This is the repository for our storytelling project, which can be found here. This project was created for the TWEB 2017 class at the School of Business and Engineering Vaud. We decided to take a look at Donald Trump's Twitter and, using an archive of all his tweets, explore his social media presence.
All the source data used in this project was taken from the Trump Twitter Archive. We originally used a CSV format to download and store data, however a few tweets contained a '\r' character which adds a newline to the textbox containing the tweets and messes up the rest of the data. We therefore had to transition to the JSON format.
For the word cloud, we used a Java 8 program to stream every single tweet body and map each word to a number of occurences. We ignored pronouns and some articles since they would always be the most used words. Words with less than 3 occurences were removed from the map. The results are stored in data/wordoccurence.csv.
The first 250 words are used to generate the word cloud, with words increasing in size based on their number of occurences.
The interactive charts use the JSON data as is, with no modification. Using crossfilter, we can use dimensions to filter the data however we like. The charts are generated using dc.js and are inspired by the tutorials found here and here.
After modifying any of the Javascript files, you must run the browserify command to regenerate the bundle.js file, which is a bundle containing all of our scripts. To generate this bundle, use the command browserify js/main.js -o js/bundle.js from the root folder (in our case, the docs folder).