Metadata proliferation

There has been a recent debate, on the www-international list, regarding the 
relationship between:

   1.   <... LANG=xxx> language tagging of HTML as per RFC 2070,
   2.   <META HTTP-EQUIV="Content-Language" Content="xxx">.

To these could be added the Dublin Core Metadata element LANGUAGE:

  3.    <META NAME = "DC.language" CONTENT = "...">

This example highlights the proliferation of various kinds of Web-related 
metadata.  This comes in at least three flavours (yet another is provided by 
PICS):

   1.   HTML-based
   2.   HTTP/META HTTP-EQUIV-based
   3.   META-based

These flavours seem to be very loosely coupled.  I am trying to understand the 
uses and relative priorities of the various flavours.  What should happen when 
they disagree?

The last two flavours are designed to exist both inside and outside an HTML 
document.  Would it be sensible to argue that where the HTML flavour is present, 
the other two should be absent/ignored?

In what directions do we want Web-related metadata to evolve?  The requirements 
driving the various flavours are different; how much do they have in common?

A note for readers not familiar with the Dublin Core, taken from 
draft-kunze-dc-00.txt:

   The ... motivation ... [is to] ... improve the prospects for resource 
   discovery on the Web.  Specifically, the goal [is] to identify a simple 
   set of common description elements that authors (or content managers) 
   could embed in their documents to promote their discovery ...  The term 
   "Dublin Core" applies to this simple core of descriptive elements.

The fifteen Dublin Core Metadata elements are: TITLE, CREATOR, SUBJECT, 
DESCRIPTION, PUBLISHER, CONTRIBUTOR, DATE, TYPE, FORMAT, IDENTIFIER, SOURCE, 
LANGUAGE, RELATION, COVERAGE and RIGHTS.

Misha

Received on Saturday, 8 March 1997 10:20:21 UTC