While I agree with Stu about the value of consensus, I'm not sure that
there is a way to achieve it without a clear statement of what it is we are
agreeing to. In looking at the various messages that have come to the list
in the past few days it seems that the fundamental tension is between those
who want to keep the DC really simple and easy, and those who would like it
to be more robust and versatile. Each of these address the same basic
problem - the difficulties inherent in discovering relevant information -
but approaches it from a different perspective.
We could characterise these two perspectives as follows - and answer Carl's
questions for each:
1. The Dublin Core enables people with Web pages to describe them. A simple
set of fielded metadata elements will more precise searching than is
supported by the present generation of webcrawing search tools.
Document type: - HTML pages [referencing any digital media?]
Users: - uninitiated
Creation Method: - manual, or simple tools/forms
User Tools: - next generation web searching tools
Role in Process: - enables more efficient discovery across the breadth of
the web
Other metadata: - no relation
What value is placed on interoperability with other types of searching
tools by this group?
2. The Dublin Core has the potential to harmonise a broad range of existing
metadata standards, to support interdisciplinary use of what have been
created as subject-specific, or domain-specific resources.
Document type: - anything you can describe, digital or otherwise [perhaps
limited by the concepts of static or bounded resources explored at the
Dublin Image Meeting, which contrasted these with dynamic, changing
resources]
Users: - tool developers and professional cataloguers
Creation Method: - generated from existing data, or assigned by a specialist
- most likely a subset or mapping of existing metadata
User Tools: - both web searching and more 'sophisticated' tools
Role in Process: - supports generic information discovery
- not meant to meet specialised information discovery needs.
- not meant for descriptive cataloging.
Other metadata: - core subset that enables inter-operation across disciplines
- mappings should be made between existing standards and
the 'core'
If my characterisations are correct, the question then becomes in the first
instance whether these two sets of requirements are in conflict at the data
requirements level, or if can we can see a core set of information
categories that serve both needs - i.e. the basic Dublin Core.
jennifer
At 8:38 am 17/10/96, Carl Lagoze wrote:
>Perhaps I was too hastly in a previous message to express the need for a
>"requirements document". This starts to sound like a scarey engineering
>task that could occupy lots of time.
>
>I won't spend time coming up with a new name for this "document" - let me
>just state some of the requirements-like questions we need to answer
>before we proceed on the DC. This is not meant to be all inclusive nor
>are my possible answers meant to be all inclusive.
>
>- What are the types of resources we wish to describe with the DC -
>textual documents? HTML documents? images and text? all networked
>documents? simple (one per file) documents? complex (many files/many
>locations) documents?
>
>- What is the community of content providers that we are aiming at?
>professional catalogers? web-masters? casual authors?
>
>- How are these content providers/creators going to create Dublin Core
>records? manually? tools? second-party additions?
>
>- What are the types of tools do we expect will be "users" of the Dublin
>Core? Lycos/Alta-Vista like? on-line catalogues?
>
>- Where do we see the Dublin Core fitting in to the general Information
>Discovery/domain-specific searching/browsing/serendipity/retrieval
>process?
>
>- How does the Dublin Core relate to other metadata sets? Is it a
>summarization of other sets of metadata? Which sets of metadata does it
>summarize?
>
>Feel free to add to this list or argue with it. But we need to have firm
>answers to these questions to properly engineer this thing. With all due
>respects to Stu's consensus-building motivations, it simply is not
>sufficient as we try to really create something that is useful to the
>community and will have some lasting value.
>
>Carl
--------
J. Trant [log in to unmask]
Collections and Standards Development
Arts and Humanities Data Service +44 (0)171 873 5076 [phone]
King's College, The Strand, London WC2R 2LS +44 (0)171 873 5080 [fax]
http://www.kcl.ac.uk/projects/ahds/
--------
|