November 30, 2010 Vol. 3, Issue 11
Princeton computer scientists have developed a new way to pinpoint the origins of new ideas and how they spread.
A new computer algorithm developed by computer scientists at Princeton University takes a different approach to identifying the influence and spread of ideas.
David Blei, an assistant professor in the Computer Science department at Princeton, and his post-doctoral student, Sean Garrish, sought to be able to track the origin of an idea to a single document. While Google uses visitor traffic to rank pages and scholarly journals use citations to measure paper impact, the Princeton duo came up with an algorithm that focuses on how language within documents changes over time.
They put their algorithm to the test on three scientific journals: Nature, Proceedings of the National Academy of Sciences, and Association for Computational Linguistics Anthology. When compared, the two methods agreed about 40 percent of the time. However, there were some cases where frequently cited papers did not have much impact on the language used in the field, and other instances when high-impact papers were not often cited.
The new idea-mapping method is not meant to replace other techniques, said the researchers. It is simply another method for measuring where ideas get started and how they spread throughout documents. Blei and Garrish are continuing their work to see if they can find patterns in the way language changes over time that will eventually lead to a capability to predict the next big idea.
