The purpose of the project was to create a browser extension that prevents users from loading phishing websites. The project consisted of three components. The first component was a browser extension - a user interface for the project. The second component was an API translating raw URLs into data for the machine learning model. The last component, the machine learning model, analysed the data and predicted the probability of phishing. If the result from the server had a high probability, a user was blocked from loading the page. Several models were trained with a different set of features to find the most effective model.
After installation the default behaviour of the extension is to block the phishing websites.
Left clicking extension icon shows a popup window with basic setings:
Using the green slider the phishing check can be disabled or enabled.
The button with text "Options" navigates to more advanced settings. The effect of clicking this button is the same as right clicking the extension icon and selecting "options". For details see the Options section below.
The button navigates to this page.
The options page also includes slider to disable or enable phishing check, but additionally contains a whitelist functionality.
The URLs in the whitelist field can be edited. The list can be either coma separated list, or one URL can be entered per line as shown in the image.
To save the whitelist, the button must be clicked.
On this page the address of the potentially phishing website is shown in bold.
By clicking the button the user can navigate to previous page. By clicking the button the user can visit the site ignoring the warning and automatically add it to the whitelist.