‘Map of Science,” reflects the new bit-sized world

-A A +A
By Roger Snodgrass

A team of cyber-scientists at Los Alamos National Laboratory has found a novel approach for measuring the global scientific enterprise, bit by bit.

Johan Bollen of LANL’s Mathematical Modeling and Analysis Group led the investigation, published this week as “Clickstream Data Yields High Resolution Maps of Science” in the current issue of the Public Library of Science (PLoS).

Bollen said the findings may change the way science is funded.

Current measures of scientific activity and influence rely heavily on an outdated system of citations and a more subjective interpretation of press clippings and public impact, but that is about to change.

After compiling records of nearly 1 billion online interactions from a broad sampling of scholarly web portals, Bollen believes that the patterning and analysis of these Internet sessions that he calls “clickstream data” provides a more representative picture of what is actually happening in the scientific world.

“It will help funding agencies and policy makers make better decisions about science,” Bollen said.

While citations have a kind of scriptural authority, Bollen compares the practice of making inferences from scholarly citation in various specialized journals to formative charts used by early explorers. There is historical value to be gained, but citations in particular tend to be outdated by the time a particular influence can be observed.

He compares it to astronomers looking at a distant galaxy, which can only be seen as they were millions of years ago, because that’s how long it took the light to travel through the intervening space.

“When we look at science, we’re going to be faced with significant publication delays,” he said. “The paradigm we should be looking at is not the traditional print publications. We live in a digital environment.”

The informational landscape has changed rapidly in the last 15 years under the accelerating influence of the World Wide Web.

Everywhere in the world, researchers and scholars have access to indexed, searchable lodes of expanding content.

Lee L. Zial, lead program director of the National Science Digital Library Program, has called this a fundamental change in the relationship between people and knowledge.

“The democratization of access to data and information has altered not just the ‘where’ and ‘when’ of learning but increasingly the ‘how’ and ‘by whom’ that authority or certification of expertise has been granted,” he wrote in a study for the Council for Information and Library Resources last year.

“Science nowadays is becoming much more interdisciplinary,” Bollen said.

His research suggests that the citational criteria has an inherent bias toward the natural science and underestimates the role of the humanities and social sciences in the creative process.”

Also, the scientific process communicates in many more ways than formal papers. There are also databases, algorithms, data sets, operating systems, sound clips, multi-media and many other kinds of informational content that wouldn’t necessarily be found in the footnotes or bibliography of a paper.

As he collects more data and finds new ways to analyze it, Bollen said he is also working on prediction.

“As usage accumulates in certain areas, like biodiversity and ecology, these journals swell; they are read and visited a lot, an indication that the scientific community is converging on those subjects,” he said. “Scientific interest is like attention.”

He is developing algorithms that will project shifts in momentum and accumulations of interests. Although the published map is a snapshot and a composite of two year’s activity, he said he was working hard to be able to spot bursts of activity, perhaps even in real time.

The research team includes Bollen, Herbert Van de Sompel, Ryan Chute, and Lyudmila Balakireva of LANL’s Digital Library Research and Prototyping Team and Aric Hagberg, Luis Bettencourt and Marko A Rodriguez of LANL’s Mathematical Modeling and Analysis Group, and LANL's Center for Nonlinear Studies. Bettencourt also is part of the Santa Fe Institute.