Don’t lose your preferences!

Sign-in or sign-up to save your personalization preferences to return to later
FeedChevron right imageData ScienceChevron right imageAI text classifier for WikipediaChevron right image
Post main image

AI text classifier for Wikipedia

Differentiate real articles from AI generated text

I built this web app with Wikipedia in mind because, with the rise of AI, it is easier than ever to spread false information to the masses on Wikipedia, with strong technology comes the need for strong counter-technology! I used GPT2 to generate 1,000 fake articles after training it on 1,000 real Wikipedia articles. Next, I used a Deep Neural Network to classify the real articles from the fake ones. Then I used the Flask framework to build this web app which I deployed to Heroku. The current model has 94% accuracy overall and was better at identifying the fake articles, the model is slightly overfit. I am currently working to fix this by increasing the training data and looking into a cloud option.

Video demo
Comment post

Skip straight to final coding interviews with
Remote Companies!

Top Makers

Most active posts

Top projects