Frank A. Roos wrote:
>Consider the following "chain" of factors that influence the finding of
>relevant information on the Web. These factors are all related to
>"quality".
>
>[lots of excellent stuff]
>
>Therefore. Will all DC elements likely be indexed? Will qualifiers be
>indexed, or what is their role in the indexing process?
The view from Reuters :-)
There are two kinds of DC metadata: codes and free text. The following
elements can, at least in controlled environments, be treated as codes:
DATE, TYPE, FORMAT, IDENTIFIER, LANGUAGE, COVERAGE, RELATION, RIGHTS,
SOURCE. The following elements are more likely to be treated as free text:
TITLE, CREATOR, SUBJECT, DESCRIPTION, PUBLISHER, CONTRIBUTOR.
By "controlled environments", I mean environments where the metadata is
inserted, and/or validated, by software.
In the case of codes, the role of qualifiers is to inform the indexing
engine how to index the data. If we have, say, three recognised coding
schemes for LANGUAGE, the designers of indexing engines will choose the
internal format they prefer, and will use mapping tables to transcode the
other formats to the internal format.
This approach is relevant also to text transcoding. Increasingly, designers
are choosing to store all text in Unicode and to map to/from Unicode on the
boundary. This is done, for instance, by Java Virtual Machines and by
Windows NT.
>[more good stuff]
Misha
|