CRANV1P1
ASLIB Cranfield Research Project: Factors Determining the Performance of Indexing Systems: VOLUME 1. Design, Part 1. Text
Formation of Index Languages
chapter
Cyril Cleverdon
Jack Mills
Michael Keen
Cranfield
An investigation supported by a grant to Aslib by the National Science Foundation.
Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government.
- 62 -
The use of quasi-synonyms* to enlarge classes is a permissive device; since
the final, expanded classes do not totally exclude the continued separate use of the
terms confounded (as is the case with real synonyms) no figure showing the exact
degree of vocabulary reduction is possible. For example, the expanded class for
Bow is Bow + Bowing + Ahead + Front + Forward + Forebody; the expanded class for
Ahead is Ahead + Forward + Upstream. Clearly, the expansion of Bow does not
result in the obliteration of the separate class 'Ahead', which not only continues to
exist but is in turn expandable by the addition of other quasi-synonyms. In an index
language which confounds true synonyms only, the reduction is once and for all and
the terms no longer have a separate identity.
At this stage of our thinking about the function of vocabulary size as the main
determinant of recall and precision, it seemed desirable to have as exact a measure
of this parameter as possible. So the first testing of hierarchy took the form of a
fixed ( and therefore accurately measureable} reduction in vocabulary size.
Hierarchical reduction
The two methods of measuring hierarchy as a recall device (i. e. , by obligatory,
block reductions in vocabulary size and by selective searching through different
hierarchical paths} are demonstrated below. The first example is one of a 'concept'
hierarchy - i.e., one not restricted to single terms and one place per term. This
is in order to show more clearly the two methods, and also to emphasize the distortion
which results from restriction to a 'one-place' hierarchy of single terms. The
latter results in the exclusion of some terms which are located in more general
categories and the result is seen in the second part of the example. A hierarchical
notation is attached to this example in order to make the permissive search clearer
(in the schedules actually used, notation was purely ordinal}.
'Concept' hierarchy demonstratin[OCRerr] hierarchical reduction
a
ab
aba
abaa
abab
abac
abad
abb
abba
abbb
abbc
abbd
abbe
abc
abca
abd
Experimental wind tunnel methods for investigating flow
Visualization methods
Using smoke, vapours, etc. (3}
Vapour screen (1)
Fog (1)
Wood smoke (1)
Oil smoke (1)
Using coatings, flows, etc. (3)
Oil flow (i)
Oil film (1 }
China clay (1)
Phosphorescent lacquer (1)
Ink flow (1)
Using spectrum (3)
X-ray spectrography (1)
Using Stroboscope (3)
[OCRerr],,Quasi-synonyms' are terms which can be used synonymously in certain contexts,
but which are not true synonyms. (see page 68}