A Deep Learning based Smishing Detection System that identifies malicious SMS messages using CNN, LSTM, and Bidirectional LSTM models.
To improve smishing detection accuracy while reducing false positives using advanced deep learning techniques and natural language processing.
- SMS Text Preprocessing using NLTK
- TF-IDF Feature Extraction
- LSTM Model
- BiLSTM Model
- CNN + LSTM Hybrid Model
- CNN + LSTM + BiLSTM Extended Model
- Flask Based Web Application
- High Accuracy (98.96%)
- Python
- TensorFlow
- Keras
- Scikit-Learn
- Pandas
- NumPy
- NLTK
- Flask
SMS Message ↓ Data Preprocessing ↓ Tokenization ↓ Stopword Removal ↓ Stemming & Lemmatization ↓ TF-IDF Feature Extraction ↓ Dataset Splitting ↓ Model Training ↓ CNN + LSTM + BiLSTM ↓ Evaluation ↓ Prediction
The proposed hybrid deep learning architecture achieved excellent performance in detecting smishing messages.
| Model | Accuracy |
|---|---|
| GRU | Evaluated |
| LSTM | Evaluated |
| BiLSTM | Evaluated |
| CNN + LSTM | 98.83% |
| CNN + LSTM + BiLSTM | 98.96% |
- Real-Time SMS Detection
- Android Application Integration
- Multilingual Support
- BERT-Based Detection Models
- numpy
- pandas
- matplotlib
- seaborn
- scikit-learn
- tensorflow
- keras
- nltk
- flask
After creating the repository:
- git init
- git add .
- git commit -m "Initial commit"
- git branch -M main
- git remote add origin https://github.com/SambitKumarBisoi/enhancing-smishing-detection-using-deep-learning.git
- git push -u origin main
enhancing-smishing-detection-using-deep-learning/ ├── README.md ├── requirements.txt ├── LICENSE ├── .gitignore ├── datasets/ ├── notebooks/ ├── models/ ├── screenshots/ ├── app/ ├── report/ └── presentation/
Project Lead & Deep Learning Model Development
Data Preprocessing & Feature Engineering
Dataset Collection & Analysis
Flask Deployment & Web Interface
✔ Achieved 98.96% Accuracy
✔ Reduced False Positives
✔ Implemented CNN + LSTM + BiLSTM
✔ Built Flask-Based Web Application
✔ Applied NLP Techniques for SMS Analysis