Hi,
I am following this discussion with keen interest. I am working on the
maintenance and development of one of bibliographic classification.
On Monday I attended The Essentials of Metadata and Taxonomy conference
(http://www.damusers.com/metadata/programme.html). Speakers were mainly
from corporate environment and very much concerned with the quality of
information delivery. Categorization and classification in this sector
is assumed to be common sense. Their customers do not approach system
*hoping* to get relevant results on the first page, they expect nothing
but. So it is understandable why they do not like to guess what users
mean by 'pitch' or 'deal' or 'management'.
Following discussion here - it seems that not many actually use
classification for semantic search expansion, controlling search terms,
knowledge browsing, cross vocabulary mapping, supporting term match in
automatic indexing, support program-assisted indexing, cross-language,
cross-repository information integration.
Could it be that the only way classification is envisaged to be used in
repositories is the one dating 30 years back - moving classification
notation from 6xx MARC to dc.subject - and hoping that someone will make
sense of it?
Objections to Dewey with cc 30,000 classes seem to be the lack of
specificity - while LCC cc 300,000 classes jolly old baroque giant seems
to be better suited. If one decides to use classification created for
shelf-arrangement of physical objects - where the only concern is the
systematic linear order of books - it is logical to think that there may
be some differences when it comes to digital collections. Especially if
nobody is actually planning to make any use of linear ordering in result
display or in online browsing.
If, however, there is a need to collocate resources that deal with the
same subject to a useful (and scalable!) level of granularity -
independently from many ways the subject may have been verbally
described by authors, indexers and users. Then maybe it would be worth
thinking how to implement classification so that this function can be
enabled without users and indexers being too much concerned with various
PE4178 or 811.111-1 etc.
But most importantly classification symbols differ from social tags - as
they also represent subjects' position in the knowledge area and in
hierarchy, in some systems they express relationships between different
subjects and type of these relationships.
IMHO it is not classification system that makes difference it is how one
implements it.
Regards
Aida Slavic
|