Add preprocessing for replacing polish characters
This commit is contained in:
parent
3f02c41b1f
commit
d38fbee2ec
@ -2,13 +2,13 @@ import re
|
||||
|
||||
from acts import hello_act, name_request_act
|
||||
|
||||
hello = ['dzie[ńn] dobry', 'dobry wiecz[oó]r', 'witam', 'witaj', 'siema', 'elo', 'cze[śs][ćc]']
|
||||
request_name = ['imi[eę]', 'nazywasz']
|
||||
hello = ['dzien dobry', 'dobry wieczor', 'witam', 'witaj', 'siema', 'elo', 'czesc']
|
||||
request_name = ['imie', 'nazywasz']
|
||||
question = ['\?$']
|
||||
|
||||
acts = {hello_act: [hello],
|
||||
name_request_act: [request_name, question]}
|
||||
|
||||
|
||||
def analizator_jezyka_naturalnego(text):
|
||||
text = text_preprocess(text)
|
||||
frame = act_check(text)
|
||||
@ -16,6 +16,18 @@ def analizator_jezyka_naturalnego(text):
|
||||
|
||||
def text_preprocess(text):
|
||||
text = text.lower()
|
||||
text = text.replace("ą", "a")
|
||||
text = text.replace("ć", "c")
|
||||
text = text.replace("ę", "e")
|
||||
text = text.replace("ł", "l")
|
||||
text = text.replace("ń", "n")
|
||||
text = text.replace("ó", "o")
|
||||
text = text.replace("ś", "s")
|
||||
text = text.replace("ź", "z")
|
||||
text = text.replace("ż", "z")
|
||||
text = text.replace("\n", " ")
|
||||
text = text.replace("\t", " ")
|
||||
text = text.replace(" ", " ")
|
||||
return text
|
||||
|
||||
def act_check(text):
|
||||
|
Loading…
Reference in New Issue
Block a user