mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-11 20:28:20 +03:00
13 lines
664 B
Python
13 lines
664 B
Python
|
# coding: utf8
|
||
|
|
from __future__ import unicode_literals
|
||
|
|
|
||
|
|
# stop words as whitespace-separated list.
|
||
|
|
# Source: https://raw.githubusercontent.com/dohliam/more-stoplists/master/yo/yo.txt
|
||
|
|
|
||
|
|
STOP_WORDS = set(
|
||
|
|
"a an b bá bí bẹ̀rẹ̀ d e f fún fẹ́ g gbogbo i inú j jù jẹ jẹ́ k kan kì kí kò "
|
||
|
|
"l láti lè lọ m mi mo máa mọ̀ n ni náà ní nígbà nítorí nǹkan o p padà pé "
|
||
|
|
"púpọ̀ pẹ̀lú r rẹ̀ s sì sí sínú t ti tí u w wà wá wọn wọ́n y yìí à àti àwọn á "
|
||
|
|
"è é ì í ò òun ó ù ú ń ńlá ǹ ̀ ́ ̣ ṣ ṣe ṣé ṣùgbọ́n ẹ ẹmọ́ ọ ọjọ́ ọ̀pọ̀lọpọ̀".split()
|
||
|
|
)
|