This project is about using sentiment analysis and Blender, a 3d modelling program, to visualise the negativity of Twitter. The goal is to excavate society’s shadows through social media. What does that mean? Well, following Carl Jung’s idea of the Shadow, I use “shadows” to refer to the sides of ourselves that we’re unaware of more often than not, the sides we refuse to accept because we judge them as negative or “bad”. This concept can also be applied collectively, and in this project I try to give a face to this collective shadow, the dark sides of society.
To do so, I learned to access the Twitter API and wrote a script to obtain the current top 50 trending topics in my local area, the city of Edinburgh, and find 10 popular tweets in each of these topics.
The script then performs sentiment analysis on these tweets to obtain negativity and positivity scores, and aggregates other data about them, like retweets, likes, length, etc., and builds 500 cubes in Blender, each of which has properties tied to the attributes of the corresponding tweet: size according to popularity, colour according to negativity, etc.
This can then be rendered into an image, an animation or even a VR scene to make the visualisation more impactful.
generated on April 24th, 2022 at 3:11 PM
Here, each cube represents a tweet. White, luminous cubes are tweets classified as positive by the VADER algorithm, while purple cubes are the ones classified as negative – the shadows that are the focus of this project.
The size of each cube is proportional to its popularity: number of retweets + number of likes.
I really liked doing this project because it allowed me to use together very different skills I had obtained in varied contexts in a new way. Mainly, I was able to take advantage of my coding skills in a new way: as a medium to produce a provocative piece, rather than just as a way of showcasing my work or a tool to solve analytical problems. Furthermore, it allowed me to learn three things that I was interested in: using the Twitter API, applying sentiment analysis, and visualising data in 3D environments using code.