[627] in ad-lib
B.Index query
daemon@ATHENA.MIT.EDU (sbyrd@MIT.EDU)
Wed May 17 17:24:40 1995
From: sbyrd@MIT.EDU
To: opac-lib@MIT.EDU
Cc: ad-cat@MIT.EDU, gyoung@MIT.EDU, ganderso@MIT.EDU
Date: Wed, 17 May 1995 17:24:23 EDT
Hello fellow Opaccers:
The final B.Index frontier is about to be crossed. As part of the
B.Index changes we requested, we asked that 3 numeric fields be added
to the PUBNUM index: 027 (tech report number), 088 (tech report
number) and 074 (GPO item number). When the changes were made, they
were indexed differently from the numbers already in the PUBNUM index
(086--gov doc number, and 028 subfield a--music publisher number).
(All of these searches, by the way, are alphanumeric, even though we
refer to them as number searches.)
In the interest of not wanting to drive our patrons (and staff) any
crazier than they already will be after Advance goes live, I've been
conferring with Nina, Eric, and the Ad-cat group about ways to make
all searches using the PUBNUM index consistent. (By the way, be on
the lookout for my followup issue/question about search labels.)
Please take a look at what I will be proposing to Michel Ridgeway
tomorrow afternoon and let me know if you have any questions or other
suggestions by Thursday afternoon. I apologize for the short
turnaround time, but since it's a change we're asking Geac to make to
the B.Index I want to make sure we give them as much time as possible
before the 2nd GMA.
Thanks,
Sam
--- ----------------------------------------------------
The following searches will be indexed in the PUBNUM index after the
2nd GMA:
027 tech report number indexed as is
028 subfield a music publisher number indexed as all caps
074 GPO item number indexed as is
086 Gov doc number indexed as all caps
088 Tech report number indexed as is
These are all alphanumeric fields, containing numbers and letters
that may appear in the record in either lower or upper case, or both.
"Indexed as is" means the search has to match the way the number
appears in the record. If the number has upper case letters, the
search has to have upper case letters in order to retrieve the
number. "Indexed as all caps" means that no matter what case the
letters are in in the record, the search has to be in all upper case.
The problem is that they are inconsistent. Originally, only 028 and
086 were in the pubnum index, and when we found out the search
had to be in all caps, we were bummed but resigned. Then Michel
Ridgeway added 027, 074, and 088 to the pubnum index, but he
added them "as is" so that the search has to match exactly the way it
appears in the record. This made us realize that the items in the
pubnum index could be indexed differently, and if there's a
possibility of changing them , we should ask at the very least that they all be
consistent. I would go further and say that I think they should all
be one case or the other, upper or lower, because we can't expect the
searcher to know in which case the letters appear in the record.
So here's what I propose asking Michel:
We would like every search in the Pubnum index to be consistent.
This means 027, 028, 074, 086, 088. Here are our priorities for how
they should be indexed:
1. TOP PRIORITY: every search in the Pubnum index should be
case-insensitive. In other words, index them so they are retrievable
whether searched by upper case, lower case, or any combination.
2. If this is not possible, index them with indexing normalized to
lower case (because this wouldn't require the searcher to hit an extra
key, i.e. the shift key or the caps lock key).
3. Lowest priority: index them with indexing normalized to upper
case.
Regardless of how they're indexed, our main concern is that every
search in the Pubnum index be done the same way. Otherwise it's
inherently confusing to the user.