- From: John S. Erickson <john.erickson@hp.com>
- Date: Thu, 26 Jun 2003 11:33:35 -0400
- To: <www-rdf-dspace@w3.org>
Farrukh sez: > An approach that side steps the hair splitting over what > is content and what is metadata embedded in the content, > is to use pluggable content cataloging. Content cataloging > works similar to database indexing. When content of a specific > type is submitted, one or more plugable content specific > cataloging services are invoked which automatically generate > metadata from content in a content specific manner. Such > services could also use knowledge external to the content > such as context knowledge. > > The result is the metadata can be generated in a standard > form in an automated manner directly from content... JSE: The technique that Farrukh refers to can also be thought of as applying a "template": for an artifact of a particular "type", instantiate a record of pre-defined structure as specified by the template (or schema). This might include both manual and automatically-generated data elements, the later (perhaps) derived from some transformation of the submitted artifact. Clearly, the distinction between content and metadata is an application-level concern; this must be up to the interpreting application. A given data element might be useful as "metadata" to one class of interpreter, but as "content" to another... This also ties back to the observations about application profiles made earlier in the week... John
Received on Thursday, 26 June 2003 11:36:59 UTC