Re: AI4LIB

I really liked this quote:

  "...we have reached the point where our risk aversion as an institution actually has become the major risk to the institution/profession... where we end up basically outsourcing the profession."

  Bernie Sloan

Alexander Johannesen <alexander.johannesen_at_GMAIL.COM> wrote:
  Here's a mail I received a few days ago. I have his permission to post
this here ;

Hi,

There actually are successful machine learning (A.I.) projects that are
ongoing in libraries and have been for some time. The INFOMINE/iVia/Data
Fountains projects is one example. We have been pursuing aspects of
machine learning, re: automated/semi-automated resource discovery and
metadata generation (i.e., collection building), continuously for about
a decade. A small project working in a very challenging area, we have
had some modest successes. iVia/DF is open source code BTW and available
to all who are interested. There are other projects I could mention as
well if you're interested.

How to get started? We just got busy and started fund raising to get the
wherewithal we knew we needed to employ good programmers along with good
librarians/application developers. We realized that we would have to go
outside our community a bit and tap into advanced programming expertise
and serious machine learning knowledge (i.e., we needed to move beyond
usual in-house Web interface expertise). The good news is that the
faculty experts that are working with us have benefited as much as we in
exploring/dealing with the opportunities/challenges of this
inter-pollination. Text processing and machine learning have been and
remain HUGE now in CS and Info Sci. Note as well that high level library
managers love it when you have a path breaking project and can do a bit
of successful fund raising. It makes the whole Library look good of
course in addition to paying for important research/development. The
down side is that there are real challenges in keeping primarily grant
based projects going, keeping continuity and advancing the same threads
of research and development. Still, we've done it for over 10 years.

The way we look at the whole area is that many of the traditional
activities and services of libraries just don't scale well anymore
considering the amount of high quality information available that our
patrons need. Our intent then is to develop tools that will amplify and
augment librarian expertise and effort. We're not looking at the work as
"A.I." writ large but as something more akin to having servo-mechanisms
that save, at least at first, 10-20 minutes in developing, say, a
description for a resource. This time adds up to serious resource
savings. There is no intent to replace librarian expertise because.
Given the maturity of this area, it just isn't possible and wont be
possible for a long time to come. Instead, there is the intent of
reducing laborious, repetitive tasks that really do not require the
level of expertise that we have...the re-purposed time would allow us to
do a lot of good stuff.

I think libraries need to ask themselves about the major opportunity
costs of NOT getting involved; of not shaping and guiding this
technology. If we don't engage, others will simply develop it less well
than we could have and sell it back to us at a premium. Their spin to
high level campus/library managers could very well be librarian
downsizing... not developing new, more empowering and useful roles and
services for library user communities. We've outsourced so much, much of
it not even consciously, simply by not picking up the challenges of the
times that we have lived through...not rising to the occasion. Some of
us, myself included, think that we have reached the point where our risk
aversion as an institution actually has become the major risk to the
institution/profession... where we end up basically outsourcing the
profession. And that would be too bad as most of us are very service
oriented, know how to structure and maintain the best of human knowledge
for the general good, and are very concerned about having fairly open
flows of information in society (think of the great work of ALA and
other library affiliated efforts put out to question the Patriot Act).
At any rate, you might consider making the point that there IS a lot at
stake when we choose not to involve ourselves in major, current
technological or service opportunities/challenges.

Cheers,

Steve Mitchell


--- end of forwarded mail ----


Alex
--
---------------------------------------------------------------------------
Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------



---------------------------------
Take the Internet to Go: Yahoo!Go puts the Internet in your pocket: mail, news, photos & more.