Natural Language Database Advisors for Libraries

[Gerry Mckiernan]

               _Natural Language Database Advisors for Libraries_

   In response to my recent post on Database Advisors
[ http://sunsite.berkeley.edu/Web4Lib/archive/0112/0257.html ]

   I received the response below from the Senior Web Engineer of Ask Jeeves
[NOTE: The excerpted text is for informational purposes only and does not constitute an endorsement.; it  is reposted with the permission of the author, Paul Hounshell]

    I believe that the technologies described below offers new for navigating library resources most notably Natural Language input to digital knowledge bases and OPAC.

   Are My Web Colleagues aware of any applications of AskJeeves (or related) technologies to  querying databases, OPAC, e-journals(?), etc. or other digital *library* resources? Articles, papers, reports, and other relevant literature relating to the technologies (and their applications) in a *library* environment are of particular interest.

  Regards,

/Gerry McKiernan
Natural Librarian
Iowa State University Library 
Ames IA 50011
gerrymck at iastate.edu 

                    ****************************************
>>> Paul Hounshell <PHounshell at askjeeves.com> 01/02/02 07:41PM >>>
I received this email via a friend and I have a [biased] opinion.  However,
I'm not sure what you're asking for exactly.  Ask Jeeves has a few products
which go along this route.  The first is called simply "Advisor".  It's a
system that guides users through a question process to determine what they
are looking for and recommend the best item for them on that basis.  I think
of it as a complicated "fuzzy" parametric search.  But, since you're
affiliated with a library, that probably is not exactly what you are looking
for, as the amount of work required could be substantial.  There's an
example available at:
http://advisor.ask.com/advisor/AJ_Advisor_1.asp?cid=6&cat=digital+cameras 

The other product is JeevesOne.  It's their natural language search engine
boiled down for searching individual web sites.  As I understand it, you run
the app and it spiders a site looking for all documents then users can
search using normal english phrases, such as "What books are available on
the effect of spider venom on infants?" (OK, OK, I'm not good at making
questions up, but you get the idea).  That's being used by Ford, Nike,
Purina, Washington State and a couple other big hitters.

The last product is called Answers for Databases.  It's essentially that
natural language engine, but designed to run specifically on databases
instead of web pages.  It sounds like that is what you're most interested
in.  The only clients I know of that product are Office Depot and Nike.
Nike's is live at http://nike.ask.com/index.asp where the larger text box
near the top is for a question, while a more parametric search can be done
below.  It can actually take fairly complex requests.


                                      [snip]