Spaces:
Sleeping
Sleeping
File size: 22,077 Bytes
8a3639a c03ee07 8a3639a c03ee07 8a3639a c03ee07 8a3639a c03ee07 8a3639a c03ee07 8a3639a da2ceec 8a3639a c03ee07 8a3639a c03ee07 8a3639a |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 |
import gradio as gr
import torch
import torch.nn as nn
from transformers import AutoTokenizer, AutoModel
import pandas as pd
import json
from datetime import datetime
import plotly.graph_objects as go
import plotly.express as px
class BERTScamClassifier(nn.Module):
"""BERT-based classifier for scam detection"""
def __init__(self, model_name='bert-base-multilingual-cased', n_classes=2, dropout=0.3):
super(BERTScamClassifier, self).__init__()
self.bert = AutoModel.from_pretrained(model_name)
self.dropout = nn.Dropout(dropout)
self.classifier = nn.Linear(self.bert.config.hidden_size, n_classes)
def forward(self, input_ids, attention_mask):
outputs = self.bert(
input_ids=input_ids,
attention_mask=attention_mask
)
pooled_output = outputs.pooler_output
output = self.dropout(pooled_output)
return self.classifier(output)
class GradioScamDetector:
"""Gradio web app for scam detection"""
def __init__(self, model_path='bert_scam_detector.pth'):
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
self.model = None
self.tokenizer = None
self.id2label = {0: 'trust', 1: 'scam'}
self.max_length = 128
self.prediction_history = []
# Load model
self.load_model(model_path)
def load_model(self, model_path):
"""Load the trained model"""
try:
checkpoint = torch.load(model_path, map_location=self.device)
model_name = checkpoint.get('model_name', 'bert-base-multilingual-cased')
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
self.model = BERTScamClassifier(model_name)
self.model.load_state_dict(checkpoint['model_state_dict'])
self.model.to(self.device)
self.model.eval()
self.max_length = checkpoint.get('max_length', 128)
self.id2label = checkpoint.get('id2label', {0: 'trust', 1: 'scam'})
print("✅ Model loaded successfully for Gradio app!")
return True
except Exception as e:
print(f"❌ Error loading model: {e}")
return False
def predict_message(self, message):
"""Predict if a message is scam or trust"""
if not message or not message.strip():
return "⚠️ Please enter a message", 0.0, "No prediction", {}
message = message.strip()
# Tokenize message
encoding = self.tokenizer(
message,
truncation=True,
padding='max_length',
max_length=self.max_length,
return_tensors='pt'
)
input_ids = encoding['input_ids'].to(self.device)
attention_mask = encoding['attention_mask'].to(self.device)
with torch.no_grad():
outputs = self.model(input_ids, attention_mask)
probabilities = torch.nn.functional.softmax(outputs, dim=1)
_, prediction = torch.max(outputs, dim=1)
predicted_label = self.id2label[prediction.item()]
confidence = probabilities[0][prediction.item()].item()
trust_prob = probabilities[0][0].item()
scam_prob = probabilities[0][1].item()
# Format result with emoji
if predicted_label == 'scam':
result_text = f"🚫 SCAM DETECTED"
color = "red"
else:
result_text = f"✅ TRUSTED MESSAGE"
color = "green"
# Confidence level description
if confidence >= 0.9:
conf_desc = "Very High"
elif confidence >= 0.75:
conf_desc = "High"
elif confidence >= 0.6:
conf_desc = "Medium"
else:
conf_desc = "Low"
# Store prediction history
self.prediction_history.append({
'timestamp': datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
'message': message[:50] + "..." if len(message) > 50 else message,
'prediction': predicted_label,
'confidence': confidence,
'trust_prob': trust_prob,
'scam_prob': scam_prob
})
# Create probability chart
prob_chart = self.create_probability_chart(trust_prob, scam_prob)
# Detailed results
details = f"""
**Prediction:** {result_text}
**Confidence:** {confidence:.1%} ({conf_desc})
**Device:** {self.device}
**Message Length:** {len(message)} characters
"""
return result_text, confidence, details, prob_chart
def predict_api(self, message):
"""API-friendly prediction function for webhooks"""
if not message or not message.strip():
return {
"status": "error",
"message": "Empty message",
"prediction": "unknown",
"confidence": 0.0
}
message = message.strip()
try:
# Tokenize message
encoding = self.tokenizer(
message,
truncation=True,
padding='max_length',
max_length=self.max_length,
return_tensors='pt'
)
input_ids = encoding['input_ids'].to(self.device)
attention_mask = encoding['attention_mask'].to(self.device)
with torch.no_grad():
outputs = self.model(input_ids, attention_mask)
probabilities = torch.nn.functional.softmax(outputs, dim=1)
_, prediction = torch.max(outputs, dim=1)
predicted_label = self.id2label[prediction.item()]
confidence = probabilities[0][prediction.item()].item()
trust_prob = probabilities[0][0].item()
scam_prob = probabilities[0][1].item()
# Format result
if predicted_label == 'scam':
result_text = "🚫 SCAM DETECTED"
alert_level = "HIGH"
else:
result_text = "✅ TRUSTED MESSAGE"
alert_level = "LOW"
# Store prediction history
self.prediction_history.append({
'timestamp': datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
'message': message[:50] + "..." if len(message) > 50 else message,
'prediction': predicted_label,
'confidence': confidence,
'trust_prob': trust_prob,
'scam_prob': scam_prob,
'source': 'API'
})
return {
"status": "success",
"message": message[:100] + "..." if len(message) > 100 else message,
"prediction": predicted_label,
"result_text": result_text,
"confidence": round(confidence, 4),
"trust_probability": round(trust_prob, 4),
"scam_probability": round(scam_prob, 4),
"alert_level": alert_level,
"timestamp": datetime.now().isoformat()
}
except Exception as e:
return {
"status": "error",
"message": f"Prediction failed: {str(e)}",
"prediction": "unknown",
"confidence": 0.0
}
def create_probability_chart(self, trust_prob, scam_prob):
"""Create probability visualization"""
fig = go.Figure(data=[
go.Bar(
x=['Trust', 'Scam'],
y=[trust_prob, scam_prob],
marker_color=['green', 'red'],
text=[f'{trust_prob:.1%}', f'{scam_prob:.1%}'],
textposition='auto',
)
])
fig.update_layout(
title="Prediction Probabilities",
yaxis_title="Probability",
xaxis_title="Classification",
showlegend=False,
height=300,
margin=dict(l=20, r=20, t=40, b=20)
)
return fig
def batch_predict(self, file):
"""Batch prediction from uploaded file"""
if file is None:
return "⚠️ Please upload a file", None
try:
# Read file based on extension
if file.name.endswith('.csv'):
df = pd.read_csv(file.name)
if 'message' in df.columns:
messages = df['message'].tolist()
else:
messages = df.iloc[:, 0].tolist() # First column
elif file.name.endswith('.txt'):
with open(file.name, 'r', encoding='utf-8') as f:
messages = [line.strip() for line in f if line.strip()]
else:
return "❌ Unsupported file format. Use CSV or TXT files.", None
# Process messages
results = []
for i, message in enumerate(messages[:100]): # Limit to 100 messages
if message and message.strip():
pred_label, confidence, _, _ = self.predict_message(message)
results.append({
'Message': message[:100] + "..." if len(message) > 100 else message,
'Prediction': pred_label,
'Confidence': f"{confidence:.1%}"
})
# Create results DataFrame
results_df = pd.DataFrame(results)
# Summary
scam_count = len([r for r in results if 'SCAM' in r['Prediction']])
trust_count = len(results) - scam_count
summary = f"""
📊 **Batch Processing Complete**
- Total Messages: {len(results)}
- 🚫 Scam Messages: {scam_count}
- ✅ Trusted Messages: {trust_count}
- 📈 Scam Rate: {scam_count/len(results):.1%}
"""
return summary, results_df
except Exception as e:
return f"❌ Error processing file: {str(e)}", None
def get_prediction_history(self):
"""Get prediction history as DataFrame"""
if not self.prediction_history:
return pd.DataFrame({'Message': ['No predictions yet']})
df = pd.DataFrame(self.prediction_history[-20:]) # Last 20 predictions
df['Confidence'] = df['confidence'].apply(lambda x: f"{x:.1%}")
df['Prediction'] = df['prediction'].apply(lambda x: f"🚫 {x.upper()}" if x == 'scam' else f"✅ {x.upper()}")
df['Source'] = df.get('source', 'Manual')
return df[['timestamp', 'message', 'Prediction', 'Confidence', 'Source']].rename(columns={
'timestamp': 'Time',
'message': 'Message',
})
def clear_history(self):
"""Clear prediction history"""
self.prediction_history = []
return pd.DataFrame({'Message': ['History cleared']})
def get_sample_messages(self):
"""Get sample messages for testing"""
return {
"Swahili Scam": "Hongera! Umeshinda Sh 5,000,000. Tuma PIN yako sasa kupokea zawadi yako!",
"English Scam": "CONGRATULATIONS! You've won $1,000,000. Send your bank details immediately!",
"Swahili Trust": "Habari za leo? Natumai uko salama na kila kitu ni sawa",
"English Trust": "Hi there! How was your day today? Hope everything is going well",
"Mixed Language": "Hi, kikao kitafanyika kesho at 2 PM. Don't forget!",
"Suspicious": "URGENT: Your account will be suspended. Click link to verify now!"
}
# Global detector instance for API endpoints
detector = None
def create_gradio_app():
"""Create and configure Gradio interface"""
global detector
# Initialize detector
detector = GradioScamDetector()
# Custom CSS for better styling
css = """
.gradio-container {
max-width: 1200px !important;
}
.result-box {
font-size: 18px !important;
font-weight: bold !important;
text-align: center !important;
padding: 20px !important;
border-radius: 10px !important;
}
.scam-result {
background-color: #ffebee !important;
color: #c62828 !important;
border: 2px solid #f44336 !important;
}
.trust-result {
background-color: #e8f5e8 !important;
color: #2e7d32 !important;
border: 2px solid #4caf50 !important;
}
"""
# Create Gradio interface
with gr.Blocks(css=css, title="🛡️ BERT Scam Detector", theme=gr.themes.Soft()) as demo:
# Header
gr.Markdown("""
# 🛡️ BERT Scam Detector
### Intelligent SMS Scam Detection for Swahili & English
This AI system uses advanced BERT language models to detect scam messages in both Swahili and English.
Simply enter a message below to check if it's legitimate or potentially fraudulent.
""")
# API Information Tab
with gr.Tab("🔌 API Integration"):
gr.Markdown("""
## 📡 API Endpoints for IFTTT/Zapier Integration
### For IFTTT Webhook:
```
URL: https://jacksonwambali-bert.hf.space/api/predict
Method: POST
Content-Type: application/json
Body: {"data": ["Your SMS message here"]}
```
### For Zapier Webhook:
```
URL: https://jacksonwambali-bert.hf.space/api/predict
Method: POST
Content-Type: application/json
Payload: {"data": ["{{sms_text}}"]}
```
### Response Format:
```json
{
"data": [
{
"status": "success",
"prediction": "scam" or "trust",
"result_text": "🚫 SCAM DETECTED" or "✅ TRUSTED MESSAGE",
"confidence": 0.95,
"alert_level": "HIGH" or "LOW"
}
]
}
```
### Quick Test:
Use the form below to test your API integration:
""")
with gr.Row():
with gr.Column():
api_test_input = gr.Textbox(
label="📱 Test SMS Message",
placeholder="Enter SMS to test API response...",
lines=3
)
api_test_btn = gr.Button("🧪 Test API Response", variant="primary")
with gr.Column():
api_response = gr.JSON(label="📊 API Response")
api_test_btn.click(
fn=lambda msg: detector.predict_api(msg) if detector else {"error": "Model not loaded"},
inputs=api_test_input,
outputs=api_response
)
# Main prediction interface
with gr.Tab("🔍 Single Message Detection"):
with gr.Row():
with gr.Column(scale=2):
message_input = gr.Textbox(
label="📝 Enter SMS Message",
placeholder="Type or paste your SMS message here...",
lines=4,
max_lines=8
)
with gr.Row():
predict_btn = gr.Button("🔍 Analyze Message", variant="primary", size="lg")
clear_btn = gr.Button("🗑️ Clear", variant="secondary")
# Sample messages
gr.Markdown("### 📋 Quick Test Samples:")
sample_messages = detector.get_sample_messages()
with gr.Row():
for name, msg in list(sample_messages.items())[:3]:
gr.Button(name, size="sm").click(
lambda msg=msg: msg, outputs=message_input
)
with gr.Row():
for name, msg in list(sample_messages.items())[3:]:
gr.Button(name, size="sm").click(
lambda msg=msg: msg, outputs=message_input
)
with gr.Column(scale=2):
# Results
result_text = gr.Textbox(
label="🎯 Prediction Result",
interactive=False,
elem_classes=["result-box"]
)
confidence_slider = gr.Slider(
label="📊 Confidence Level",
minimum=0,
maximum=1,
interactive=False,
show_label=True
)
details_md = gr.Markdown(label="📋 Detailed Analysis")
prob_chart = gr.Plot(label="📈 Probability Distribution")
# Batch processing tab
with gr.Tab("📁 Batch Processing"):
gr.Markdown("### Upload a file with multiple messages for batch analysis")
with gr.Row():
with gr.Column():
file_upload = gr.File(
label="📄 Upload File (CSV or TXT)",
file_types=[".csv", ".txt"]
)
batch_btn = gr.Button("🚀 Process Batch", variant="primary")
with gr.Column():
batch_summary = gr.Markdown(label="📊 Summary")
batch_results = gr.Dataframe(
label="📋 Batch Results",
interactive=False,
wrap=True
)
# History tab
with gr.Tab("📚 Prediction History"):
with gr.Row():
refresh_btn = gr.Button("🔄 Refresh History", variant="secondary")
clear_history_btn = gr.Button("🗑️ Clear History", variant="secondary")
history_df = gr.Dataframe(
label="📋 Recent Predictions",
interactive=False,
wrap=True
)
# About tab
with gr.Tab("ℹ️ About"):
gr.Markdown("""
## 🤖 About This System
### How It Works
- **Model**: BERT (Bidirectional Encoder Representations from Transformers)
- **Languages**: Swahili and English
- **Training**: Fine-tuned on SMS scam detection dataset
- **Accuracy**: High precision scam detection
### Features
- ✅ Real-time message analysis
- 🌍 Multilingual support (Swahili & English)
- 📊 Confidence scoring
- 📁 Batch processing
- 📚 Prediction history
- 🔌 API integration for IFTTT/Zapier
### SMS Integration
- Connect with IFTTT for automatic SMS scanning
- Webhook support for real-time alerts
- Batch processing for multiple messages
### Usage Tips
- Enter complete SMS messages for best results
- The system works with both languages simultaneously
- Higher confidence scores indicate more reliable predictions
- Check the probability distribution for detailed insights
### Safety Notice
- This is an AI assistant - use your judgment
- Report suspicious messages to authorities
- Never share personal information with untrusted sources
---
**Powered by BERT & Gradio** | Made with ❤️ for SMS security
""")
# Event handlers
predict_btn.click(
fn=detector.predict_message,
inputs=message_input,
outputs=[result_text, confidence_slider, details_md, prob_chart]
)
clear_btn.click(
fn=lambda: ("", 0, "", None),
outputs=[message_input, confidence_slider, details_md, prob_chart]
)
batch_btn.click(
fn=detector.batch_predict,
inputs=file_upload,
outputs=[batch_summary, batch_results]
)
refresh_btn.click(
fn=detector.get_prediction_history,
outputs=history_df
)
clear_history_btn.click(
fn=detector.clear_history,
outputs=history_df
)
# Auto-refresh history on prediction
predict_btn.click(
fn=detector.get_prediction_history,
outputs=history_df
)
return demo
def main():
"""Launch the Gradio app"""
print("🚀 Starting BERT Scam Detector Web App...")
# Create and launch app
app = create_gradio_app()
# Launch with custom settings
app.launch(
server_name="0.0.0.0", # Allow external access
server_port=7860, # Default Gradio port
share=True, # Set to True for public link
debug=False,
show_error=False,
quiet=False,
inbrowser=True # Auto-open browser
)
if __name__ == "__main__":
main() |