Shymaa2611 commited on
Commit ·
2afdd48
1
Parent(s): 05e8fd6
Add Application File
Browse files- dataset.py +7 -5
dataset.py
CHANGED
|
@@ -8,11 +8,13 @@ from nltk import word_tokenize
|
|
| 8 |
import string
|
| 9 |
import nltk
|
| 10 |
import os
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
nltk.
|
| 15 |
-
nltk.download("
|
|
|
|
|
|
|
| 16 |
|
| 17 |
|
| 18 |
def remove_html_tags(text):
|
|
|
|
| 8 |
import string
|
| 9 |
import nltk
|
| 10 |
import os
|
| 11 |
+
import nltk
|
| 12 |
+
nltk_data_dir = os.path.expanduser("~/app/nltk_data")
|
| 13 |
+
os.makedirs(nltk_data_dir, exist_ok=True)
|
| 14 |
+
nltk.data.path.append(nltk_data_dir)
|
| 15 |
+
nltk.download("stopwords", download_dir=nltk_data_dir)
|
| 16 |
+
nltk.download("punkt", download_dir=nltk_data_dir)
|
| 17 |
+
nltk.download("wordnet", download_dir=nltk_data_dir)
|
| 18 |
|
| 19 |
|
| 20 |
def remove_html_tags(text):
|