def praposes(a):
#a = teks
a = pps.preprocessing(a)
a = pps.removePunc(a)
a = ct.norm_typo(a)
a = nkbi.norm_kbbi(a)
a = stm.stemmer_kata(a)
a = stm.stop_word(a)
return a
def praposes(a, normalisasi=True):
#a = teks
if normalisasi == True:
a = pps.gantiKarakter(a)
a = pps.preprocessing(a)
a = pps.removePunc(a)
if normalisasi == True:
a = ct.cek_typo(a)
a = nkbi.norm_kbbi(a)
a = stm.stemmer_kata(a)
a = stm.stop_word(a)
return a
def praproses(teks):
a = tknzr.tokenize(teks)
a = " ".join(a)
a = pps.preprocessing(teks)
a = pps.removePunc(a)
a = ct.norm_typo(a)
a = nkbi.norm_kbbi(a)
a = stm.stemmer_kata(a)
a = re.sub(' +', ' ',a)
a = a.lstrip()
return a
def saves():
stm.save_kta()
def savek():
nkbi.save_gdiganti()
ct.save_gdiganti()
stm.save_kta()
def stem_dulu(list_kata):
n_list = list()
for i in list_kata:
i = stm.stemmer_kata(i)
n_list.append(i)
return list(set(n_list))