Fake News Detector

Know what’s real. A simple tool to check facts and filter out fake news.

Comment

Inspiration

I was inspired to create the Fake News Detector after the recent Air India crash. While browsing Google, I came across many conflicting news articles, each giving different reasons for the crash. It became difficult to identify what was true and what was misinformation. This experience made me realize the need for a tool that could help people verify news quickly and easily, so I built this project to solve that problem.

What it does

Once we go to the website we first have to sign in using an email account. If you do not have an account created then you can sign up. This was done using the Auth0. After signing in you are able to upload a screenshot, image, or pdf of the article. Then once you press analyse the bot will analyse the image and give a reault whether it is Real or Fake. Not only that but we have also integrated an AI Chatbot. This can be accessed by clicking the bot button on the bottom right of the screen. In there you can chat to an AI bot built using Gemini API.

How we built it

  1. The Authentication (Login and Logout): Handled using Auth0
  2. Chatbot: Developed using Gemini API
  3. Fake News Bot: A Custom bot which is trained to recognize Fake News using Fake news and real news datasets. The datasets were aquired from Kaggle (https://www.kaggle.com/datasets/clmentbisaillon/fake-and-real-news-dataset)

Challenges we ran into

We ran into multiple challanges:

  1. One of the biggest hurdles was integrating Auth0 into the main website.
  2. Another issue was with the chatbot; it initially wouldn’t respond due to a small bug in the code.
  3. The most difficult part was building the custom bot, as I had never created a model like this before. It required a lot of trial, error, and learning.

Accomplishments that we're proud of

An accomplishment that I am proud of is getting the Custom bot to have an accuracy of 99.8%. For a first-time attempt at making a classifier bot, this was a huge personal milestone.

What we learned

This project taught me a lot of new skills. I had never worked with Auth0, Gemini API, or developed a custom fake news detection bot before. The concept of this project was 100% new for me and was a fun learning experience.

What's next for Fake News Detector

Moving forward, I plan to:

  1. Make the bot self learning. Right now the bot is trained on specific news articles (20,000+). In the future I plan to make the model teach it self from time to time from different sources. This will make sure the bot stays updated about future news.
  2. Visualize data from the custom bot to identify trends and patterns in misinformation over time.
  3. Build a mobile app version of Fake News Detector and release it on the Play Store to make it more accessible.

Built With

Share this project:

Updates