[562] in Public-Access_Computer_Systems_Forum
Third Generation OPACs
daemon@ATHENA.MIT.EDU (Public-Access Computer Systems For)
Tue Jun 23 09:51:33 1992
Date: Tue, 23 Jun 1992 08:46:42 CDT
Reply-To: Public-Access Computer Systems Forum <PACS-L%UHUPVM1.BITNET@mitvma.mit.edu>
From: Public-Access Computer Systems Forum <LIBPACS%UHUPVM1.BITNET@mitvma.mit.edu>
To: Multiple recipients of list PACS-L <PACS-L%UHUPVM1.BITNET@mitvma.mit.edu>
3 Messages, 144 Lines
*-----
From: dhoyman@fammed.wisc.edu (Dirk Herr-Hoyman)
Subject: Re: OPAC Functionality
>From: "Clyde W. Grotophorst" <WALLYG%fen1.gmu.edu@gmuvax2.gmu.edu>
>Subject: Re: OPAC Funtionality
>
>I'm ready to bet the next generation of OPACs will be virtual--
>they'll 'exist' only during a particular use and they won't look
>the same to any two users. Instead, an ever expanding collection of
>information servers (computers handling local databases) will be
>linked by networks to provide the user with the comforting illusion
>of a single system.
>
>Users will determine the interface. Vendors will offer a variety
>of personal information management packages for these little
>computers and most users will use a personally designed/customized
>interface.
This sounds alot like reports of the 1933 (I think) World Fair, which
described the world in 1960. While some of those predictions came true,
many, if not most are still pie.
There is alot of good ideas in Clyde's vision, but I have two problems with
it.
First, where is the money going to come from? My wife just interviewed at
a library that could not find the money to provide 13 more dumb terminals,
much less this high tech vision. I have not seen any numbers, but it
appears to me that the majority of libraries already have an automation
system. How are you going to sell throwing away the old system for this
new vision?
Second, with sophistication comes complexity. It's all find and well to
have on-line access to 1000's of libraries, but how are you going to
navigate thru this? Again at the library my wife was at, I could look at
the holdings from about 17 libraries in the region, but which ones are
here? Finally, I noticed that the local holdings were brighter. I would
say that this is very confusing for most patrons and is the type of thing
that will scare them away. Well, if you haven't really dealt well with
this type of scenario, what is it going to be like in the very networked
vision of Clyde?
I am reading alot of wonderful, grandious visions in this thread, but it
sounds alot like the techical elite talking (which, by the way, I am a part
of :-). In my journeys through various libraries and their OPAC I have
been uniformly disappointed. There is a definite gap, nay canyon, between
the reality I see out there and these visions. My rule of thumb is that
OPAC is successful when it is as useful and easy to use and the card
catalog. What I see in use today is not. My question is, how are we to
bring OPAC up this level?
>From: LYNCH@jade.bucknell.edu
>Subject: RE: OPAC Funtionality
>>Dirk Herr-Hoyman says:
>>We MUST get away from a command line, terminal type of interface. Touch
>>screens, like you see in mall kiosks, or light pens come to mind as far
>>better interfaces.
>
>Hang on a second, here. I personally hate touch screens and especially
>light pens. I much prefer a command line interface. As Millard Johnson
>suggested here, why not have the interface be tailored to the user?
>Why not have touch screens for some folks, voice activated for others
>(e.g. blind or handicapped), point and shoot for gui afficionados,
>and command lines for folks like me?
>
Sounds great. But again, who is going to pay for ALL of these nifty
interfaces.
I am suggesting that for a PUBLIC (not the one on your desk) catalog, a
pointing device is a good, intuitive way to go. Voice would be better, but
that's too pricey at this point. Within pointing devices, direct screen
pointing is better for the novice (and the novice level is what I believe
we should shoot for), than say a mouse.
Dirk Herr-Hoyman | Are we
University of Wisconsin-Madison | kinder and gentler
Dept. of Family Medicine and Practice | yet?
dhoyman@fammed.wisc.edu |
(608) 262-6368 voice |
(608) 263-5813 fax |
*-----
Subject: Re: Post-Boolean OPACS (fwd)
From: "Joan Eslinger <wombat@kilimanjaro>" <wombat@kilimanjaro>
I'm new to this list, so forgive me if you already know this.
Are people familiar with the work of Ted Nelson and
Xanadu/Autodesk/AMIX? For the last 20 years or so Ted Nelson has been
interested in making information easily available electronically. You
might be interested in some of their ways of cataloging, classifying,
and otherwise marking information.
Xanadu is a system being built to support that. Someday it may even be
finished and released. (Autodesk bought all or part of Xanadu out.)
In the meantime, some of the same people have been involved with AMIX,
the American Information Exchange. I don't know much about it, but it's
supposed to debut soon if it hasn't already. For more information, you
could write to janet@xanadu.com.
*-----
From: dhoyman@fammed.wisc.edu (Dirk Herr-Hoyman)
Subject: Post-Boolean OPACS
I have seen a search method that is neither boolean nor exactly keyword
called Latent Semantic Indexing (LSI). This comes out of Bell Core, which
used to be Bell Labs. With LSI, one can do a keyword search using terms
that may not even be specified nor even existing in the text. The idea
is that there is a larger vocabulary used to describe concepts than any one
individual uses. Bell Core found that a person typically uses only 30% of
the vocabulary. LSI attempts to utilize the underlying 70%, which is
termed latent. Thus, the name LSI.
LSI utilizes a technique from social science statistics, factor analysis.
It groups together terms/words found in a document (that's a factor) and is
run on a group of documents. Thus, a term could get a hit on a document
when it was not in that document because it was in a factor with some term
that was in the document.
A search using LSI will return a value between 0 and 1, where the higher
values indicate greater relevance. Unlike a boolean or keyword search, LSI
gets what we might call partial hits. This makes LSI very valuable for
searching large numbers of documents or any other type of text (like MARC
records), as you can get a group of hits which are ranked by this index,
again, unlike boolean or keywords where its a 0/1 proposition.
From a users perspective, searching with this type of indexing is far
easier than boolean. This was another area that the Bell Core research
explored. They found that novice users were much more successful using
LSI, than boolean searching.
If someone wishes to see the references to this work, I can drudge them up
for you.
Dirk Herr-Hoyman | Are we
University of Wisconsin-Madison | kinder and gentler
Dept. of Family Medicine and Practice | yet?
dhoyman@fammed.wisc.edu |
(608) 262-6368 voice |
(608) 263-5813 fax |
