Words are Wind: Sentiment Data Visualization

Data visualizations can help us to see things in ways we otherwise wouldn’t, and to make connections we might not see without the correlation mapped out. In a beta visualization that draws from script dialogue and other sources, Wind and Words explores, analyzes, and visualizes Game of Thrones characters’ vocabularies, direct interactions, and attitudes during those interactions. In other words, they use sentiment data.

Character sentiment data visualization is particularly interesting in the context of B2B applications of buyer intent data. Buyer intent signals can correlate company / user behavior with a likelihood of making a purchase, and the correlations are based off many data points: which topics are actively researched, which content is consumed, historical purchase and tech install information, and at which point in the buyer’s journey a user is.

In the Wind and Words beta, conceived and built by Impossible Bureau, characters’ vocabularies, instances of dialogue, and their sentiment is tracked over time, letting the user predict whether a given character will succeed or fail in the Game of Thrones.

Hint: You’ll want to use the directional arrows on your keyboard to navigate through the Wind and Words visualization, and click with your mouse.

Similarly, to predict whether a prospect will ultimately result in a win or loss, Aberdeen tracks buyer intent over time, using machine learning and natural language processing to parse 12 billion web pages and 480,000 actively researched and targeted keywords.

Whereas the Wind and Words beta tracks all the dialogue, all the din (check out the data tabs for each season; In season 1, each word averaged a 7.555 Scrabble score, there were 3,679 spoken determinative words, and there were an average of 24.5 profanities per episode!), all the vocabulary, and all the sentiment behind each spoken line (can you believe Ned Stark had only one positive sentiment the entire first season?), Aberdeen’s massive sets of buyer intent data include only the active research within actual buyer journeys, and intent signals are captured from within the noise of normal Internet activity.

This creates the most accurate way to predict which signals are coming from someone who’s in-market for a purchase.

And while Aberdeen’s machine learning-powered, intent data-capturing tech cannot accurately tell us who will win and who will die in the Game of Thrones, it can offer up to 91% accuracy in predicting purchase intent, as shown in blind tests run by clients.