The Geotags web app intelligently combines related topics of discussion across multiple social media sources to provide a navigable map of trends across the United States.
We use Tweepy and PRAW to query and aggregate posts made to Twitter and Reddit respectively, storing information about geolocation and content.
We pipe data to Mallet, which uses Hierarchical Latent Dirichlet Allocation and Pachinko Allocation, to compute a topic distribution of the data interpolated with the United States.
We query a Postgres Database with a PHP backend to receive our topic data, then display it in a React/Redux web app featuring an interactive AmCharts chloropleth map.