Trumpalyser is a new piece of artificial intelligence designed to read the tweets of Donald Trump and his Twitter followers to determine what they are talking about and their attitude towards the topics. It is based on two separate models, one for named entity recognition and the other for sentiment analysis. These two models are large, deep learning neural networks trained on large datasets of pre-tagged tweets. There is also an embedded layer to transform the words from the tweets into vectors using the word2vec structure. The tweets are streamed in real time through to the models and pushed to the front-end of the website using websockets so that no refreshing is needed for the sentiment to change live as the user is viewing it.
There is a large graph at the top of the website to show the sentiment of Trump and his followers over time, this aims to show any changes in the popularity of Trump through one of his favourite mediums, Twitter!
I was lucky enough to be able to be a large part in building this as a part of BrightMinded.
The project is currently live here