Menu

Home / portfolio / graduate thesis

My Sister in Data: Visualizing an online relationship

Advisors: Ali Qadeer and Isabel Meirelles

My Sister in Data is a two-year graduate thesis project to fulfill the requirements of the Masters of Design in Digital Futures at OCAD University in Toronto. It is a data visualization entirely composed of the messages between my sister and I on Facebook, ranging from the first message sent to me by my sister Zoya in 2010, to the 14,256 messages sent between the two of us in 2018. Facebook Messenger was chosen because it is the platform we use most often to communicate. The project uses D3.js, a JavaScript library for data visualization, and Tableau, a visualization prototyping software, in order to explore and visualize pre-collected message history data that I had downloaded from my personal Facebook account.
Chat messages are seen as ephemeral, and yet this project makes Facebook’s data hyperpublic, exposing seemingly fleeting dialogue and displaying it in a completely different context. This visualization intentionally exposes collected personal data to an audience as a way to reclaim the content that has been posted to an internet platform, where personal data is debatably not solely our own property anymore. In some ways, this is an empowerment of identity and relationships online, as well as an attempt to understand a rapidly changing social world and my own personal relationship with my sister throughout time. Each Facebook message was assigned an emotional value through sentiment analysis in the scripting language, R. Every word in a message was compared to the Tidy Text library, which categorizes certain words into specific emotional groupings. In addition to running the sentiment analysis in R, I went through the dataset manually in order to ensure that every message’s sentiment value was accurately assigned. In the visualization, the messages are shown over time. Every message is stacked according to the month it was sent, creating a dense column for each month, colored by each message’s sentiment value. The legend at the top of the visualization is interactive, allowing the user to toggle off or on messages categorized by that emotion. Hovering over a message reveals its contents.


image
image
image
image
image