Home → News → Computational Linguists Predict Time's 'person of... → Full Text

Computational Linguists Predict Time's 'person of the Year' Using Computer Model

By Radboud University Nijmegen

December 12, 2013

[article image]

Computational linguists from Radboud University, the Meertens Institute, the University of Antwerp, and the University of Illinois at Urbana-Champaign have developed a quantitative model that predicted Time's "Person of the Year" for 2013 would be President Barack Obama.

Using only articles published by Time during the previous year, the model assessed the cultural impact of people in texts via the Learning to Rank technology that is frequently used by online search engines. The team tested the model's accuracy in forecasting Persons of the Year for 1923-2006, and found that the correct person was identified among the model's top 10 candidates 78 percent of the time. The correct person was ranked at the top of the list 29 percent of the time.

Following Obama, the model's top five, in order of rank, included Vladimir Putin, Miley Cyrus, George W. Bush, and Angelina Jolie. The researchers say the results demonstrate the growing role of computers in scientific research, particularly the digital humanities.

From Radboud University Nijmegen
View Full Article


Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA


No entries found