SOME BASIC COMMENTS ON RETRIEVAL TESTING
| Pages | 267-270 |
| Date | 01 April 1965 |
| DOI | https://doi.org/10.1108/eb026376 |
| Published date | 01 April 1965 |
| Author | ROBERT A. FAIRTHORNE |
| Subject Matter | Information & knowledge management,Library & information science |
SOME BASIC COMMENTS ON RETRIEVAL TESTING*
ROBERT A. FAIRTHORNE
Herner and
Company,
Washington,
DC, USA
TO TEST is not to evaluate. Tests reveal to what extend a retrieval system
performs in some specific way; what value is or should he put upon such
performance
is
another matter altogether. Clearly one should test only such
performance
as is
necessary to the aim of the system, because
one
cannot test
everything. Thus we must first decide what is the aim of the system, then
test
to
find how well it
achieves
this
aim.
These matters can
be
examined and
discussed usefully without going into profound matters such
as
man's place
in the universe, the nature of unrecorded knowledge, the library's place in
next year's budget, or whether some individual
user's
knowledge
is
changed
to his individual satisfaction.
Even to define the aim
is
difficult, because we deal with essentially social
systems. We can lay down necessary, but not sufficient conditions of per-
formance. What, for
instance,
are the conditions sufficient to inform some-
body about something? Even if known they would take us well out of the
field of 'Information Sciences' into psychology and beyond it into the per-
sonal psychical structure and history of each individual user of the system.
Sufficient conditions are out of the question; we must concentrate on
neces-
sary ones. As our knowledge grows, so will the conditions known to be
necessary.
We certainly know that before anyone can be informed, he must
be told. Before users can be informed by documents they must be told or
notified about them. This
is
not sufficient, but it is certainly necessary.
DOCUMENTARY ACTIVITIES IN DOCUMENTARY TERMS
To separate aim and achievement from the values of the aims and achieve-
ments, we must consider them in terms of documentation, not of the ex-
ternal applications or
intentions.
For example, the same documentary tech-
niques can be used to look up such expressions as those denoting the square
root of thirteen, the melting point of tungsten, the destination and intended
time of departure of an aircraft, or the last line of a limerick. These may be
demanded for many reasons or
non-reasons.
They can be looked up or re-
* This paper is sponsored by the Information Sciences Directorate, US Air Force Office of
Scientific Research, under Contract No. AF 49(638)-1427.
267
To continue reading
Request your trial