Promise and pitfalls of extending Google's PageRank algorithm to citation networks.

The misuse of journal impact factor in hiring and promotion decisions is a growing concern. This article is one in a series of invited commentaries in which authors discuss this problem and consider alternative measures of an individual's impact. The number of citations is the most commonly used metric for quantifying the importance of scientific publications. However , we all have anecdotal experiences that citations alone do not characterize the importance of a publication. Some of the shortcomings of using citations as a universal measure of importance include the following. (1) It ignores the importance of citing papers: a citation from an obscure paper is given the same weight as a citation from a groundbreaking and highly cited work. (2) The number of citations is ill suited to compare the impact of papers from different scientific fields. Due to factors such as size of a field and disparate citation practices, the average number of citations per paper varies widely between disciplines. An average paper is cited ϳ6 times in life sciences, 3 times in physics, and Ͻ1 times in mathematics. (3) Many groundbreaking older articles are modestly cited due to a smaller scientific community when they were published. Furthermore, publications on significant discoveries often stop accruing citations once their results are incorporated into textbooks. Thus, citations consistently underestimate the importance of influential old papers. These and related shortcomings of citation numbers are partially obviated by Google's PageRank algorithm (Brin and Page, 1998). As we shall discuss, PageRank gives higher weight to publications that are cited by important papers and also weights citations more highly from papers with few references. Because of these attributes , PageRank readily identifies a large number of scientific " gems " : modestly cited articles that contain groundbreaking results. In a recent study (Chen et al., 2007), we applied Google's PageRank to the citation network of the premier American Physical Society (APS) family of physics journals (Physical Review A–E, Physical Review Letters, Reviews of Modern Physics, and Physical Review Special Topics). Our study was based on all 353,268 articles published in APS journals since their inception in 1893 until June 2003 that have at least one citation from within this dataset. This set of articles has been cited a total of 3,110,839 times by other APS publications. Our study is restricted to internal citations— citations to APS articles from other APS articles. Other studies (Bollen et …