Abt HA. The Institute for Scientific Information and the Science Citation Index" Organizations and strategies in Astronomy, 4(296):197-204, 2003

[Eugene Garfield]

E-mail Address - Helmut Abt  : abt at noao.edu

Author  :  Abt, HA
           Kitt Peak National Observatory
           P.O. Box 26732, Tucson, AZ 86726-6732

Title   :  The Institute for Scientific Information and the
           Science Citation  Index

Source  :  Organizations and Strategies in Astronomy, 4(296):197-204, 2003.
           A. Heck (editor)

Publisher: Kluwer Academic Publishers, PO Box 17, 3300 AA Dordrecht,
           The Netherlands



Abstract:
The history, goals, and capabilities of the Science Citation Index, which is
produced by the Institute for Scientific Information, are reviewed.  Recent
statistics and data about astronomy are given.  We address questions such as
the importance of self-citations, whether citations really measure
importance, and whether citations in different fields can be fairly compared.

Introduction
The only quantitative method that we have to judge the importance or
usefulness of scientific papers is to count citations (references) to them.
 People have worried about whether a high citation count invariably
indicates an important paper, giving, as an example, counts that depend upon
many self-citations.  Several people have explored the statistical
significance of self-citations.  For instance, Trimble (1986) states: "About
15% of all citations in astronomical papers published durig January 1983
were self-citations, in the sense that the cited and citing papers had at
least one author in common... [It] varies surprisingly little among
journals, countries, subdisciplines, and epochs".  Considering that citation
counts vary by two or three orders of magnitude between excellent and
unimportant ones, an error of 15% is not significant.  Others (e.g. Garfield
1955) were concerned whether incorrect papers might also receive more
citations than they deserve.  That has rarely proven to be true.

By studying citations to a set of 53 papers judged by senior astronomers to
be the most important ones of the 20th century, Abt (2000) found that 92% of
them had an average citation count 5 (after 1950) to 11 (before 1950) times
the average and average half-lives 2.5 times the average.  He concluded that
high citation counts almost invariably implies importance or usefulness.

As early as 1955 Eugene Garfield (1955) considered the idea of producing a
multi-disciplinary citation index for science.  The idea was not new.  As
early as 1873 lawyers had a legal "citator" covering case law.  Shepard's
Citations was published, first in New York and then after World War II moved
to Colorado Springs, CO.  Shepard's provides brief summaries of court cases
in all states and lists all subsequent cases that refer to them.  It did not
originally cover legal literature.  However, producing a citator for science
was a far greater undertaking.  Consider that in 1998 the printed Science
Citation Index (SCI) included 16,780,146 citations to 770,591 papers in 3542
journals, and was printed on 38,422 pages for the main index alone.  It also
included citations to thousands of books and patents.

There are several reasons for producing a citation index.  Its primary
purpose is to help conduct literature searches that are highly specific and
independent of terminology.  The Science Citation Index spawned the field of
scientometrics - charting the growth of sub-disciplines relative to one
another.  The journal Scientometrics (now published by Kluwer Academic
Publishers - see Schubert 2001) was launched in 1978.  Shortly after SCI was
launched, it was recognized that it could be used to evaluate the importance
and impact of (1) an individual paper, (2) a scientist's set of papers, (3)
the papers from a given institution, (4) various journals, and (5) to select
the most important papers (CitationClassics) in any field of research.