def sent_anly_prediction():
if request.method == 'POST':
text = request.form['text']
Sentiment = ''
max_review_length = 500
print(text + "\n")
start = time.time()
Util = util()
end2 = time.time()
### total time taken
print(f"Runtime of util is {end2 - start}")
t = [Util.text_util_final(text)]
end3 = time.time()
### total time taken
print(f"Runtime of text_util_final is {end3 - start}")
dict_tfidf = Dict_Tfidf(PATH)
end4 = time.time()
### total time taken
print(f"Runtime of Dict_Tfidf is {end4 - start}")
vectorizer = dict_tfidf.create_dict_tfidf()
end5 = time.time()
### total time taken
print(f"Runtime of create_dict_tdidf is {end5 - start}")
vector_tfidf = vectorizer.transform(t)
end6 = time.time()
### total time taken
print(f"Runtime of transform is {end6 - start}")
label = model.predict(vector_tfidf)
class1 = str(label[0])
end7 = time.time()
### total time taken
print(f"Runtime of predict is {end7 - start}")
end = time.time()
### total time taken
print(f"Runtime of the program is {end - start}")
print(class1 + "\n")
###Compare label if == "1" => negative and set sad emoji else positive and set smiling emoji
if class1 == "1":
sentiment = 'Negative'
img_filename = os.path.join(app.config['UPLOAD_FOLDER'],
'Sad_Emoji.png')
else:
sentiment = 'Positive'
img_filename = os.path.join(app.config['UPLOAD_FOLDER'],
'Smiling_Emoji.png')
print(sentiment)
#####write output contains class and content (input data) into csv file
with open("output/output_sentiment_analysis.csv",
'a',
newline='',
encoding="utf8",
errors="ignore") as out:
filenames = ['class', 'content']
writer = csv.DictWriter(out, filenames)
writer.writeheader()
writer.writerow({'class': sentiment, 'content': text.strip("\n")})
return render_template('home.html',
text=text,
sentiment=sentiment,
image=img_filename)
def classify():
ap = argparse.ArgumentParser()
ap.add_argument("-t", "--text", help="nhap doan text!!!")
args = vars(ap.parse_args())
text = args["text"]
Util = util()
text = Util.text_util_final(text)
if (text.find("nhưng", 0, len(text)) > -1
or text.find("nhung", 0, len(text)) > -1):
exception.exist_nhung(text)
exit(0)
if ((text.find("được mỗi", 0,
len(text) > -1))
or (text.find("được cái", 0, len(text)) > -1)):
print("Day la binh luan tieu cuc!")
exit(0)
text = [Util.text_util_final(text)]
dict_tfidf = Dict_Tfidf(PATH)
vectorizer = dict_tfidf.create_dict_tfidf()
vector_tfidf = vectorizer.transform(text)
model = joblib.load('./models/best_model.pkl')
label = model.predict(vector_tfidf)
if (label[0] == 1):
print("Day la binh luan tieu cuc!")
else:
print("Day la binh luan tich cuc!")
def create_dict_tfidf(self):
start8 = time.time()
ds = DataSource()
dict_data = pd.DataFrame(ds.load_data(self.path)).review
dict_data = dict_data.tolist()
Util = util()
A = []
for i in range(len(dict_data)):
text = dict_data[i]
text = Util.text_util_final(text)
A.append(text)
end8 = time.time()
# total time taken
print(f"Runtime of preprocess is {end8 - start8}")
vectorizer = TfidfVectorizer(max_features=100000, ngram_range=(1, 3))
end9 = time.time()
# total time take
print(f"Runtime of tfidfvectorizer is {end9 - start8}")
vectorizer.fit(A)
end10 = time.time()
# total time taken
print(f"Runtime of fit is {end10 - start8}")
return vectorizer
def application(environ, start_response):
# Set up initial parameters and utility instances
request = webob.Request(environ)
params = json.loads(request.body) if request.body else {
'username': '******'
} # Default to USDI
utl = utils.util(environ, start_response)
try:
# Authorize the Twitter API `auth` object using OAuth tokens
auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_token, access_secret)
# Set the `auth` object for tweepy's API
twitter_api = tweepy.API(auth)
# Perform a search
## tweets = twitter_api.search(q='from%3A{usernm}%20OR%20%23{usernm}%20OR%20%40{usernm}'
## .format(usernm=params['username'])) # from:nasa OR #nasa OR @nasa
tweets = twitter_api.search(q='from%3A{usernm}'.format(
usernm=params['username'])) # from:interior
# Iterate over first 6 tweets and add information to response payload
payload = []
for tweet in tweets[:6]:
tweet_props = {
"text": tweet.text, # Tweet contents
"time": tweet.created_at.strftime(
"%-m/%-d/%y at %I:%M %p"), # Datetime formatted
"num_words": len(tweet.text.split(
)) # Number of words in content split by whitespace
}
payload.append(tweet_props)
# Log a successful result
res = True
except Exception as e:
res = False
err_texts = []
excptn, err_msg, tb = sys.exc_info()
for file_name, line_no, func_name, text in traceback.extract_tb(tb):
err_text = '{} (LINE {}) in {}() -> `{}`'.format(
os.path.basename(file_name), line_no, func_name, text)
err_texts.append(err_text)
err_texts.insert(0, '{} {}'.format(e.message, type(e)))
tb_error = '\n'.join(err_texts)
finally:
data = collections.OrderedDict()
data['result'] = str(res)
data[
'traceback'] = tb_error if not res else None # Comment out in production
if res:
data['payload'] = payload
resp = json.dumps(data)
return utl.send_data(resp)
def create_dict_tfidf(self):
ds = DataSource()
dict_data = pd.DataFrame(ds.load_data(self.path)).review
dict_data = dict_data.tolist()
Util = util()
A = []
for i in range(len(dict_data)):
text = dict_data[i]
text = Util.text_util_final(text)
A.append(text)
vectorizer = TfidfVectorizer(max_features=100000, ngram_range=(1, 3))
vectorizer.fit(A)
return vectorizer
def create_tfidf_vector(path):
dict_tfidf = Dict_Tfidf(PATH)
vectorizer = dict_tfidf.create_dict_tfidf()
#load du lieu
ds = DataSource()
train_data = pd.DataFrame(ds.load_data(path))
x_train = train_data.review
y_train = train_data.label
# chuan hoa lai du lieu
x_train = x_train.tolist()
Util = util()
A = []
for i in range(len(x_train)):
text = x_train[i]
text = Util.text_util_final(text)
A.append(text)
#Chuyen ve vector tfidf
x_train_tfidf = vectorizer.transform(A)
return x_train_tfidf, y_train
def __init__(self, CV_fold=5, produceResult=False, forceRegen=False):
self.forceRegen = forceRegen
if self.forceRegen:
print(
'Dataset force regeneration is active, regenerating all datasets'
)
elif not self.forceRegen:
print(
'Dataset force regeneration deactivated, using existing datasets'
)
#Create working directories
self.dirCreate()
#Instantiate utility class
self.utils = util()
self.readData()
self.raw_standarization()
self.class_split()
if produceResult or self.forceRegen:
self.correlationMx()
self.CV_feature_plot()
self.feature_selection()
self.CV_datasetGen(fold=CV_fold, pca_type='Vani', test_ratio=0.2)
self.CV_datasetGen(fold=CV_fold, pca_type='Ker', test_ratio=0.2)
def __init__(self, bot):
self.bot = bot
self.db = ub.DBUtils()
self.u = ii.util(self.bot)
self.client = api.Client(None)
self.utils = ii.util(self.bot)