Thursday, December 11, 2008

Map of science



This citation map of science (click for larger version) is from eigenfactor.org. It shows the importance of basic sciences like physics and molecular biology. Applied fields still depend on advances in basic science.

Eigenfactor score is a PageRank-like algorithm, in which citations are links. See here for more.

Orange circles represent fields, with larger, darker circles indicating larger field size as measured by Eigenfactor score™. Blue arrows represent citation flow between fields. An arrow from field A to field B indicates citation traffic from A to B, with larger, darker arrows indicating higher citation volume.

The map was creating using our information flow method for mapping large networks. Using data from Thomson Scientific's 2004 Journal Citation Reports (JCR), we partitioned 6,128 journals connected by 6,434,916 citations into 88 modules. For visual simplicity, we show only the most important links, namely those that a random surfer traverses at least once in 5000 steps, and the modules that are connected by these links.

13 comments:

Anonymous said...

the utility of the natural sciences may be measured by the employment prospects of its graduates.

to a first approximation science is applied engineering.

Anonymous said...

Partially off-topic: See this paper by Carl Bergstrom (U of WA), one of the people behind Eigenfactor.org.

Building trust by wasting time

Rather than try to justify the prevalence of other-regarding behavior in economic exchange by means of a model in which economic structure is jury-rigged on top of a system of values that were perhaps adaptive in smallgroup interpersonal interactions, we invert the causal chain. We argue that markets need not rest upon values that arose before them; instead, markets may create the values which allow them to function effectively. Markets are human constructions; in creating them, the participants engage in a process of mechanism design, selecting the rules of the strategic games in which they will be involved. These rule choices give rise to conventions of behavior — and where such conventions are granted normative force, they may appear to us as values.

Anonymous said...

Two observations:
1. Why operations research is not directly connected to prob. & statistics?
2. It'll be interesting if this study takes into account some discounting factor by the years the research has been done. This will probably highlight what are the fastest growing research area.

Anonymous said...

So you didn't correct for the tendencies for different fields to have different quantities of citations?

Carson C. Chow said...

Mathematics definitely shows why page rank can be misleading. No one cites Newton or Leibniz when they use calculus, etc. The math dot should at least be as large as physics.

Steve Hsu said...

Carson: I think using the Thomsen data is good for seeing the impact of *recent* research on other fields. Any recent math results used by, e.g., physicists, would be cited directly. No one would debate the importance of 18th or 19th century math to all of science, but even someone like me uses very little late 20th century math!

Ken: the links that are shown are ones that would have been traversed by a random surfer in 5000 hops. There are other links not shown, but the citation flow would be small.

Carson C. Chow said...

Precisely my point. The impact of math may not come for a century or more. Theories may come and go but a theorem is forever.

Anonymous said...

higher math is very seldom used in engineering practice in chemistry or biology. I'm an actuary and the exams are full of math, but in practice it is never used.

math's primary use is as a bona fides for academics. if most people can't understand you, you're set.

Anonymous said...

One big problem with this type of analysis is that it does not take into account sociological factors. Relevant factors which distort the graph include conventions about who/what gets cited, and the fact that good researchers in come fields publish many more papers than good researchers in other fields. The first factor is probably hard to quantify, but the second is not.

There are indeed some interesting connections missing; for example mathematics and biology, as well as those mentioned earlier. It would be interesting if these gaps persisted if books were included. I imagine that even late 20th century mathematics being used in other disciplines, for example, appears in lecture notes or other texts and thus the papers themselves are not being cited.

It would also be interesting to see a graph with the disciplines weighted (inversely, of course) by government funding levels. Can one quantitatively describe how much research in one field contributes to another in appropriate units?

Anonymous said...

Fail.

There's no dot for Materials Science, which should show strong citation links to physics, chemistry, and chemical engineering. They must have mis-categorized a dozen or two high impact journals.

Adam said...

If you hunt through the eigenfactor.org data, it looks like a number of major materials science and physical chemistry journals (Advanced Materials, Nano Letters, Langmuir, Materials Science and Engineering Reports) are actually being categorized as Physics.

Given the number of other dots there, I'm not sure why so many journals like this are lumped into physics--they contain some physics, but can be better categorized.

Anonymous said...

Also, where is linguistics? There's sociolgy, political science, the incredibly nebulous "education," but nothing about the science of spoken and written language?

payday loans online said...

Science advances our knowledge of this universal dimension and gives us opportunities to experience and learn and grow as individuals. Spirituality helps us individually with; who we eternally are, understand our purpose in life, and appreciate our chosen path to fulfillment here on planet Earth.

Blog Archive

Labels