KNOWLEDGE CREATION BY INFORMATION RETRIEVAL
| Date | 01 April 1990 |
| Published date | 01 April 1990 |
| Pages | 365-367 |
| DOI | https://doi.org/10.1108/eb026867 |
| Author | DAVID ROBERTS |
| Subject Matter | Information & knowledge management,Library & information science |
DOCUMENTATION NOTE
KNOWLEDGE CREATION BY INFORMATION RETRIEVAL
DAVID ROBERTS
Medical Information Service
British Library Document
Supply Centre,
Boston
Spa, West
Yorkshire,
LS23 7BQ
DAVIES' INTERESTING REVIEW of the possibilities and methods of
deriving new knowledge from bibliographic databases has implications for
key factors in the design of such databases
[1].
These factors include coverage
and retrieval.
A recent letter published in Lancet referred to the apparently not
uncommon assumption that to search Medline is to search the entire medical
journal literature [2]. It was pointed out that Medline covers at most about
40%
of this literature. Even when including a second major
medical
database,
Excerpta Medica, coverage
is
far from
complete.
There may
be
a considerable
quantity of worthwhile medical journal literature not covered in any publicly
available information database or service. Some areas, for example the
literature of the professions allied to medicine, are particularly neglected, as
some recent and pending articles have shown
[3,4,5,6].
This selective coverage
of the literature by bibliographic databases is likely to increase the difficulties
of deriving new knowledge from published information and to limit the
possibilities.
Turning to information retrieval this has of course been extensively studied
[7].
As Ellis has observed however [8], progress has been disappointing
particularly with respect to practical recommendations for the design of new
and ongoing
systems.
Specifically on retrieval and indexing, the literature has
implied a fairly widely accepted consensus (amongst researchers) on the
respective merits of searching by free text or title words and by assigned
subject headings. This is that the former is equal to or even superior to the
latter and that consequently in the age of computer systems the necessity for
complex and expensive indexing is past. This consensus seems to have
originated in or at least been strongly influenced by the studies at Cranfield [9]
and of Medlars [10] though Ellis has suggested that the relationship between
these studies has been
widely
misinterpreted
[8].
Whatever
the
truth of
this,
the
consensus is contradicted by the practices of database producers and the
experiences of commercial hosts and search service intermediaries. Many of
the fifty or so online databases in medical subjects were originated without
assigned subject indexing
systems.
Almost all have subsequently adopted such
Journal
of
Documentation,
Vol.
46, No. 4, December 1990, pp. 365-367.
365
To continue reading
Request your trial