What is required to make this sort of addition useful is people who are
willing and able to add the information. The rare materials cataloging
community is used to doing this sort of thing, both for the local catalog
and to existing records in "OCLC." I would suspect that some music
libraries also go to the trouble of adding these terms. I suspect that
enabling searching on such terms is a matter of including the subfield in
the index.
No search gets you "all" the resources in a catalog; it simply gets you
the resources that have been indexed, so, e.g., articles in a Festschrift
are frequently not retrieved. Unless you have normalized or established
access points, the search will not pull up "everything." So, the problem
of incompleteness should not deter us from the attempt. Besides, as
Ranganathan pointed out, the future is longer than the past. Until the
apocalypse, of course.
Laurence S. Creider
Special Collections Librarian
New Mexico State University
Las Cruces, NM 88003
Work: 575-646-7227
Fax: 575-646-7477
lcrei...@lib.nmsu.edu
On Wed, 27 Jan 2010, Jonathan Rochkind wrote:
I don't know about large proprietary systems, but allowing searching like this
is quite possible in various open source 'next generation' discovery layers,
for instance those baesd on the open source text indexing package Solr.
In the implementation of a discovery layer I'm working on right now, it
would be easy to allow limits based on "relationships" like this -- show
me "all the resources in a catalog where Clint Eastwood is an actor
rather than a director (or where he both acts and directs) "
But after considering such a feature, I realized it wouldn't' work,
because a tiny minority of our data (our marc records) have this info.
So it would kind of be a lie: We _can't_ show you ALL of the resources
in a catalog where Clint Eastwood is an actor. We can come close to
showing you all of the resources where Clint Eastwood has been mentioned
as a "related person" (in a 700 or 100). But how many items in the
catalog where Clint Eastwood IS indeed an actor, will not only have
Clint Eastwood as a 700, but ALSO have the a relator code in there? Very
very few. So it would be misleading to suggest that the system can show
you this.
This is perhaps why few have done this, even in systems that make it possible.
I still think it's much more useful, even in the present environment, for _display_,
rather than index/searching. It is rather odd that, for most of our records, even if
they have a bunch of 700 name entries, all the system can say is that all of these
entries are "related people". What does that mean to a user? It would be much
better, even just on display, if we could say that Clint Eastwood is an actor in it, and
some other person is the director. For index/search on these relators to work well
without being misleading, you need a certain critical mass of records that have the data
filled out. But for display on an individual record, every little bit helps, every record
that has relator codes is one more record with a more useful display.
To me, it seems even more clear that this is important with name-title entires. It seems especially weird
that, of a 700 name-title, all the system can display is that it's a "related work." A
"related person" is _usually_, but not always, some kind of a contributor. But a related work can
be almost anything. That the system is only capable of listing "related works" without telling the
user in what way they are related seems a serious flaw to me; whether used as an access point or as a
display, how is the user to know why they got a hit on the title, or why the title is being listed as
'related'?
It may also be worth experimenting with using relator codes in "partiioned" or "facetted"
results. Okay, my system can't really promise to show you ALL of the records where Clint Eastwood is an
actor. But what if you first search for Clint Eastwood, and THEN, of the 300 results, the system can tell
you that in 50 of them he's an actor, in 10 of them he's a director, and, sadly, in the other 240, all we can
say is he's a "related person". So the system isn't erroneously telling you it can find ALL the
records where he's an actor, but it is telling you what it can about your results. That might be another way
to deal with the fact that this data, while potentially useful, can be misleading if used inappropriately
since it exists in such a small part of most of our corpuses. (corpii?). As far as I know, nobody has
experimented much with this yet, but I think it would be worth investigating if someone had the
time/resources to.
Jonathan
________________________________________
From: Resource Description and Access / Resource Description and Access
[rd...@listserv.lac-bac.gc.ca] On Behalf Of Adam L. Schiff
[asch...@u.washington.edu]
Sent: Wednesday, January 27, 2010 7:39 PM
To: RDA-L@LISTSERV.LAC-BAC.GC.CA
Subject: Re: [RDA-L] Use of $4 and $e for 700's
On Wed, 27 Jan 2010, J. McRee Elrod wrote:
I see no need for either $4 or $e after 100's, 700's ...
As I tried to explain earlier, if we encode the specific relationships, we
are able (system-willing) to do more refined searches, such as looking for
all the resources in a catalog where Clint Eastwood is an actor rather
than a director (or where he both acts and directs) or where Leonard
Bernstein is the conductor rather than the composer. There's no way to
make these kinds of searches or limits (and I'm not aware of a system that
does this either - does anyone know of one?) if we don't encode these
kinds of relationships in a way that a machine can act on them. Maybe
it's too much extra work for us to do given the staffing most of us have,
but I for one think that some users would appreciate being able to search
at this level of granularity.
Adam
**************************************
* Adam L. Schiff *
* Principal Cataloger *
* University of Washington Libraries *
* Box 352900 *
* Seattle, WA 98195-2900 *
* (206) 543-8409 *
* (206) 685-8782 fax *
* asch...@u.washington.edu *
**************************************