As a network scientist, Janosov Milán likes to apply the analogies of his profession in the most unexpected contexts. And so, in addition to contemporary art, brain science, and urbanism, he has also taken a scientific look at the popular TV series Game of Thrones.
In the project Network Science Predicts Who Dies Next in Game of Thrones, Milan used predictive modeling and machine learning methods to predict which key characters would die in the seasons of the series.
“Game of Thrones is a complex world in which social position and true friendships play an important role, so I used network science to quantify the social interaction patterns of each character,” says Milan. He built on the subtitled version of the series, which includes not only the spoken text but also the names of the speakers, and paid attention to the separation of scenes. By combining the two pieces of information, he extracted a list of the characters who appear in each scene, from which he constructed a social map of Westeros. Every major character is represented by a network node, and two characters are linked if they appear in the same scene. He then marked who had died in the first six seasons. Finally, he applied a widely used linear model to predict which of the still-living characters were likely to meet their faith. A more detailed description of the project and the methodology, as well as a spoiler conclusion, can be found on Research Gate.