Fwd: RE: RE: [SAC-FAST] Re: SAC Subcommittee on FAST

Tue Jan 25 19:29:29 EST 2005

One tricky problem for converting LCSH dates into a single period FAST 
heading will be with records containing $y's following two different kinds 
of topical subdivision. For example, the LCSH periodizations for $x History 
and $x Politics and government and $x Intellectual life under a country 
name are often not identical, yet they do fairly often appear together on 
records. Likewise, periodizations vary for different countries, and the 
record for a comparative study will often contain both. In such cases, the 
narrower range is more likely to be accurate for the piece, but it sounds 
like the FAST algorithm will come up with the most encompassing period. Is 
the FAST conversion algorithm able to distinguish between sequential spans 
and overlapping spans? If so, it might be worth trying to send these down 
two different conversion pathways.

The question which reviewing the sample records won't be able to answer is 
how successful searching is in a file which combines LCSH derived periods 
and item-specific periods. It seems inevitable that there will be a loss of 
consistency and reliability in a mixed file of machine and manually derived 
FAST headings--but maybe not to the point of user frustration. We can 
consider as a subcase of the problem of applying a rule of specificity. It 
sounds like there is a rule of specificity envisioned for FAST at least in 
the case of chronology which is not embodied in the authorities themselves. 
I'd be interested in hearing more about FAST rules of application and how 
they would be documented and promulgated.

Stephen

>X-From_: sac-fast-bounces at mailman.yale.edu Tue Jan 25 17:24:24 2005
>X-Umn-Remote-Mta: [N] mtain-c.tc.umn.edu [160.94.128.20] #+LO+NM
>X-Umn-Remote-Mta: [N] studentcomputing.its.yale.edu [130.132.52.22] 
>#+NR+UF+CU (A,-)
>X-Umn-Report-As-Spam: 
><http://umn.edu/mc/s?DE@EBhFBxpx5I7u6$JsPEEg@C6lf@DIEqEWluUtanZJWDHQQ$x2vS7mB$TQRtTZi3wnH22dKLcqo>
>X-Original-To: SAC-FAST at mailman.yale.edu
>Delivered-To: sac-fast_mm at studentcomputing.its.yale.edu
>Subject: RE: RE: [SAC-FAST] Re: SAC Subcommittee on FAST
>Date: Tue, 25 Jan 2005 17:44:07 -0500
>X-MS-Has-Attach:
>X-MS-TNEF-Correlator:
>Thread-Topic: RE: [SAC-FAST] Re: SAC Subcommittee on FAST
>Thread-Index: AcUDK4BzgpjFNaIvSp63nGlkbZ3R6gAABzfQ
>From: "O'Neill,Ed" <oneill at OCLC.ORG>
>To: "Sherman Clarke" <sherman.clarke at nyu.edu>
>X-YaleITSMailFilter: Version 1.1e (attachment(s) not renamed)
>X-Mailman-Approved-At: Tue, 25 Jan 2005 18:22:30 -0500
>Cc: "Dyer, Anya" <dyera at OCLC.ORG>,
>     "Kammerer, Kerre" <kammerer at OCLC.ORG>,
>     SAC-FAST at mailman.yale.edu,
>     "Vizine-Goetz,  Diane" <vizine at OCLC.ORG>,
>     "Dean, Becky" <deanr at OCLC.ORG>,
>     "Childress, Eric" <childree at OCLC.ORG>
>X-BeenThere: sac-fast at mailman.yale.edu
>X-Mailman-Version: 2.1.5
>List-Id: ALA SAC Subcommittee on FAST Project <sac-fast.mailman.yale.edu>
>List-Unsubscribe: <http://mailman.yale.edu/mailman/listinfo/sac-fast>,
>         <mailto:sac-fast-request at mailman.yale.edu?subject=unsubscribe>
>List-Archive: <http://mailman.yale.edu/pipermail/sac-fast>
>List-Post: <mailto:sac-fast at mailman.yale.edu>
>List-Help: <mailto:sac-fast-request at mailman.yale.edu?subject=help>
>List-Subscribe: <http://mailman.yale.edu/mailman/listinfo/sac-fast>,
>         <mailto:sac-fast-request at mailman.yale.edu?subject=subscribe>
>Sender: sac-fast-bounces at mailman.yale.edu
>
>Sherman,
>
>We are still working our through some of these questions--Tomorrow's
>answer may be different than today's.  That said, other members of the
>FAST may differ from my answer.
>
>For manually assigned headings, the actual date "1985-1995" would be
>used.  Because of the faceting, it is very difficult to pick the dates
>from the deconstructed LCSH period subfields.
>
>Regarding the second question, when converting from LCSH, we convert to
>the shortest period covering both periods, 1800-1929 for the Picasso
>example. "1900-1919" is included even thought that period in not
>relevant. When manually assigned, it would be up to the cataloger to
>decide if both periods are important. The good news is that these
>examples are very rare--the bad news is that FAST doesn't deal with them
>well.  We made an intentional decision to keep FAST simple which works
>well most of the time, but not always.  Having a single numeric date
>range does simply retrieval.
>
>Thanks, Ed
>
>-----Original Message-----
>From: Sherman Clarke [mailto:sherman.clarke at nyu.edu]
>Sent: Tuesday, January 25, 2005 5:16 PM
>To: O'Neill,Ed
>Cc: SAC-FAST at mailman.yale.edu
>Subject: Re: RE: [SAC-FAST] Re: SAC Subcommittee on FAST
>
>Ed,
>If the book on rock music covers 1985 to 1995, would the "right" FAST
>chronological heading be "1985-1995"? Or do you have to pick from
>deconstructed $y subfields?
>
>Is the restriction on one period heading related to a particular topic
>or to the whole record? Let's say our item is about 19th century
>African sculpture and the influence on Picasso in the 1920s. Or a
>comparison of the prehistoric Venus of Willendorf and modern feminist
>renditions of the earth mother?
>
>Thanks.
>Sherman
>
>----- Original Message -----
>From: "O'Neill,Ed" <oneill at oclc.org>
>Date: Tuesday, January 25, 2005 3:40 pm
>Subject: RE: [SAC-FAST] Re: SAC Subcommittee on FAST
>
> > Arlene is correct, FAST normally uses exact periods.  However,
> > when the
> > periods are derived from LCSH, that (usually longer) period is used
> > since more specific dates are not available.  In addition, FAST only
> > permits one period headings per record, so that multiple LCSH periods
> > are combined.  For example (LCCN: 2000-344708) the LCSH headings:
> >
> >       Rock Music $y 1981-1990
> > and   Rock Music $y 1991-2000
> >
> > convert to the single FAST period
> >
> >       1981 - 2000
> >
> > Multiple periods in LSCH are relatively rare and almost always
> > representeither contiguous or overlapping periods.
> >
> > Ed
> >
> > -----Original Message-----
> > From: sac-fast-bounces at mailman.yale.edu
> > [mailto:sac-fast-bounces at mailman.yale.edu] On Behalf Of Arlene Taylor
> > Sent: Tuesday, January 25, 2005 11:46 AM
> > To: Sherman Clarke
> > Cc: SAC-FAST at mailman.yale.edu
> > Subject: Re: [SAC-FAST] Re: SAC Subcommittee on FAST
> >
> > It is my understanding that FAST *does* use exact time periods.
> > That'swhy I wrote what I did.  The examples we'll be looking at
> > that Ed
> > creates
> > for us at OCLC will deconstruct LCSH into FAST headings.
> > Therefore, the
> > chronological headings will be LCSH time periods.  But if a person
> > created
> > FAST headings for the item from scratch, the time periods would be
> > exactcoverage of the item being described.  So, I'm asking if we
> > can evaluate
> > this part of the sample that Ed is going to create for us.
> >
> > --Arlene
> >
> > On Tue, 25 Jan 2005, Sherman Clarke wrote:
> >
> > > > 6.If there are chronological headings, do the time periods
> > > > make sense for the item being described?  (Given the way these
> > > > examples will be created,
> > > > i.e., deconstructing LCSH, the chronological headings will
> > > > represent LCSH time periods, not the exact period covered by
> > > > the item as FAST is intended to do; so I'm not sure what we
> > > > can evaluate here.)
> > >
> > > Since the chronological headings could not be reconstructed with
> > their
> > > topic, we should perhaps consider recommending that FAST would
> > fit
> > > better if they moved to giving the exact period covered by the
> > item,
> > > rather than fussing with the pre-existing chronological
> > subdivisions.
> > > This does invite more variety in chronological subdivisions and
> > it
> > > would be good to have indexing that would be smart about time
> > periods.
> > > For example, it would find an item with "1913-1922" if you
> > searched
> > > 1910s or 1920s or 20th century or pre-1945.
> > >
> > > Sherman
> > >
>_______________________________________________
>SAC-FAST mailing list
>SAC-FAST at mailman.yale.edu
>http://mailman.yale.edu/mailman/listinfo/sac-fast

****************************************************
Stephen Hearn
Authority Control Coordinator
Database Management Section Head
University of Minnesota
160 Wilson Library               Voice: 612-625-2328
309 19th Avenue South              Fax: 612-625-3428
Minneapolis, MN 55455      E-mail: s-hear at tc.umn.edu 