------- Forwarded Message

Return-Path: [log in to unmask]
Received: from elmer.harvard.edu (elmer.harvard.edu [128.103.151.248]) by 
elmer.Harvard.EDU (8.6.12/8.6.12) with SMTP id QAA18054 for 
<[log in to unmask]>; Thu, 10 Apr 1997 16:57:55 -0400
Date: Thu, 10 Apr 1997 16:57:55 -0400 (EDT)
From: "R. Wendler" <[log in to unmask]>
Reply-To: [log in to unmask]
To: [log in to unmask]
Subject: Re: Coverage-Element - working draft for discussion 
In-Reply-To: <[log in to unmask]>
Message-Id: <[log in to unmask]>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

Interesting work, folks! Comments below...

On Wed, 9 Apr 1997 [log in to unmask] wrote:

> a. There is some feeling that Coverage should be used only for what might be 
> termed non-fiction works - for example, a scanned map of Montana - and not for 
> such works as a novel that takes place in Montana. It would be difficult to 
> explain this difference to Webpage users, and I suggest we keep Coverage for 
any 
> sort of spatial or temporal coverage, rather than say, sometimes it's Coverage 
> and sometimes it's Subject or Keyword.
> Or, to put it another way, that Coverage be used only for certain aspects.  
For 
> example - a ceramic pot; should Coverage - Spatil refer to where the pot were 
> made, or where it were found?  I suggest that since it is a repeatable field, 
> one may have a Coverage field for each of these.
> b. There is a viewpoint that Coverage is a type of Subject or Keyword and 
should 
> be subsumed under that element.  This is a topic that catalogers in libraries 
> have been arguing about for years - that is, whether geographic area is a 
> subject or not.  It is included in subject headings; but it is dealt with as a 
> separate type of heading from thematic headings. For example, in USMARC, 
> thematic terms (e.g., Geology; Anthrpology; etc.), geographic terms, and 
> chronologic terms are each given different field codes; these are usually $x 
for 
> thematic, $z for geographic, and $y for chronologic headings.

It is true that the distinction between topical and geographical 
subjects in preserved in MARC data.  However, at least in the U.S., 
library system functionality rarely makes use of that distinction. 
While they could easily allow users to make this distinction in 
searching, the vast majority of libraries do not do so.  What about 
libraries outside the U.S.?  (Or does someone in the U.S. disagree with 
the observation?)

In Canberra a lot of disparate opinions were on view about when 
the subject element should be used vs. when coverage should be used; 
if we can't keep it straight, how do we expect a searcher to? 
The very word "coverage" may contribute to my unease -- a book may 
be said to "cover" the women's suffrage movement, may it not?
I am (I think...) sympathetic to the subject camp for certain cases:

1) There is conceptual difference between what a thing *is* and 
what a thing is *about*.  For example, data *is* collected during 
a given interval; the data represents (*is*) the climate of North 
Africa.  This pot *is* from 16th century China whereas that novel 
is *about* an imaginary pub in Islington.  Coverage could be seen 
as an *is* element, so to speak.  (Yeah, yeah -- it's not always 
clear, but you understand my point, I hope.)  It may be a disservice 
to lump those things together for retrieval.  

2)  If I understand correctly, DC was originally conceived to improve 
retrieval of network resources across domains rather than just within 
domains, but it seems unlikely that the wide range of types, 
formats, and intentions of the information in the coverage element 
can be meaningful if the qualifiers cannot be interpreted. 
(Canberra Qualifiers Rule, yes?)  

This state of affairs might be fine if we were dealing with textual 
data, where you won't search Timbuktu and mistakenly retrieve the 
Gilded Age, but when the data can be numeric values coded according 
to a wide variety of schemes, the data seems meaningless without 
qualification.  

For that reason I'd love to isolate textual *subjects* which have 
geographic or temporal components from the admittedly critically 
important coded data about capture or representation needed for 
sophisticated geo-spatial retrieval.  Unlike other fields where, 
when you can't interpret the qualifier, you may continue to process the 
data, many of us may be reduced to simply dropping any coverage field 
whose qualifier is unknown on the assumption that without a specialized 
retrieval environment, it will be noise.  

Some geographic and temporal information will be put in the subject 
element no matter how this is decided.  Consider Library of Congress 
Subject Heading Strings, which are composed of topical, geographic, 
chronological, and form sub-elements which themselves may be less useful 
if decomposed.  If the Dublin Core guidelines instruct the use of 
subject for topics and coverage for temporal/geographic data, you'd see 
things like:

   ELEMENT=subject CONTENT=Scholars, Muslim--Mali--Tombouctou--Biography.
   ELEMENT=coverage CONTENT=Tombouctou (Mali)--Biography.
  
So you'd have to index them together or risk giving people only half the 
story.

(As usual, I'm coming from a metadata-import/use rather than a 
metadata-creation perspective.)

Hi to the Canberrites!  Tans faded yet?

Robin Wendler  ........................     work  (617) 495-3724	
Office for Information Systems  .......     fax   (617) 495-0491
Harvard University Library  ...........     [log in to unmask]
Cambridge, MA, USA 02138  .............


------- End of Forwarded Message