def __init__(self):
parameters = Parameters()
outputFolder = parameters.paths['FACT_FEEL_db']
print('{}'.format('creating fact feel\n'))
self.afinn = Afinn()
data_path = os.path.expanduser(FactFeel.corpora_path)
self.path = str(data_path + FactFeel.nlds_path)
self.facts = []
self.feels = []
self.doclist = {FactFeel.FACT: [], FactFeel.FEEL: []}
self.doc_list_path = str(self.path + 'doclist.json')
if self.doc_list_exists():
self.doclist = self.parse_doc_list()
else:
self.generate_doc_list()
list_of_documents = []
for doc in self.doclist:
pat = re.compile(
'(train|dev|test).(fact|feel).(fact|feel)(_\d+)(.txt)')
for other in self.doclist[doc]:
l = re.findall(pat, other)
capturegroup = l.pop()
parent = str(capturegroup[0:1][0]) + '/' + str(
capturegroup[1:2][0])
docleaf = str(capturegroup[2:3][0]) + str(capturegroup[3:4][0])
extension = capturegroup[4:5]
new_doc = Document(parent, docleaf, outputFolder)
list_of_documents.append(new_doc)
## this line prepares a doclist to be used for our flask server that feeds the annotation viewer
self.trasformed_doclist = self.transform_doclist()
remakeDB = False
if remakeDB:
self.docs = self.create_db(self.path, list_of_documents)
else:
temp_docs = []
for doc in list_of_documents:
temp = self.read_db(doc)
temp_docs.append(temp)
self.docs = temp_docs
self.facts = [
fact for fact in self.docs if 'fact' in fact.get_doc_leaf()
]
self.feels = [
feel for feel in self.docs if 'feel' in feel.get_doc_leaf()
]
from sklearn.base import TransformerMixin, BaseEstimator
from utils.Parameters import Parameters
import numpy as np
from pymongo import MongoClient
parameters = Parameters()
from afinn import Afinn
afinn = Afinn()
"""
configs:
0- No feature
1- sentence polarity TODO:(needs parameterization )
2- num_polarity_words / num_neutral_words
3- mean_pos mean_neg
4- median_pos median_neg
"""
class AfinnTransformer(TransformerMixin, BaseEstimator):
features = {'value': 0 }
def __init__(self, featureSetConfiguration = 1 ):
self.featureSetConfiguration = featureSetConfiguration
def transform(self, X, **transform_params):
if(self.featureSetConfiguration == 0): # not active
features = np.array([
AfinnTransformer.features.keys()
for s in X
])