BOUESTI B.Sc Computer Science Final Year Project
Intelligent Spam Email Detection System
Machine Learning-Powered Classification for Secure Email Communication using TF-IDF Feature Extraction and High-Accuracy Supervised Classifiers.
33,716
Enron Emails Trained
4 ML
Algorithms Compared
99.00%
Accuracy Achieved
< 10ms
Real-Time Latency
How The System Works
End-to-end natural language processing pipeline converting raw email content into accurate predictions.
1
1. Input Email Text
Paste email text or choose from preset phishing and legitimate work email test samples.
2
2. Feature Extraction
Text is cleaned, tokenized, lemmatized, and converted into 10,000 TF-IDF numerical feature vectors.
3
3. Instant Prediction
Get immediate Spam vs. Legitimate status banner, confidence score meter, and highlighted trigger words.
Evaluated ML Models
Trained on 33,716 Enron emails with 5-fold cross-validation.
DEPLOYED
Logistic Regression
Accuracy:99.00%
F1-Score:0.9896
Recall:99.35%
Support Vector Machine
Accuracy:98.95%
F1-Score:0.9891
Recall:99.25%
Random Forest
Accuracy:98.44%
F1-Score:0.9838
Recall:99.04%
Multinomial Naive Bayes
Accuracy:98.39%
F1-Score:0.9832
Recall:98.28%