Phishing email detection

Web front page

Inspiration

Phishing emails are one of the most common cyber threats today. Many users fall victim because phishing messages look professional and create urgency. We were inspired to build a system that helps users identify malicious emails before they click harmful links. Our goal was to create a practical and easy-to-use phishing detection solution.

What it does

Our project detects phishing emails using a hybrid approach. It analyzes email content, suspicious keywords, links, and patterns to classify emails as Legitimate, Suspicious, or Phishing. The system assigns a risk score based on predefined rules and machine learning predictions to improve accuracy.

How we built it

We built the system using Python. The project combines: Rule-based detection for suspicious patterns Machine learning model (Logistic Regression with TF-IDF) Email text preprocessing and feature extraction Streamlit for user interface The system takes email content as input and outputs a phishing risk classification.

Challenges we ran into

One of the major challenges was handling noisy email text and cleaning the dataset properly. Another challenge was balancing accuracy and false positives. We also worked on tuning the model to improve precision and recall while maintaining explainability.

Accomplishments that we're proud of

We successfully developed a working phishing email detection system with a clear risk scoring mechanism. The hybrid approach improved reliability compared to using only rules or only machine learning. The system provides user-friendly outputs and practical cybersecurity relevance.

What we learned

Through this project, we learned about: Natural Language Processing (NLP) techniques Feature extraction using TF-IDF Machine learning model training and evaluation Importance of precision, recall, and F1-score Real-world cybersecurity challenges We also improved our problem-solving and teamwork skills.

What's next for Phishing email detection

In the future, we plan to: Integrate deep learning models for better accuracy Add real-time email scanning capability Deploy the system as a web-based security tool Improve detection of image-based phishing emails Our aim is to make the system more scalable and production-ready.