I had said: > I take it whatever people think though, no one is going
to claim such
> software will produce subject headings like...
[examples]
Alex responded: My question is (and it's a very serious one), what is
that assumption based on? Why wouldn't these systems create subject
headings like this? Why do people think they can't? Is it purely because
no one have shown you a specific example you can agree on? (Or at all,
given the current state of ILS)(end)
I understand what you are saying here Alex. But as you say, this is all
"Hush-hush" and that makes it hard. We all need evidence in life.
Karen's example about evolution does not strike me as not good, because
trust in human authority is so tightly bound up with it. How many
people who think evolution is true can really talk intelligently about
it - on a deep level - and then also deal honestly about the
shortcomings (the lingering questions people always bring up) of the
more militantly atheistic strains in particular? We are people. We
need to see the facts ourselves and also evaluate the frameworks that
the facts are being evaluated in for ourselves. Translating this to the
"trust me" re: AI, maybe with such an attitude, I'm doomed to see it at
the Googleplex as you say - in which case I'll have to take up teaching
or something else that I can do. In any case, I guess you can call me
doubting Thomas - some of Jesus' disciples may have seen and touched his
resurrected body - and even "eyewitnessed" this to Thomas - but the poor
guy still needed to see it himself.
I'm the same way. Who do you trust? Who can you trust? Who seems to
be honest - who seems to actually deal with evidence that doesn't fit
their pet ideas they cling to? It is all about evidence.
Tough-mindedness. Testing claims. Listening to evidently the best
informed folks on the opposite ends of the spectrum have it out.
That said, I don't think its because someone hasn't shown me a specific
example *I* can agree on - but that this evidently "hush-hush" stuff
(which evidently works on various kinds of books even...) *evidently*
has not been tested by people known to be excellent catalogers and other
non-catalogers who are experts in their various disciplines - i.e. it
has not been "brought out into the light". I say "evidently" for a
reason - because maybe someone has something to share for poor doubters
like me.
(in addition, you can start talking about how these folks could not be
objective and have their jobs on the line, etc., but I think that many
of the best would be forthright about when they've been beaten).
James Weinheimer I think seconds me: "To decide if a single search is
successful or unsuccessful can take some time, and a lot of knowledge.
So, this is why I am extremely skeptical when somebody says that
something is successful. I have seen lots of impressive papers written
on the topic, replete with graphs and mathematical formulae.
But personally, I don't care about the theory but in the results. This
must be determined by a disinterested group of experts after a great
deal of research."
Alex, you also said:
Well, to paraphrase from my blog, this is not about no human
intervention, or even the eradication of the cataloging profession. In
fact, this is an area where human catalogers help utilize and tool the
systems would make the library more relevant to the human race than any
other automatic attempt. I'm just sayin'.
In other words, like what Karen said about MESH and has been confirmed
here since. Fine (even though I think you also tend to give the
impression that AI is capable of doing the job with an absolute minimum
of human intervention). Then where to you and folks like James
Weinheimer and myself disagree? I keep coming back to your statement
here (to Ed):
I'm sorry that you need to see it now (and also fulfill your specific
requirements; why those?)
You had asked "why those?" (namely Ed's list which you thought was too
generalized and had lots of straw) for example, "reasonably good,
controlled vocabulary subject headings".
To me the answer is obvious. Authority control - or call it vocabulary
consistency - is an absolute must if you are going to have any useful
system at all.
In other words, content is the key.
Have you ever read this article?:
http://www.libraryjournal.com/article/CA434443.html
Let me quote Roy Tennant from the article: "But I could hardly believe
the results." (Re: his super-easy harvest of 100,000 bibliographic
records describing free online resources held by five different
libraries using Open Archives Initiative Protocol for Metadata
Harvesting [OAI-PMH] in 2004)
He also says in the article, talking about his "harvesting epiphany":
"What I had was a pile of metadata problems that in hindsight I should
have expected."
But the point is that he didn't. Why? Evidently he didn't listen.
Evidently, he didn't deal with the evidence that was available to him -
he should have paid attention to expert cataloger folks like James
Weinheimer and Bernhard Eversberg a little bit more.
Even today, many, if not most catalogers, want to do what they've always
been doing - respect some complicated rules not because they make simple
things complex (and so assure us of work), but do so because this whole
issue really is that complicated.
And I think most of us in the cataloging world would welcome some help
from AI. It sounds like it could make things quite a bit better.
In order to help along the "doubting Thomases" among the catalogers, let
me make a plea. I think it should be super, super easy to do. Why
doesn't someone start with Conal Tuohy's claim about our current
capabilities (using Bayesian statistics) re: all of the David Johnsons
James Weinheimer informed us of? I know something about science and
research, so this ought to be easy enough to empirically test. First,
get all the works (only text, I assume?) of all the David Johnsons. Of
course, *in order to even get started here* I don't see how you would be
able to avoid needing to use something like OCLC's Worldcat (made
possible with its wonderful authority control, thank you!) in order to
find most, if not all, of these works. Then all you would need to scan
them and do the test, and find out if it worked or not. I think this
would be very, very helpful - and I want help. Does anyone have the
means of doing this?
Please understand us. Why can't we all just get along. :)
Love this list. Thanks to everyone who helps educate me. And Karen -
I'll try to make time to just read some good tech stuff too!
Regards,
Nathan Rinne
Media Cataloging Technician
ISD 279 - Educational Service Center (ESC)
11200 93rd Ave. North
Maple Grove, MN. 55369
Work phone: 763-391-7183
Received on Fri Aug 31 2007 - 08:40:44 EDT