Data Protection Program is a cybersecurity awareness platform targeted towards people without technical experience. It utilizes a Django backend with a React frontend, as well as Google Sign-In API and a Hugging Face Transformers model.
I used Hugging Face Transformers to train a model to detect phishing emails. I could only find two different datasets of emails with phishing labels, which meant I could only train the model on one of those datasets so I could test the accuracy with the other. I was able to train the model to 87% accuracy. The user can either copy and paste email text or upload an email file, and then receive a result of whether or not the email is phishing.
There's a page where users can input a potential password, and a list of similar passwords from rockyou.txt (a text document of over 14 million real passwords that attackers use to breach accounts) will be displayed. The inputted password isn't stored in any way.
The guide page has a list of links to different guides or articles that help educate people on staying safe online. This includes how to set up two-factor authentication, using a password manager, and even freezing credit bureaus. The home page has links to current cybersecurity news articles using NewsAPI, so users can stay aware of current events that may lead to a breach in their security.
The sign in feature using Google Sign-In API was going to be used to give users the ability to test their knowledge with adaptive quizzes, but I ran out of time before I could implement it.
The link to the site is on the GitHub page.
No comments:
Post a Comment