Spaces:
Sleeping
Sleeping
File size: 6,164 Bytes
00a16c9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 |
import torch
import pandas as pd
import argparse
import re
import matplotlib.pyplot as plt
import seaborn as sns
from transformers import AutoTokenizer, AutoModelForSequenceClassification
from sklearn.metrics import classification_report, confusion_matrix
# Define model names
bert_model_name = "bert-base-uncased"
hatebert_model_name = "GroNLP/hateBERT"
class CyberbullyingDetector:
def __init__(self, model_type="bert"):
if model_type == "bert":
self.tokenizer = AutoTokenizer.from_pretrained(bert_model_name)
self.model = AutoModelForSequenceClassification.from_pretrained(bert_model_name)
elif model_type == "hatebert":
self.tokenizer = AutoTokenizer.from_pretrained(hatebert_model_name)
self.model = AutoModelForSequenceClassification.from_pretrained(hatebert_model_name)
else:
raise ValueError("Invalid model_type. Choose 'bert' or 'hatebert'.")
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.model.to(self.device)
self.cyberbullying_threshold = 0.7 # Confidence threshold
self.trigger_words = [
'buang', 'pokpok', 'bogo', 'linte', 'tanga', 'diputa', 'yuta mo', 'gaga',
'lagtok ka', 'addict', 'bogok', 'gago', 'law-ay', 'demonyo ka', 'animal ka', 'animal',
'bilatibay', 'yudipota', 'pangit', 'tikalon', 'tinikal', 'hambog',
'batinggilan', 'biga-on', 'bulay-ug', 'agi', 'agitot', 'alpot', 'hangag'
]
def find_triggers(self, text):
text_lower = text.lower()
return [word for word in self.trigger_words if word in text_lower]
def predict(self, text):
triggers = self.find_triggers(text)
inputs = self.tokenizer(
text,
return_tensors="pt",
truncation=True,
max_length=128,
padding=True
).to(self.device)
with torch.no_grad():
outputs = self.model(**inputs)
logits = outputs.logits
probs = torch.nn.functional.softmax(logits, dim=1)
pred_class = torch.argmax(probs).item()
confidence = probs[0][pred_class].item()
if (pred_class == 1 and confidence >= self.cyberbullying_threshold) or (len(triggers) > 0):
label = "Cyberbullying"
is_cyberbullying = True
else:
label = "Safe"
is_cyberbullying = False
return {
"text": text,
"label": label,
"confidence": confidence,
"language": "hil",
"triggers": triggers,
"is_cyberbullying": is_cyberbullying
}
def save_confusion_matrix(y_true, y_pred, filename="confusion_matrix.png"):
labels = sorted(set(y_true + y_pred))
cm = confusion_matrix(y_true, y_pred, labels=labels)
plt.figure(figsize=(6, 4))
sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=labels, yticklabels=labels)
plt.title("Confusion Matrix")
plt.xlabel("Predicted")
plt.ylabel("True")
plt.tight_layout()
plt.savefig(filename)
plt.close()
print(f"πΈ Confusion matrix saved as {filename}")
def run_predictions(input_csv=None, output_csv=None, model_type="bert"):
detector = CyberbullyingDetector(model_type=model_type)
if input_csv:
df = pd.read_csv(input_csv)
results = [detector.predict(text) for text in df['tweet_text']]
output_df = df.copy()
output_df['predicted_label'] = [r['label'] for r in results]
output_df['confidence'] = [r['confidence'] for r in results]
output_df['language'] = [r['language'] for r in results]
output_df['triggers'] = [', '.join(r['triggers']) for r in results]
output_df['is_cyberbullying'] = [r['is_cyberbullying'] for r in results]
output_df['true_label'] = output_df['cyberbullying_type'].apply(
lambda x: "Cyberbullying" if pd.notnull(x) and str(x).strip().lower() != "none" else "Safe"
)
if output_csv:
output_df.to_csv(output_csv, index=False)
print(f"\nβ
Predictions saved to {output_csv}")
print("\nπ Sample Predictions:")
print(output_df[['tweet_text', 'predicted_label', 'confidence', 'triggers']].head(10).to_string(index=False))
print("\nπ Prediction Summary:")
print(output_df['predicted_label'].value_counts())
print("\nβ
Ground Truth Summary:")
print(output_df['true_label'].value_counts())
accuracy = (output_df['predicted_label'] == output_df['true_label']).mean()
print(f"\nπ― Accuracy: {accuracy:.2%}")
print("\nπ§Ύ Classification Report:")
print(classification_report(output_df['true_label'], output_df['predicted_label'], digits=2, zero_division=0))
save_confusion_matrix(output_df['true_label'].tolist(), output_df['predicted_label'].tolist())
return output_df
else:
# Inference mode
print("\nπ Type a sentence to analyze (or 'exit' to quit):")
while True:
text = input(">>> ")
if text.lower() in ["exit", "quit"]:
break
result = detector.predict(text)
print(result)
if __name__ == '__main__':
parser = argparse.ArgumentParser(description="Cyberbullying Detector")
parser.add_argument('--input_file', type=str, help="Path to input CSV with 'tweet_text' column")
parser.add_argument('--output_file', type=str, help="Path to save results CSV")
parser.add_argument('--model', type=str, default='bert', choices=['bert', 'hatebert'], help="Model to use")
args = parser.parse_args()
if args.input_file:
print(f"π₯ Running batch predictions from {args.input_file} using {args.model.upper()}...")
else:
print(f"π§ͺ No input file. Running in interactive mode using {args.model.upper()}...")
run_predictions(args.input_file, args.output_file, model_type=args.model)
|