Using Transformers and BERT to Predict Disasters from Twitter Feeds- Abhinav Angirekula - Jefferson High School, VA
This project utilizes Bert for natural disasters. It reads tweets, and from these tweets, determines whether that tweet subject matter was linked to natural disasters or not. The reason for this was pretty simple, because natural disasters are incredibly devastating, right. And so by reading social media feeds, and using them to determine whether specific posts are talking about a natural disaster, an earthquake, a tsunami, a snowstorm etc that lets us essentially create a sort of warning system. We use our model on live data, we can apply this model to live Twitter data. Then if the majority of these tweets seem to be talking about natural disasters, we can use our warning system and validate that there's likely a natural disaster happening right now.