Factify

Fallacious text with explanations
Fallacious text with score
Youtube URL transcription text
Good text with light fallacies

Make bias, fallacies and misinformation visible

What it does

It takes audio, video, text and Youtube URLs, and identifies the logical mishaps in the words. It looks for hidden biases, fallacies and rhetorical tactics in the text and it highlights them to the user. Depending on the severity of the error, the brightness of the highlight changes. The color of the highlight varies due to the mistake it is pointing out and if the same words are deemed to be making two different kinds of errors, the colors mix. Factify gives you a summary of the mistakes by grading 4 aspects: Neutrality, Soundness, Transparency, Verifiability. The higher the score, the more logical sound and likely to be true the text is. Furthermore, you can switch to the findings tab which will give you a mistake by mistake rundown of the text and will explain what is wrong with the passage. It allows users to upload multiple files at once and they are saved in the history so that they may be referred to later. There is also a chrome extension version of the webapp. The extension allows user to highlight text and to "Factify" it. The extension highlights the mistakes on the page and shows the user what was wrong with the sentences. Both the extension and the webapp allow the user to toggle through the three types of mistakes so that they can single out a type or two.

How we built it

Teamwork makes the dream work

Challenges we ran into

Initially, the highlighting was static meaning that if the same words had two different kinds of mistakes, the colors would overlap with the dominant one showing. To allow the color mixing, we had to refactor all of the code related to the highlighting and the color logic Uploading files in a batch used to give weird state errors just also caused a headache Yt-dlp was very hard to get running because yt blocks the IP addresses of datacenters and once we deployed our website on render, it was basically impossible, given the time limit, to get it working Google developer program for the extension was very complicated and hard to navigate for no reason