Commit ·
2e4daec
1
Parent(s): 861610c
add emoji preprocessor
Browse files- functions.py +17 -6
functions.py
CHANGED
|
@@ -1,14 +1,25 @@
|
|
| 1 |
from transformers import AutoTokenizer, AutoConfig, AutoModelForSequenceClassification
|
| 2 |
from scipy.special import softmax
|
| 3 |
-
|
| 4 |
# Define the preprocess function
|
| 5 |
-
def
|
| 6 |
new_text = []
|
| 7 |
for t in text.split(" "):
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
return " ".join(new_text)
|
| 13 |
|
| 14 |
# Define the sentiment_analysis function
|
|
|
|
| 1 |
from transformers import AutoTokenizer, AutoConfig, AutoModelForSequenceClassification
|
| 2 |
from scipy.special import softmax
|
| 3 |
+
|
| 4 |
# Define the preprocess function
|
| 5 |
+
def process_text(text):
|
| 6 |
new_text = []
|
| 7 |
for t in text.split(" "):
|
| 8 |
+
# Check if the token is an emoji
|
| 9 |
+
if emoji.is_emoji(t):
|
| 10 |
+
# Add the emoji to the new text list
|
| 11 |
+
new_text.append(t)
|
| 12 |
+
else:
|
| 13 |
+
# Check if the token starts with '@' and has more than one character
|
| 14 |
+
if t.startswith('@') and len(t) > 1:
|
| 15 |
+
# Replace the token with '@user'
|
| 16 |
+
t = '@user'
|
| 17 |
+
# Check if the token starts with 'http'
|
| 18 |
+
if t.startswith('http'):
|
| 19 |
+
# Replace the token with 'http'
|
| 20 |
+
t = 'http'
|
| 21 |
+
# Add the processed token to the new text list
|
| 22 |
+
new_text.append(t)
|
| 23 |
return " ".join(new_text)
|
| 24 |
|
| 25 |
# Define the sentiment_analysis function
|