A browser extension or app that scans incoming emails for phishing threats using NLP. It flags risky links, poor grammar, sender mismatch, etc.
PhishGuard is a console-based application designed to detect phishing emails using a combination of machine learning (ML) and natural language processing (NLP) techniques. The tool aims to help users identify potential phishing attempts and enhance email security, especially for students, professionals, or anyone concerned about cybersecurity threats.
Email Content Analysis:
PhishGuard scans the content of emails for suspicious patterns, keywords, and links that are commonly found in phishing attempts. It processes the email text to detect potentially harmful messages.
Phishing Prediction:
The application uses a logistic regression ML model trained on a dataset of phishing and safe emails. The model predicts whether an email is phishing or safe based on its content, providing a confidence score.
Suspicious Link Extraction:
The app extracts URLs from the email text and checks for suspicious or blacklisted domains. It highlights any links that may lead to phishing websites.
Keyword Detection:
PhishGuard scans the email for keywords like "urgent," "verify," "update," "click," and "login," which are commonly used in phishing scams to create urgency or trick users.
Recommendations:
After scanning the email, the app gives the user feedback, such as whether the email is likely phishing or safe, and highlights red-flag keywords and links. It also provides general safety recommendations.
Python: The core language used for building the application.
Libraries: scikit-learn, pandas, numpy, nltk for NLP and ML model development.
Machine Learning: Logistic Regression with CountVectorizer for text feature extraction and prediction.
Natural Language Processing (NLP): Used for tokenizing text and identifying key phrases that indicate phishing.
Email Input: Users paste the content of an email or upload a text file with the email body.
Text Preprocessing: The application processes the input, extracting features like words, links, and suspicious terms.
Phishing Prediction: The trained ML model predicts the phishing risk and provides a confidence score.
Results Display: The app displays the phishing risk level (high, low), flags suspicious words, and lists any links found.
Expand Dataset: Add more data to improve prediction accuracy and handle various phishing tactics.
Integrate with Gmail/Outlook: Allow users to scan emails directly from their inboxes via Gmail API or Outlook API.
GUI Version: Develop a graphical user interface (GUI) for easier use, possibly using tools like Streamlit or Tkinter.
PhishGuard aims to serve as a simple but effective tool for educating users and enhancing email security by identifying phishing threats early.
Sure! Here's the entire **"Progress During Hackathon"** section rewritten as a single, smooth-flowing paragraph: --- During the hackathon, I began by identifying the rising threat of phishing attacks and finalized the concept of PhishGuard—an AI-powered tool to detect phishing emails. I chose Python as the core technology and focused on building a lightweight, console-based application that could analyze email content using simple Natural Language Processing techniques. The initial phase involved developing the core logic to scan for suspicious keywords and links, and calculating a phishing risk score based on severity. I tested the app with various sample emails—both genuine and malicious—to fine-tune the keyword sensitivity and confidence scoring. As development progressed, I enhanced the app with features like real-time link extraction and highlighted risky terms, making the results more informative and user-friendly. Finally, I prepared a demo video, created a pitch script, and documented the entire project on GitHub for submission. Phish Guard was designed, built, and tested entirely within the hackathon window, growing from an idea to a fully functional prototype.
NA