Picture this: the whole of human knowledge as a figurative mind that can selectively focus on certain areas. It’s a profound notion, and visualizing such a construct is an enormous undertaking. But with last week’s release of a new “map of science,” a team of researchers led by Johan Bollen is attempting to do just that — with a high-resolution visualization of how scientific literature is accessed based on users’ downloading and browsing behavior, known as clickstream data. This usage data was collected, aggregated, and normalized across a wide variety of journal publishers and institutions. The result is a network map with color-coded nodes (clusters of research articles from different fields) and interconnected lines (shaped by users’ clickstreams), demonstrating the connections among a comprehensive sample space of scholarly research.
A new map of science based on clickstream data. Click to enlarge. Credit: PLoSOne
This isn’t the first attempt to extract meaning from the referential loops within scientific literature. In 2006, Columbia University’s W. Bradford Paley released an influential map of science based on data from Thomson Scientific, a firm that tracks article citations across scholarly journals. More recently, Carl Bergstrom, a biologist from the University of Washington, has developed a suite of innovative visualizations based on his own citation data sets for a venture called Eigenfactor. His method draws from network science and information theory to determine how often specific articles cite other articles as part of a relative ranking system for journals.