273
DisasterMan
We can now search easily using author's name, organizational affi liation,
journal's name, and year of publication.
However, the increasing volume of scientifi c and technical literature
has provided an impetus for more automatic error detection procedures to
supplement the traditional human error detection and correction routines.
The question to be considered is whether or not the combination of human
and computer error detection systems is now able to cope with the volume
of scientifi c literature.
Errors in Data Bases
For purposes of discussion, some of the illustrations of types of
errors will be drawn from the Science Citation Index. This should not be
misconstrued as being an attempt to publicize any presumed shortcomings
of this database. On the contrary, the Institute for Scientifi c Information
has already taken extraordinary steps to correct errors in its databases.
As Sher (1) pointed out in a symposium on error control in chemical
literature during a meeting of the American Chemical Society in 1966, the
data found in Index Chemicus are sometimes more accurate than in the
original article from which the abstract was prepared. The error detecting
procedures apparently included recalculation of molecular formulas by
chemical abstracters who then requested the original author to confi rm
corrected errors.
We must also keep in mind that there may be different orders of
importance of errors. Dr. Cawkell (2) classifi ed errors in Science Citation
Index into two major classes. A class one error would be one in which the
result is that an item is very unlikely to be retrieved in consequence. A
radical misspelling of an author's name might be an example of a class
one error. A class two error would be of the kind which will usually not result
in retrieval loss. For example, a non-standard abbreviation of a journal title
might be a class two error since the cited item would appear beneath the
correct cited author, usually in juxtaposition to the same item correctly
cited (always assuming that the item has been cited more than once).
Errors in Primary Literature
Some errors originate with the author. For example, an erroneous
reference or mathematical error will be published if it is not noticed by
the referees and the editor. When the error is subsequently detected, an
erratum may be published. If a reader detects the author's error, a reader's
Summary :
The error detecting procedures apparently included recalculation of molecular formulas by chemical abstracters who then requested the original author to confi rm corrected errors. For example, a non-standard abbreviation of a journal title might be a class two error since the cited item would appear beneath the correct cited author, usually in juxtaposition to the same item correctly cited (always assuming that the item has been cited more than once).
Tags :
errors,cited,item,class,literature,which,chemical,example,author,index,authors,two,scientifi