news-analyzer

This project is a Chrome extension for analyzing online news articles. It uses Fakebox, a machine learning API that analyzes the title, content, and domain name of online news articles. Learn more about the machine learning algorithm here.

Installation

Set up and run Fakebox

1. Install Docker. This is the container the Fakebox API will run on.
2. Sign into machinebox and obtain your box key.
3. Open a terminal and run: docker run -p 8080:8080 -e "MB_KEY=$MB_KEY" machinebox/fakebox
4. Go to http://localhost:8080/ in your browser to access the Fakebox console.

Set up Chrome extension

1. Download or clone this repository. You can clone this repository by opening a terminal and running git clone https://github.com/emmaling27/news-analyzer.git.
2. Open chrome://extensions/ in Chrome.
3. Click "Developer mode" on the top right.
4. Click "Load unpacked" and select this repository. You should now see your extension in the list and in the toolbar.

Using the Chrome extension

Now that you've installed the Chrome extension and have Fakebox up and running, when you click on the News Analyzer icon, you should see a popup with a big green button that says "Analyze news." If you click, you should see a loader, and then text describing the source and the bias of the content on the current page. A bar at the bottom of the popup shows how biased the content of the current page is, where red is biased and green is impartial.

Navigating the repository

To understand how Chrome extensions work, see the documentation and this helpful blog. I will describe the important parts of the architecture of this Chrome extension in the following:

• page_scraper.js is a content script that runs when the page is loaded. It has access to the DOM of the current tab and waits for a message from popup.js. When it receives the message that the user has clicked on the "Analyze news" button in the popup, this script will get the paragraph elements of the page and strip the links and other HTML elements so only the main text of the article remains. It sends this back to popup.js.
• popup.js sends a message to the page_scraper.js to get the content from the page and makes an XMLHttpRequest to Fakebox with the content. The script processes the API response, editing popup.html to display information on the source and bias of the content.

Troubleshooting

In the test directory, test_fakebox.html and test_fakebox.js are designed to test to make sure Fakebox is up and running. If you open test_fakebox.html and see the word "nothing" on the page, Fakebox is not working. If you see the word "something", Fakebox is working. The script test_fakebox.js sends an XMLHttpRequest to Fakebox with article content and should change "nothing" to the JSON response in test_fakebox.html.