This project for EECE 5642 Data Visualization attempts to portray the important words in fake news articles to a user, allowing them to learn about the medium and avoid such news in the future. It does this using two main features: a convolutional neural network that operates on the words in the article to determine a news classification, and a Google Chrome extension that accesses this model, classifies text, and changes text sizes to identify which words/phrases the model finds important to the user.
The report for this project is included as Data_Vis_Final_Project_Report.pdf.
The presentation for this project is available HERE.
The structure for the neural network used can be seen below:
The model is exported in the model5/ directory, so that it can be imported into the Google Chrome extension.
The extension source is contained in the extension/src/ directory. To build the extension, navigate into extension/, then run yarn && yarn build. The resulting extension will be packed and available in the dist/ directory that is created. To use the extension, follow the instructions Here. In short - navigate to chrome://extensions, make sure the "Developer Mode" switch is turned on, and click Load Unpacked. Select the dist/ directory, and this will enable the extension in your browser~
To use the extension, navigate to a page you want to classify, and select the article's text. Then, right-click the text, and select the Classify News Text option.