MONO91
NIST Monograph 91: Automatic Indexing: A State-of-the-Art Report
Indexes Compiled by Machine
chapter
Mary Elizabeth Stevens
National Bureau of Standards
of today, there are at least five or six instances of citation indexes that have been pro-
duced, several different experimental investigations are under way, and new interest
has been generated by the considerations of the Weinberg panel. Thus:
"Of the newer approaches to the indexing of scientific documents, the Weinberg
Panel was particularly impressed with the citation index as a promising biblio-
graphy tool. In order to learn more about this approach, the National Science
Foundation is currently sponsoring the compilation and publication of extensive
citation indexes for the fields of genetics and also for statistics and probability;
and is supporting two kinds of experiments to evaluate different techniques for
using citation data in indexes and searching systems in the field of physics." ![OCRerr]
In general, the principle of citation indexing is based upon the hypothesis that the
bibliographic references cited by an author provide significant clues to the subject content
of the author's own paper and/or that there is a certain commonality in subject between
papers that cite the same references or that are co-cited. 2/ The principle can be applied
to the compilation of bibliographical or indexing tools in several different ways. First,
there is the method of citedness, which groups for a given item the identifications of sub-
sequent items that have cited it. The converse of this is, of course, the bibliography or
reference list of a given item. 3/ In the first case, we are concerned with `1descendants,"
and in the list of references with "ancestors". 4/
1/
Committee on Scientific Information, 1963, [l35[OCRerr], p. 16.
2/
Compare Adair, 1955, [z[OCRerr], p. 32, with respect to Shepard's Citations itself:
![OCRerr]Since all of the cases listed under a given case have cited it, it follows that
they must all be, more or less, pertinent to the case cited." See also Kessler,
1963, [32o[OCRerr], p. 1: `1This method ... originated in the hypothesis that the biblio-
graphy of technical papers is one way by which the author can indicate the
intellectual environment within which he operates, and if two papers show similar
bibliographies there is an implied relation between them."
See Salton1 [OCRerr]96Z, [5203, p.III-3: "A citation index consists of a set of biblio-
graphic references (the set of `cited1 documents), each being followed by a
list of all those documents (the `citing' documents) which include the given
cited document as a reference. A citation index is to be distinguished from a
reference index which lists all cited documents under each citing document."
See, for example, Tukey, 1962, [611], p.5: "Any user's greatest need is
likely to be for access to the latest information rather than to the oldest, but
the latest items are children, not ancestors. Genealogy is important, but
progress requires tracing descendants lung and Vandeputte, 1960, [2913, p.11,
make a similar distinction between "histoire" (antecedents) and "filiation"
(successors).
29
3/
4/