Visualizing public discourse through user comments on the New York Times website with natural language processing and sentiment analysis.

User comments visualized geographically, colorized by the ratio of positive to negative words

As public discourse surrounding journalism moves online, conversations scale rapidly. When a single article on the New York Times website inspires thousands of comments, it is impossible to reveal the collective voice contained within. We conceived and built VoxPop as a tool for abstracting online discourse, and visualizing sentiment through time and space.

User comments visualized over time
Detail view showing comments from a specific location, and which words have been tagged positive or negative.

By siphoning reader comments from the New York Times API through a custom Natural Language Processing toolchain, VoxPop classifies positively and negatively charged words, and then abstracts the underlying reader sentiment across hundreds of thousands of comments in popular topics of conversation.

VoxPop allows viewers to explore reader sentiment geographically and chronologically through an open, standards compliant, web application.


2012 Adobe Design Achievement Award (Winner)

SXSW Interaction Award (Finalist)