Spaces:
Sleeping
Sleeping
Commit ·
9d9f0fa
1
Parent(s): 0e69286
lalal
Browse files- feature_extract.py +5 -1
feature_extract.py
CHANGED
|
@@ -1,8 +1,13 @@
|
|
| 1 |
import nltk
|
|
|
|
|
|
|
|
|
|
| 2 |
try:
|
| 3 |
stopwords.words("english")
|
| 4 |
except LookupError:
|
| 5 |
nltk.download("stopwords", quiet=True)
|
|
|
|
|
|
|
| 6 |
try:
|
| 7 |
from nltk.corpus import twitter_samples
|
| 8 |
twitter_samples.fileids()
|
|
@@ -14,7 +19,6 @@ import string
|
|
| 14 |
import numpy as np
|
| 15 |
from nltk.stem import PorterStemmer
|
| 16 |
from nltk.tokenize import TweetTokenizer
|
| 17 |
-
from nltk.corpus import stopwords
|
| 18 |
|
| 19 |
# --- constants & tools ---
|
| 20 |
pronouns = {
|
|
|
|
| 1 |
import nltk
|
| 2 |
+
from nltk.corpus import stopwords
|
| 3 |
+
|
| 4 |
+
# tải stopwords
|
| 5 |
try:
|
| 6 |
stopwords.words("english")
|
| 7 |
except LookupError:
|
| 8 |
nltk.download("stopwords", quiet=True)
|
| 9 |
+
|
| 10 |
+
# tải twitter_samples nếu cần
|
| 11 |
try:
|
| 12 |
from nltk.corpus import twitter_samples
|
| 13 |
twitter_samples.fileids()
|
|
|
|
| 19 |
import numpy as np
|
| 20 |
from nltk.stem import PorterStemmer
|
| 21 |
from nltk.tokenize import TweetTokenizer
|
|
|
|
| 22 |
|
| 23 |
# --- constants & tools ---
|
| 24 |
pronouns = {
|