Nature:
The number of citations a paper receives is a common measuring
stick for evaluating the worth of the paper. Dashun Wang of the
IBM Thomas J. Watson Research Center in New York and
international collaborators have developed a model, based on
data from the first five years after a paper was
published, that accurately predicts how many citations it
will receive in its lifetime. The model ignores the author,
subject, and journal of publication. Instead it considers
only three factors: The underlying appeal of the ideas;
how quickly the paper gains its first citations; and the fact
that over time, the number of new citations approaches zero.
Wang and his colleagues tested the model using physics papers
published in the 1960s. They found that 93.5% of the papers fit
the predicted range of citations 25 years after publication.
They found similar success over shorter durations using papers
from the 1990s. However, many papers showed a wide range of
uncertainty. The remaining 6.5% of papers were slow to receive
their initial citations but then became significant more than 5
years after publication. Wang hopes to refine the model with
the inclusion of other factors such as topic or location of
publication.
