Citation index vs Google Page Rank

(from physorg.com, vía robotwisdom.com):

"Researchers Patrick Chen and Sidney Redner at Boston University, along with their colleagues Huafeng Xie and Sergei Maslov at Brookhaven National Lab, recently applied the PageRank algorithm to all 353,268 articles published by the Physical Review between 1893 and 2003. It comes as no surprise that on average, GPR correlates nicely with the citation index. More interesting are the outliers—those articles that somehow achieve a high ranking with relatively few incoming references.

After applying PageRank, Chen et al. sorted the papers in this network by their GPR values. Their recent article provides a sampling of famous papers from the top hundred results. Number 85, with only three citations, is a startling poster child of this new approach! The paper in question is a classic example of delayed influence. While it was the first to present a model which today sees widespread use, its result was refined and popularized by other researchers in a separate article. The “child paper” has accumulated 680 citations but makes only ten references to other works itself. The original paper thus collects a large share of its child’s impressive impact.

Nor is this the only example! Among the papers with over a hundred citations, most of the papers with an unusually high GPR are easily recognizable as seminal works. Such works compare favorably in overall influence with the very small population having over a thousand citations.

While “influence” may be easy to measure crudely, it is hard to measure reliably. These results show that although the two methods are comparable, Google’s PageRank algorithm seems to identify important scientific papers more reliably than a simple citation index.

If there is a lesson here, it is this: in giving due credit, one should not be short-cited! "

Paper by Chen, Redner, Xie, Maslov