Tikalon Blog by Dev Gualtieri

Sleeping Beauties of Science

July 13, 2015

Most people associate the Sleeping Beauty tale with the 1959 Disney film, but this folk tale goes back at least as far as the early 14th century. It was published as the far less romantic, Histoire de Troïlus et de Zellandine, in the 1528 book, Perceforest. In that version, which is certainly not a Disney Princess story, the beauty is raped while in her deep coma, and she delivers a child without awakening. Music lovers might enjoy Tchaikovsky's score of the The Sleeping Beauty ballet,[1] but I prefer The Nutcracker.[2]

Gustave Dore illustration of Sleeping Beauty, 1897

Detail of an illustration of Sleeping Beauty by Gustave Doré (1832–1883).

This 1867 illustration is from the book, Les Contes de Perrault.

(Via Wikimedia Commons. Click for larger image.)

Although "sleeping beauties of science" might be a reference to the older professors who nod-off during seminars, a research team from the Center for Complex Networks and Systems Research of the School of Informatics and Computing, Indiana University (Bloomington, Indiana), have a different definition. The sleeping beauties that they write about in a recent paper in the Proceedings of the National Academy of Sciences are published research papers that remain dormant for years, then later are found to be of great importance.[3-4]

Generally, a scientific paper will generate the most interest when it's first published, and then its citation rate will decline over a few years, finally reaching zero for most papers. As I wrote in an earlier article (Modeling Scientific Citation, December 16, 2011), this was demonstrated by physicists from the University of Fribourg (Fribourg, Switzerland) in an analysis of 450,000 papers published from 1893 to 2009 in American Physical Society journals , as shown below.[5]

Time decay of paper relevance, which is a function of its citations in 91-day intervals, as a function of time. The data are grouped into three sets, as defined in ref. 5. This is a simplified version of fig. 1 of ref. 5.

(Via the arXiv Preprint Server).[5]

Since searching and retrieving older articles is now easier in our Internet age, older articles are being cited more often. As I wrote in an earlier article (Citation in the Electronic Age, December 15, 2014), a 2008 study published in the Journal of the American Society for Information Science and Technology found that the age of cited articles has increased continuously since the mid-1960s.[6]

A study by a Google, Inc. research team confirmed this finding.[7-8] It found, also, that 21% of citations were to articles 15 years old and older in 2013, an increase of 30% over the 1990 level. Thirteen percent of citations were to articles 20 years old or older, an increase of 36% over the 1990 level.[7] In 2013, 36% of citations were for articles at least 10 years old, up from 28% in 1990.[7]

In contrast to citation decay of most papers, a sleeping beauty paper has a long period in which it is rarely cited, followed by a sudden surge in its popularity.[3] The term, "sleeping beauty," was coined by the Dutch statistician, Anthony F.J. van Raan, in 2004, and anecdotal evidence suggested that there were just a few of these. In physics, the Einstein, Podolsky, and Rosen (EPR) "paradox" paper is one of these.[3] EPR was published in 1935, but its citations bloomed in 1994. In statistics, Karl Pearson's 1901 paper in Philosophical Magazine wasn't much cited until 2002.[4]

Aside from a few examples such as these, sleeping beauties were thought to be rare. The Indiana University study, however, examined a very large dataset - 380,000 publications from the American Physical Society and 22.4 million from the Web of Science.[3-4] The dataset covered multiple disciplines in the span of more than a century.[4]

A "beauty coefficient" was calculated for each paper in this dataset. The value of this coefficient depended on a comparison of a paper's citation history to others based on publication year, the maximum number of citations received in a year, and the year when maximum citation was achieved. Analysis found that sleeping beauties are not that rare, as shown in the figure.[3-4]

The top twenty disciplines producing sleeping beauties in science.

(Indiana University image. Click for larger image.)[4]

There's a continuous spectrum of sleeping beauty behavior with a power-law behavior, thereby suggesting a common mechanism working at all scales.[3] Among the interesting statistics of the phenomenon, four of the top fifteen sleeping beauties were published more than a century ago.[4] Physics, chemistry, multidisciplinary science, mathematics, and general and internal medicine were the disciplines with the highest frequency of sleeping beauty behavior, and several papers in these fields weren't noticed for up to 70 years.[4] The Proceedings of the National Academy of Sciences, where these results were published, Nature, and Science were the top journals publishing sleeping beauties.[4]

Future developments in science determine whether or not a paper is destined for sleeping beauty stardom. Alessandro Flammini, an associate professor of informatics at Indiana University and the corresponding author of the study, notes the following:[4]

"The potential application of some studies are simply unforeseen at the time'... The second-ranked sleeping beauty in our study, published in 1958, concerns the preparation of graphite oxide, which much later became a compound used to produce graphene, a material hundreds of time more resistant than steel and therefore of great interest to industry."

"This study provides empirical evidence that a paper can truly be 'ahead of its time... A 'premature' topic may fail to attract attention even when it is introduced by authors who have already established a strong scientific reputation."

Alessandro Flammini of Indiana University

Indiana University's Alessandro Flammini, the corresponding author of the "Sleeping Beauty" study.

(Indiana University photo.)[4]

Statistics, a discipline that's become more relevant with the availability of large computerized datasets, has had a surge in its number of sleeping beauties. Along with Pearson's hundred year paper, there's one by Edin Bidwell Wilson that was rediscovered after seventy years.[4] Many sleeping beauties have attained prominence when their results have been found to be applicable to other fields.[4]

The authors write that their study provides evidence against the use of short-term citation metrics for deciding a paper's scientific impact.[3] This study was supported by the National Science Foundation.[4]

Sleeping Beauties of Science

References: