News moods

Data visualisation @Parsons MFA DT,
supervised by Daniel Sauter, spring 2015

News moods is an interactive data visualization that aims to exploit the non-objective, flavored and emotional side of journalism, with a hope to oversee social mood swings in the 20th century.

The dataset consists of news article headlines from New York Times that include the word "progress" starting the year 1900 till 2013. The headlines are then run through a sentiment analysis to evaluate, word by word, either they are overall positive, negative or neutral (based on AFINN). Macro view displays a yearly histogram of these values while micro-view helps to discover the actual news headlines, and the way positive or negative keywords appear within them.

Interactive exploration of yearly news article headlines

Macro histogram view

Micro view. "Good" news

"Bad" news


The data was collected from New York Times API through Node.js, and later processed using AFINN-based sentiment analysis module by Andrew Sliwinski. Total size of text-based data taken from NYT is 340Mb. Interactive visualization was made with p5.js.