanmamışsa bilsin ki şüphesiz Biz inkarcılar için çılgın alevli cehennemi hazırlamışızdır"
#sentence_eng = "And whoever does not accept faith in Allah and His Noble Messenger – We have indeed kept prepared a blazing fire for disbelievers"
#azer-turkish
#sentence = "O yer üzünü sizin üçün beşik etmiş orada sizin üçün yollar salmış və göydən su endirmişdir"
#sentence_turk = "O ki yeri size beşik yaptı ve onda sizin için yollar açtı gökten bir su indirdi"
#sentence = sentence.split()
#python -m spacy download en
spacy_eng = spacy.load("en")
tokenize_custom = lambda x: x.split()
def tokenize_eng(text):
return [tok.text for tok in spacy_eng.tokenizer(text)]
src_tokenizer = None
trg_tokenizer = None
if set_source_to == "azerbaijani" or set_source_to == "turkish":
src_tokenizer = tokenize_custom
elif set_source_to == "english":
src_tokenizer = tokenize_eng
# TODO aprimorar ementa
ementa = re.search('provido', tempo)
if fim == None or ementa == None:
return None
return sent[ementa.start()+7:fim.start()+14]
def relatorios(n, cases):
final = []
i = 0
while i < n:
if getRelatorio(cases.julgado[i]) != None:
final.append(getRelatorio(cases.julgado[i]))
i += 1
return final
npl = spacy.load("pt_core_news_sm")
relat = relatorios(1000, train_df)
relat[6]
train_df.resultado[range(50)]
opcoes = ['provido', 'improvido', 'parcial', 'não conhecido', 'desconhecido']
i=0
relat_npl = []
while i < 4000:
x = npl(relat[i])
relat_npl.append(x)
i = i+1
import os
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# use spacy small model
# nlp = en_core_web_lg.load()
import spacy
nlp = spacy.load('en_core_web_lg')
from lists_patterns import load_lists,fpath
# dependency markers for subjects
SUBJECTS = {"nsubj", "nsubjpass", "csubj", "csubjpass", "agent", "expl"}
# dependency markers for objects
OBJECTS = {"dobj", "dative", "attr", "oprd","pobj"}
# POS tags that will break adjoining items
BREAKER_POS = {"CCONJ", "VERB"}
# words that are negations
NEGATIONS = {"no", "not", "n't", "never", "none"}
# does dependency set contain any coordinating conjunctions?
def contains_conj(depSet):
return "and" in depSet or "or" in depSet or "nor" in depSet or \
def getNouns(raw_parsed_result):
nlp = spacy.load("en_core_web_sm")
doc = nlp(raw_parsed_result)
nouns = [chunk.text for chunk in doc.noun_chunks]
return nouns
day = int(tokens[0])
year = 2020
else:
day = int(tokens[0])
month = int(tokens[1])
year = 2020
elif len(tokens) == 1:
day = int(tokens[0])
year = 2020
return (day, month, year)
# !python3 -m spacy download en_core_web_lg
import spacy
sp_lg = spacy.load('en_core_web_lg')
import nltk
import re
query_date = {}
for idx, row in enumerate(rows):
date_occurences = [(ent.text.strip(), ent.label_)
for ent in sp_lg(row).ents if ent.label_ == 'DATE']
query_date[row] = []
for date in date_occurences:
try:
date_token = re.split('\s+|/|-|:', date[0])
day, month, year = decode_date(date_token)
query_date[row].append((f'{year}-{month}-{day}'))
row = row.replace(date[0], "", 1)
def load_spacy(language):
model = "en_core_web_lg" if language == "en" else "xx_ent_wiki_sm"
return spacy.load(model)
# stdlib
import re
# third party
import toml
import spacy as sp
import toolz as fp
import pandas as pd
# load spacy model
print(
"[WARN] Loading `en_core_web_md` model from spacy. Might take a few seconds."
)
nlp = sp.load("en_core_web_md")
# load in default_regex strings
with open("optimus/etc/regexes") as handle:
replace = toml.load(handle)
# helper for regex
def default_cleaner(string, regex_dict=replace):
"""
default_cleaner(string, regex_dict={"regex":"replacement"})
A default cleaner for text. The goal for this is to remove
the unnecessary words and other things such as numbers from the
text. The default dictionary for this is in the ``optimus/etc/regexes``
file.
A version of this function that maps across a list exists
def main():
import numpy as np
import pickle
import pandas as pd
import streamlit as st
import tweepy
import pandas as pd
import re
import emoji
import nltk
import datetime
import spacy
from nltk.corpus import stopwords
from nltk.stem.wordnet import WordNetLemmatizer
import string
st.title("Streamlit (Topic_Modelling App)")
html_temp = """ <div style="background-color:tomato;padding:10px">
<h1 style="color:white;text-align:center;">Topic_Modelling</h1>
</div>"""
st.markdown(html_temp, unsafe_allow_html=True)
#Creating search Box for search
Date1 = st.sidebar.date_input(
'start date',
datetime.date.today() - datetime.timedelta(days=7))
Date2 = st.sidebar.date_input('end date', datetime.date.today())
# set variables for keys and tokens to access the Twitter API
mykeys = open('API Twitter.txt', 'r').read().splitlines()
api_key = mykeys[0]
api_key_secret = mykeys[1]
access_token = mykeys[2]
access_token_secret = mykeys[3]
auth = tweepy.OAuthHandler(consumer_key=api_key,
consumer_secret=api_key_secret)
auth.set_access_token(access_token, access_token_secret)
api = tweepy.API(auth, wait_on_rate_limit=True)
#Featching the data from twitter
search_words = "news"
date_since = Date1
data_until = Date2
tweets = tweepy.Cursor(api.search,
q=search_words,
lang="en",
tweet_mode='extended',
since=date_since,
until=data_until,
result_type="recent").items(300)
# Collect tweets
tweets = tweepy.Cursor(api.search,
q=search_words,
lang="en",
since=date_since).items(300)
# Iterate and print tweets
s = []
for tweet in tweets:
s.append(tweet.text)
print(s)
df = pd.DataFrame({'tweet': s})
import nltk
words = set(nltk.corpus.words.words())
tweet = np.array(df.tweet)
cleaned_tweet = []
for i in df.tweet:
no_punc_text = i.translate(str.maketrans('', '', string.punctuation))
no_punc_text = re.sub("(RT)?(ht)?", "",
no_punc_text) # to remove RT and ht word
no_punc_text1 = re.sub(
"[\W\d]", " ",
no_punc_text) #to remove not word character and numbers
no_punc_text2 = re.sub(
"[^a-zA-Z]", " ",
no_punc_text1) #to remove forien language word character
no_punc_text2=" ".join(w for w in nltk.wordpunct_tokenize(no_punc_text2) \
if w.lower() in words or not w.isalpha())
cleaned_tweet.append(no_punc_text2)
df['cleaned_tweet'] = cleaned_tweet
df1 = df.copy()
corpus = df1.cleaned_tweet.unique()
# import vectorizers
from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer
# import numpy for matrix operation
import numpy as np
# import LDA from sklearn
from sklearn.decomposition import LatentDirichletAllocation
#nltk.download('wordnet')
# Lemmatize with POS Tag
from nltk.corpus import wordnet
import nltk
#nltk.download('averaged_perceptron_tagger')
def get_wordnet_pos(word):
"""Map POS tag to first character lemmatize() accepts"""
tag = nltk.pos_tag([word])[0][1][0].upper()
tag_dict = {
"J": wordnet.ADJ,
"N": wordnet.NOUN,
"V": wordnet.VERB,
"R": wordnet.ADV
}
return tag_dict.get(tag, wordnet.NOUN)
# Apply Preprocessing on the Corpus
# stop loss words
stop = set(stopwords.words('english'))
stop.update([
"new", "news", 'via', 'take', 'first', 'one', 'say', 'time', 'big',
'see', 'come', 'good', 'another', 'today', 'make', 'get', 'great',
'could', 'like', 'make', 'set', 'end', 'dont'
])
# punctuation
exclude = set(string.punctuation)
# lemmatization
lemma = WordNetLemmatizer()
# One function for all the steps:
def clean(doc):
# convert text into lower case + split into words
stop_free = " ".join([i for i in doc.lower().split() if i not in stop])
# remove any stop words present
punc_free = ''.join(ch for ch in stop_free if ch not in exclude)
# remove punctuations + normalize the text
normalized = " ".join(
lemma.lemmatize(word, get_wordnet_pos(word))
for word in punc_free.split())
return normalized
# clean data stored in a new list
clean_corpus = [clean(doc).split() for doc in corpus]
corpus1 = []
for i in clean_corpus:
doc = []
#j=i.split()
for z in i:
#print(len(z))
if len(z) > 2:
doc.append(z)
#print(doc)
doc = " ".join(doc)
doc1 = doc.split()
#print(doc1)
corpus1.append(doc1)
clean_corpus = corpus1
abc = [] #to create single list
for i in clean_corpus:
abc.append(' '.join(i))
abc2 = " ".join(abc)
nlp = spacy.load('en_core_web_sm')
one_block = abc2
doc_block = nlp(one_block)
#collecting 'PROPN','X','NOUN','ADJ' words
final_corpus = [
token.text for token in doc_block
if token.pos_ in ('PROPN', 'X', 'NOUN', 'ADJ')
]
imp_words = set(final_corpus)
# to remove the meaningless words
#doc=[]
corpus1 = []
for i in clean_corpus:
doc = []
#j=i.split()
for z in i:
#print(len(z))
if z in imp_words:
doc.append(z)
#print(doc)
doc = " ".join(doc)
doc1 = doc.split()
#print(doc1)
corpus1.append(doc1)
new_clean_corpus = corpus1
# Converting text into numerical representation
tf_idf_vectorizer = TfidfVectorizer(tokenizer=lambda doc: doc,
lowercase=False)
# Converting text into numerical representation
cv_vectorizer = CountVectorizer(tokenizer=lambda doc: doc, lowercase=False)
# Array from TF-IDF Vectorizer
tf_idf_arr = tf_idf_vectorizer.fit_transform(clean_corpus)
# Array from Count Vectorizer
cv_arr = cv_vectorizer.fit_transform(clean_corpus)
# Materialize the sparse data
data_dense = cv_arr.todense()
# Compute Sparsicity = Percentage of Non-Zero cells
print("Sparsicity: ", ((data_dense > 0).sum() / data_dense.size) * 100,
"%")
# Creating vocabulary array which will represent all the corpus
vocab_tf_idf = tf_idf_vectorizer.get_feature_names()
# Creating vocabulary array which will represent all the corpus
vocab_cv = cv_vectorizer.get_feature_names()
result = ""
if st.button("Search"):
result = Predict_Topics(cv_arr, vocab_tf_idf)
st.success(st.write(result))
@author: nirmalenduprakash
Identifies important entities in a document and discovers sentiment towards these
entities
"""
import spacy
import numpy as np
import pandas as pd
#!pip install git+htt!ps://github.com/huggingface/neuralcoref.git
import nltk
#nltk.download('punkt')
import neuralcoref
from wordcloud import WordCloud, STOPWORDS
nlp = spacy.load("en_core_web_sm")
stopwords = set(STOPWORDS)
pronouns = ['i', 'he', 'she', 'you', 'we', 'they', 'them', 'it', 'his', 'who']
def read_file(filepath):
df = pd.read_table(filepath, sep=' ')
if ('POS_TAGS' in df.columns):
df.drop(['POS_TAGS'], inplace=True, axis=1)
return df
df_ug = read_file(
'/Users/nirmalenduprakash/Documents/Project/NLP/Sentiment Mining/IBM_Debater_(R)_SC_COLING_2018/LEXICON_UG.txt'
)
df_bg = read_file(
def __init__(self):
path = os.path.dirname(os.path.realpath(__file__))
self.df = pd.read_csv(os.path.join(path, "../data/countries.csv"))
self.utils = nlpUtils()
self.nlp = spacy.load("en_core_web_sm")
self.nationality_matcher = Matcher(self.nlp.vocab)
nat_pattern = list()
nat_pattern.append([{
'LEMMA': 'be'
}, {
'POS': 'DET'
}, {
'ENT_TYPE': {
"IN": ["GPE", "NORP", "LANGUAGE"]
},
'OP': "*"
}, {
'POS': {
"IN": ["NOUN", "PROPN", "PUNCT", "ADJ", "SYM"]
},
"OP": "*"
}, {
'POS': {
"IN": ["NOUN", "PROPN", "ADJ"]
},
"OP": "+"
}])
nat_pattern.append([{
'LEMMA': 'be'
}, {
'POS': 'DET'
}, {
'ENT_TYPE': {
"IN": ["GPE", "NORP", "LANGUAGE"]
},
'OP': "*"
}, {
"DEP": {
"IN": ["punct", "compound", "amod", "nmod"]
},
"OP": "*"
}, {
'POS': 'NOUN'
}, {
"POS": {
"IN": ["PUNCT", "NOUN", "ADJ", "PROPN"]
},
"OP": "*"
}, {
'ORTH': 'and'
}, {
'POS': {
"IN": ["NOUN", "PROPN", "PUNCT", "ADJ"]
},
"OP": "*"
}, {
'POS': {
"IN": ["NOUN", "PROPN", "ADJ"]
},
"OP": "+"
}])
self.nationality_matcher.add("nationality", nat_pattern)
self.influence_matcher = Matcher(self.nlp.vocab)
influence1 = list()
influence1.append([{
'LEMMA': {
"IN": ["inspire", "influence"]
},
"POS": 'VERB'
}, {
'ORTH': 'by'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence1", influence1)
influence2 = list()
influence2.append([{
'LEMMA': {
"IN": ["cite", "refer", "list", "mention", "credit", "claim"]
},
"POS": 'VERB'
}, {
"OP": "*"
}, {
'LEMMA': {
"IN": ["as", "among"]
}
}, {
"OP": "*"
}, {
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
"OP": "*"
}])
influence2.append([{
'LEMMA': {
"IN": ["cite", "refer", "list", "mention", "credit", "claim"]
},
"POS": 'VERB'
}, {
"OP": "*"
}, {
'LEMMA': 'be'
}, {
"OP": "*"
}, {
'LEMMA': 'influence',
"POS": 'NOUN'
}])
self.influence_matcher.add("influence2", influence2)
influence3 = list()
influence3.append([{
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
'ORTH': 'include',
"POS": 'VERB'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence3", influence3)
influence4 = list()
influence4.append([{
'ORTH': 'influences',
"POS": 'NOUN'
}, {
'ORTH': 'cited'
}, {
'ORTH': 'by'
}, {
"OP": "*"
}, {
'ORTH': 'include',
"POS": 'VERB'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence4", influence4)
influence5 = list()
influence5.append([{
'LEMMA': 'cite',
"POS": 'VERB'
}, {
'ORTH': ','
}, {
"ORTH": "as"
}, {
"OP": "*"
}, {
'ORTH': 'influences',
"POS": 'NOUN'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence5", influence5)
influence6 = list()
influence6.append([{
'LEMMA': 'state',
"POS": 'VERB'
}, {
"OP": "*"
}, {
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
'LEMMA': 'be'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence6", influence6)
influence7 = list()
influence7.append([{
'ORTH': 'influences',
"POS": 'NOUN'
}, {
"ORTH": "?"
}, {
"ORTH": "such"
}, {
"ORTH": "as"
}, {
"OP": "*"
}])
self.influence_matcher.add("influence7", influence7)
influence8 = list()
influence8.append([{
'LEMMA': {
"IN": ["cite", "name"]
},
"POS": "VERB"
}, {
"OP": "*"
}, {
"ORTH": "as"
}, {
"ORTH": "one"
}, {
"ORTH": "of"
}, {
"OP": "*"
}, {
"ORTH": "'s"
}, {
'LEMMA': 'influence',
"POS": 'NOUN'
}])
self.influence_matcher.add("influence8", influence8)
influence9 = list()
influence9.append([{
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
"ORTH": "including"
}, {
"OP": "*"
}])
self.influence_matcher.add("influence9", influence9)
influence10 = list()
influence10.append([{
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
"OP": "*"
}, {
"ORTH": "from"
}, {
"OP": "*"
}])
self.influence_matcher.add("influence10", influence10)
influence11 = list()
influence11.append([{
'ORTH': 'citing',
"POS": 'VERB'
}, {
"ORTH": "as"
}, {
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence11", influence11)
influence12 = list()
influence12.append([{
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
'LEMMA': 'be'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence12", influence12)
influence13 = list()
influence13.append([{
'LEMMA': 'influence',
"POS": 'NOUN'
}, {
'ORTH': 'of'
}, {
"OP": "*"
}])
self.influence_matcher.add("influence13", influence13)
influence14 = list()
influence14.append([{
'LEMMA': 'inspiration',
"POS": 'NOUN'
}, {
'ORTH': {
"IN": ["from", "include"]
}
}, {
"OP": "*"
}])
influence14.append([{
'LEMMA': 'cite',
"POS": 'VERB'
}, {
"OP": "*"
}, {
"ORTH": "as"
}, {
'LEMMA': 'inspiration',
"POS": 'NOUN'
}])
self.influence_matcher.add("influence14", influence14)
self.mappa = dict()
self.mappa[self.nlp.vocab.strings["influence1"]] = "influence1"
self.mappa[self.nlp.vocab.strings["influence2"]] = "influence2"
self.mappa[self.nlp.vocab.strings["influence3"]] = "influence3"
self.mappa[self.nlp.vocab.strings["influence4"]] = "influence4"
self.mappa[self.nlp.vocab.strings["influence5"]] = "influence5"
self.mappa[self.nlp.vocab.strings["influence6"]] = "influence6"
self.mappa[self.nlp.vocab.strings["influence7"]] = "influence7"
self.mappa[self.nlp.vocab.strings["influence8"]] = "influence8"
self.mappa[self.nlp.vocab.strings["influence9"]] = "influence9"
self.mappa[self.nlp.vocab.strings["influence10"]] = "influence10"
self.mappa[self.nlp.vocab.strings["influence11"]] = "influence11"
self.mappa[self.nlp.vocab.strings["influence12"]] = "influence12"
self.mappa[self.nlp.vocab.strings["influence13"]] = "influence13"
self.mappa[self.nlp.vocab.strings["influence14"]] = "influence14"
def extract(lang, tweet, params) :
nlp = spacy.load(lang)
nlp.max_length = SPACY_MAX_LENGTH
tweet_dict = OrderedDict()
if lang == 'en':
pos_dict = {
'SPACE': 0,
'NOUN': 0,
'PART': 0,
'PRON': 0,
'INTJ': 0,
'SYM': 0,
'ADJ': 0,
'CCONJ': 0,
'PUNCT': 0,
'X': 0,
'VERB': 0,
'ADP': 0,
'ADV': 0,
'PROPN': 0,
'NUM': 0,
'DET': 0,
'AUX': 0,
'SCONJ' : 0
}
tense_dict = {
'Past': 0,
'Pres': 0
}
n_stopwords = 0
for word in nlp(tweet['text']):
# n_pos features
pos = word.pos_
pos_dict[pos] = pos_dict[pos] + 1
# n_stopwords
if word.is_stop:
n_stopwords += 1
# n_tense features
tag = word.tag_
if tag in params['tense']['present']:
tense_dict['Pres'] = tense_dict['Pres'] + 1
elif tag in params['tense']['past']:
tense_dict['Past'] = tense_dict['Past'] + 1
for feature, params in params.items():
if feature == 'tense':
tweet_dict['n_tense_past'] = tense_dict['Past']
tweet_dict['n_tense_pres'] = tense_dict['Pres']
elif feature == 'pos_counts':
tweet_dict['n_pos_space'] = pos_dict['SPACE']
tweet_dict['n_pos_noun'] = pos_dict['NOUN']
tweet_dict['n_pos_par'] = pos_dict['PART']
tweet_dict['n_pos_pron'] = pos_dict['PRON']
tweet_dict['n_pos_intj'] = pos_dict['INTJ']
tweet_dict['n_pos_sym'] = pos_dict['SYM']
tweet_dict['n_pos_adj'] = pos_dict['ADJ']
tweet_dict['n_pos_conj'] = pos_dict['CCONJ']
tweet_dict['n_pos_punct'] = pos_dict['PUNCT']
tweet_dict['n_pos_x'] = pos_dict['X']
tweet_dict['n_pos_verb'] = pos_dict['VERB']
tweet_dict['n_pos_adp'] = pos_dict['ADP']
tweet_dict['n_pos_adv'] = pos_dict['ADV']
tweet_dict['n_pos_propn'] = pos_dict['PROPN']
tweet_dict['n_pos_num'] = pos_dict['NUM']
tweet_dict['n_pos_det'] = pos_dict['DET']
elif feature == 'word_counts':
tweet_dict['n_words'] = len(tweet['text'].split())
elif feature == 'stopword_counts':
tweet_dict['n_stopwords'] = n_stopwords
return tweet_dict
import pytest
import spacy
import string
import numpy as np
from alibi.explainers import AnchorText
from alibi.explainers.anchor_text import Neighbors
from alibi.explainers.tests.utils import get_dataset
from alibi.explainers.tests.utils import predict_fcn
from alibi.utils.download import spacy_model
# load spaCy model
model = 'en_core_web_md'
spacy_model(model=model)
nlp = spacy.load(model)
def find_punctuation(text: str) -> int:
"""
Returns nb of punctuation marks in a string.
"""
punctuation = set([s for s in string.punctuation])
tokens = set(text.split())
return len(tokens & punctuation)
@pytest.mark.parametrize('lr_classifier', ((get_dataset('movie_sentiment')), ),
indirect=True)
def __init__(self, model="en_core_web_sm"):
if model is None:
model = "en_core_web_sm"
self._nlp = spacy.load(model)
def __init__(self, path='data/input/'):
self.nlp = spacy.load('en', disable=['ner', 'textcat'])
self.pi = ProcessInput()
self.pub_df = self.pi.load_publication_input(path=path)
def __init__(self):
self.nlp = spacy.load("en_core_web_sm")
def __init__(self):
self.spacynlp = spacy.load('en_core_web_lg')
def default_nlp():
return spacy.load('en_core_web_sm')
def howParse(sentence, answer):
''' Parser for questions that begin with the adverbial modifier how'''
arg1 = []
arg2 = []
rel = []
argument = False
relTrue = False
Object = False
poss = False
objTrue = False
if sentence[0].lower_ != "how":
return None
if sentence[1].lower_ == "much" or sentence[1].lower_ == "many":
arg1 = answer
argument = False
for child in sentence:
if "subj" in child.dep_:
_, arg2 = descendants(sentence, child, True)
if "obj" in child.dep_:
Object = True
_, rel = descendants(sentence, sentence.root, True)
rel = [token for token in rel if not token in arg2]
stopwords = ['much', 'many']
arg2 = [word for word in arg2 if word.lower_ not in stopwords]
rel = [token for token in rel if token.lower_ not in stopwords]
rel = [
child for child in rel
if "aux" not in child.dep_ or child.lower_ == "to"
]
if len(arg2) != 0:
if sentence[1].lower_ == "much" and arg2[0].dep_ != "prep":
arg2.insert(0, "in")
if sentence[1].lower_ == "many" and Object == False:
rel.insert(0, "number")
elif sentence[1].lower_ == "did" or sentence[1].lower_ == "is":
for child in sentence:
if "obj" in child.dep_:
_, rel = descendants(sentence, child, True, sentence.root)
if "subj" in child.dep_:
_, arg1 = descendants(sentence, child, True)
arg2.append(answer)
nlp = spacy.load('en')
answerDep = nlp(answer)
if sentence[1].lower_ != "is":
for token in answerDep:
if token.dep_ == "poss":
arg2.insert(0, "with")
poss = True
if poss == False:
arg2.insert(0, "by")
else:
arg2.insert(0, "as")
argument = True
elif "comp" in sentence[1].dep_ or sentence[1].dep_ == "advmod":
for child in sentence:
if "obj" in child.dep_:
objTrue = True
if objTrue == True and sentence[1].dep_ == "advmod":
for children in sentence:
if "obj" in children.dep_:
_, arg1 = descendants(sentence, children, True)
rel = sentence.root.lower_
arg2.append(answer)
for advmod in sentence:
if advmod.dep_ == "advmod":
arg2.append(advmod.lower_)
if objTrue == True and "comp" in sentence[1].dep_:
for token in sentence:
if "subj" in token.dep_:
_, arg1 = descendants(sentence, token, True)
break
rel = sentence.root.lower_
arg2.append(answer)
for comp in sentence:
if "comp" in comp.dep_:
arg2.append(comp.lower_)
else:
return None
print("How parse")
arg1, rel, arg2 = extractHelper(arg1, rel, arg2)
return Extract(arg1=arg1, rel=rel, arg2=arg2)
import random
import math
import time
import collections
from collections import Counter
import math
import numpy as np
import subprocess
# from google.colab import drive
# drive.mount('/content/drive')
#
# !python -m spacy download en
# !python -m spacy download fr
spacy_en = spacy.load('en')
spacy_de = spacy.load('fr')
# preparing data
SEED = 1234
random.seed(SEED)
np.random.seed(SEED)
torch.manual_seed(SEED)
torch.cuda.manual_seed(SEED)
torch.backends.cudnn.deterministic = True
# spacy_de = spacy.load('de')
def generate_features(df):
"""Generate the following features:
'days_yelping',
'word_count',
'unique_words_perc',
'no_adjectives',
'perc_unique_verbs',
'perc_unique_adjectives',
'norm_of_wordvecs',
'adj_wordvecs'
'verb_wordvecs',
'adj_maj_sent__negative',
'adj_maj_sent__objective',
'adj_maj_sent__positive',
'adj_maj_sent__no_sentiment_assigned',
'verb_maj_sent_negative',
'verb_maj_sent_objective',
'verb_maj_sent_positive',
'verb_maj_sent_no_sentiment_assigned',
'useful_review_compliments_received_for_other_reviews'
"""
print('----- Feature Generation -----')
feature_data = df[[
'useful_01', 'text', 'stars', 'average_stars', 'compliment_more',
'compliment_hot', 'compliment_photos', 'compliment_writer',
'compliment_plain', 'fans', 'review_count', 'yelping_since', 'user_id',
'useful_review'
]]
feature_data = feature_data.merge(pd.DataFrame(feature_data.groupby(feature_data['user_id']).useful_review.sum()), \
how='left', on='user_id',
suffixes=('', '_compliments_received_for_other_reviews'))
feature_data['useful_review_compliments_received_for_other_reviews'] = \
feature_data['useful_review_compliments_received_for_other_reviews'] - feature_data['useful_review']
nlp = spacy.load('en_core_web_lg', disable=['ner', "parser"])
# define for each token the attribute _.ignore:
ignore_getter = lambda token: (
token.is_stop or # remove stop words
token.lower_ in STOP_WORDS
or # ignore stop words independent of their case
# token.like_num or # ignore stuff that looks like a number
# token.is_digit or #ignore tokens consisting of digits
token.is_punct or # ignore punctuation tokens
token.is_left_punct or # ignore left punctuation
token.is_right_punct or # ignore right punctuation
token.is_space or # ignore tokens consisting of spaces
token.is_bracket or # ignore brackets
token.is_quote or # ignore quotation marks
not token.is_alpha) # ignore everything that is not only letters
# (this might be too strict, but without it special characters
# like +$% etc will stay). With it, however, most of the previous
# stuff is not needed..
Token.set_extension('ignore', getter=ignore_getter,
force=True) # add the _.ignore attribute
def get_days_yelping(input_timestamp):
days_yelp = datetime.now().date() - datetime.strptime(
input_timestamp.split()[0], '%Y-%m-%d').date()
return days_yelp.days
def get_tokens(nlp_text):
return [
token for token in nlp_text
if not (token._.ignore or not (len(token) > 1))
]
def get_unique__perc(token_list):
if not len(token_list) == 0:
return len(set([str(tok) for tok in token_list])) / len(token_list)
else:
return 0
def get_word_vectors(token_list):
if not len(token_list) == 0:
word_vectors = [token.vector for token in token_list]
return np.linalg.norm(sum(word_vectors) / len(token_list))
else:
return 0
def get_sentiment(token_list, nlp_review):
"""Get sentiment using SentiWordNet"""
# WordNet (and thus SentiWordNet) use different POS tags than spacy output
# here we transform from spacy to WordNet
wordNet_pos_dict = {"NOUN": "n", "VERB": "v", "ADJ": "a", "ADV": "r"}
def posTag_to_wordNetTag(tag):
if tag in wordNet_pos_dict:
return wordNet_pos_dict[tag]
else:
return None
# to get the semantic score of a word from SentiWordNet we first need to find this word in WordNet
# This process is called Word Sense Disambiguation and we use the simple lesk algorithm for that
def get_semantic_score_with_context(token, nlp_review):
word = token.lower_ # get lowercased token text
position = token.idx # get position of word in document
pos = posTag_to_wordNetTag(
token.pos_
) # get POS of token, for better word sense disambiguation
# define how many tokens around the token of interest we look at
num_surrounding_words = 10
# careful if there are less then num_surrounding_words before our token or after our token
leftmost_word_idx = max(0, position - num_surrounding_words)
rightmostword_idx = min(len(nlp_review),
position + num_surrounding_words)
surrounding_text = nlp_review[
leftmost_word_idx:rightmostword_idx].text
# determine word with the closest sense in WordNet
# print(word,"....",surrounding_text,pos)
try:
word_with_closest_sense = simple_lesk(surrounding_text,
word,
pos=pos)
except:
word_with_closest_sense = simple_lesk(surrounding_text, word)
# print(word,pos,word_with_closest_sense)
# find the sentiment score to the word we found in wordnet
if word_with_closest_sense:
sentiword = swn.senti_synset(word_with_closest_sense.name())
sent_scores = {
"objective": sentiword.obj_score(),
"positive": sentiword.pos_score(),
"negative": sentiword.neg_score()
}
sentiment = max(sent_scores, key=sent_scores.get)
return sentiment
else:
return 'no_sentiment_assigned'
if not len(token_list) == 0:
sentiments = []
for token in token_list:
sentiments.append(
get_semantic_score_with_context(token, nlp_review))
counts = Counter(sentiments)
return max(counts, key=counts.get)
else:
return 'no_sentiment_assigned'
feature_data['days_yelping'] = feature_data.yelping_since.swifter.apply(
get_days_yelping)
del feature_data['yelping_since']
feature_data['text_processed'] = feature_data.text.swifter.apply(
lambda x: nlp(x))
feature_data['tokens'] = feature_data.text_processed.swifter.apply(
lambda x: get_tokens(x))
feature_data['word_count'] = feature_data.tokens.swifter.apply(
lambda x: len(x))
feature_data['unique_words_perc'] = feature_data.tokens.swifter.apply(
get_unique__perc)
feature_data['adjectives'] = feature_data.tokens.swifter.apply(
lambda x: [token for token in x if token.pos_ == 'ADJ'])
feature_data['no_adjectives'] = feature_data.adjectives.swifter.apply(
lambda x: len(x))
feature_data['verbs'] = feature_data.tokens.swifter.apply(
lambda x: [token for token in x if token.pos_ == 'VERB'])
feature_data['perc_unique_verbs'] = feature_data.verbs.swifter.apply(
get_unique__perc)
feature_data[
'perc_unique_adjectives'] = feature_data.adjectives.swifter.apply(
get_unique__perc)
feature_data['norm_of_wordvecs'] = feature_data.tokens.swifter.apply(
get_word_vectors)
feature_data['adj_wordvecs'] = feature_data.adjectives.swifter.apply(
get_word_vectors)
feature_data['verb_wordvecs'] = feature_data.verbs.swifter.apply(
get_word_vectors)
feature_data['adj_maj_sent'] = feature_data.swifter.apply(
lambda x: get_sentiment(x.adjectives, x.text_processed), axis=1)
feature_data['verb_maj_sent'] = feature_data.swifter.apply(
lambda x: get_sentiment(x.verbs, x.text_processed), axis=1)
feature_data = feature_data.merge(pd.get_dummies(feature_data[['adj_maj_sent', 'verb_maj_sent']], prefix = ['adj_maj_sent_', 'verb_maj_sent']), \
left_index = True, right_index = True)
# feather.write_dataframe(feature_data, os.getcwd() + 'feature_data.feather')
return feature_data
in American history, including the Declaration of Independence, the Emancipation Proclamation,
and the United States Constitution. Early in his speech, King alludes to Abraham Lincoln's
Gettysburg Address by saying "Five score years ago ..." In reference to the abolition of
slavery articulated in the Emancipation Proclamation, King says: "It came as a joyous
daybreak to end the long night of their captivity." Anaphora (i.e., the repetition of a
phrase at the beginning of sentences) is employed throughout the speech. Early in his speech,
King urges his audience to seize the moment; "Now is the time" is repeated three times in the sixth
paragraph. The most widely cited example of anaphora is found in the often quoted phrase
"I have a dream", which is repeated eight times as King paints a picture of an integrated and
unified America for his audience. Other occasions include "One hundred years later", "We can
never be satisfied", "With this faith", "Let freedom ring", and "free at last". King was the
sixteenth out of eighteen people to speak that day, according to the official program.'''
stopwords = list(STOP_WORDS)
nlp = spacy.load('en_core_web_sm')
doc = nlp(text)
tokens = [token.text for token in doc ]
punctuation = punctuation + '\n' + ' ' + ' ' + '...' + '\n '
word_frequency = {}
for word in doc:
if word.text.lower() not in stopwords:
if word.text.lower() not in punctuation:
if word.text not in word_frequency.keys():
word_frequency[word.text] = 1
else:
word_frequency[word.text] += 1
import spacy
nlp = spacy.load("en_core_web_md")
# doc = nlp("Calderon-Zygmund operators are objects that are largely responsible for our understanding of a number of physical phenomena, from heat transfer to turbulence")
doc = nlp(
"With this award, the Chemical Structure, Dynamics and Mechanisms (CSDM-A) Program of the Division of Chemistry is funding Professor Istvan Z. Kiss and his research group at Saint Louis University to study pattern formation of electrochemical reactions"
)
for ent in doc.ents:
print(ent.text, ent.start_char, ent.end_char, ent.label_)
def __init__(self):
self.query = ""
self.nlp = spacy.load('en_core_web_sm')
# Make it work for Python 2+3 and with Unicode
try:
to_unicode = unicode
except NameError:
to_unicode = str
# Import necessary packages
import spacy
nlp = spacy.load('en')
import subprocess
from nltk.stem.wordnet import WordNetLemmatizer
lemmatizer = WordNetLemmatizer()
from nltk.corpus import wordnet as wn
import re
pattern = re.compile(r'(<IN>)*(<DT>)*(<JJ>)*(<NN>|<NNS>|<NNP>)+')
w_words = ['when', 'who', 'what', 'why', 'how', 'where']
import json
from collections import defaultdict
import pandas as pd
from flask import Flask, render_template, request, redirect
import re
import io
from fuzzywuzzy import process, fuzz
## Some import word lists
aux_verb = [
'be', 'can', 'could', 'dare', 'do', 'have', 'may', 'might', 'must', 'need',
'ought', 'shall', 'should', 'will', 'would'
]
wh_words = ['when', 'why', 'how', 'what', 'where', 'who']
else:
# Uppercase followed by lowercase
chars = [c.lower() if i % 2 else c.upper()
for i, c in enumerate(text)]
# Create augmented training example
example_dict = example.to_dict()
doc = nlp.make_doc("".join(chars))
example_dict["token_annotation"]["ORTH"] = [t.text for t in doc]
# Original example followed by augmented example
# yield example
yield example.from_dict(doc, example_dict)
return augment
dir(example.)
nlp = spacy.load("da_core_news_sm")
doc = nlp("Mit navn er Kenneth og Malte og Jakob og Kenneth.")
ent_dict = {"PER": get_names()}
example = Example(doc, doc)
lc_augmenter = create_augmenter_sponge(randomize=False)
res = next(lc_augmenter(nlp, example))
res.text
res
augment = create_augmenter(ent_dict, prob=1)
res
res = next(augment(nlp, res))
res.text
res
dir(example)