Inspiration
With the rise of the internet, and online news channels, misinformation has also skyrocketed, and sometimes, it is hard to tell what you can believe. Being able to trust the information you see online is highly important when forming a social and political view.
What it does
We wanted to create a solution which can verify information for you without any user input. We made a chrome extension that pops up when misinformation is detected in a YouTube video, and upon clicking the user is directed to another tab which has a nodal network containing all of the mis informative statements made during the video. These are ranked in terms of severity and truthfulness. These can then be interacted with to find out the sources related to each statement, and the chain of events which link each source
How we built it
Frontend: We used Flask and React (in python) along with Typescript, HTML, and CSS to create the website Backend: We scrape the YouTube video transcript with timestamp data We then extract each statement made by the speaker so we can analyse them We crawl data off Wikipedia, and use LLM,s Wikipedia and Google Fact Check API to determine whether the statement is a fact. We then check the security level for the statements made by the LLM Then, we can make a DAG with each node showing a specific article, and we can create a misinformation aggregator using text-embedding and K-means to cluster the statements
Challenges we ran into
Linking frontend and backend was quite difficult as organisation is always hectic in a hackathon. We also had some trouble near the end with finalising the website as we were quite time pressured since our project was very ambitious
Accomplishments that we're proud of
We created a novel idea which has highly important applications
What we learned
Collaborating under time pressure and working in high-stress environments.
What's next for fAIk
Larger-scale implementation and deployment to more users
Log in or sign up for Devpost to join the conversation.