NLP of LCSH

[Gerry McKiernan]

_Natural Language Processing (NLP) and Library of Congress
                    Subject Headings_

    In a recent posting on Structured Browsing for 'user-controlled'
information retrieval in Web and non-Web databases, I briefly
sketched an alternative to information access to that would make
use of the associated subject headings associated with a given
Library of Congress Subject Heading. One form of this 'neo-
conventional' functionalities certainly is the 'Related' records
and 'Sort' listing provided in the Library of Congress Experimental
Search System [accessible via OnionPatch(sm) at:

    http://www.public.iastate.edu/~CYBERSTACKS/Onion.htm

  Yesterday I have learned about a highly innovative information
system developed by folk at EOS International. Their systems
respectively, Information Quest and their Q series OPAC, have
among the most sophisticated 'neo-conventional functionality'
of which I am aware [Of course, there are others profiled in
Onion Patch (sm) [:->]. The URL for EOS International is:

                http://www.eosintl.com/

 Access to details on their Information Quest (IQ) and Q series
systems is accessible direct from

    http://www.eosintl.com/htdocs/products-services.html

or from the Products and Services link on the base homepage.

  Among the 'robust searching' technologies used in their
IQ system are Natural Language Processing (NLP) and a Word
Expansion feature that uses NLP to search for 'concepts, word
relationships, and semantic meaning' and a 'Related terms' function
and 'Query-By-Example'.

  In consideration the obvious benefit of NLP for identifying
and providing navigation of conceptual and semantic information
spaces, it occurred to me that NLP would be the ideal method
by which one could create the kind of non-syndectic associated
associations of LC subject headings that I seek in a structured
browsing environment. [BTW: I'm calling this function 'Explore'
It will permit users to explore the relationships of subjects
authors, publishers, series, etc. that are not facilitated with
conventional search options and which are not pre-defined by
conventional systems (e.g narrower, broader, terms).

  I would greatly appreciate learning other systems that have
applied (or are considering applying) Natural Language Processing
to MARC and other bibliographic databases, particularly in
providing enhanced 'subject navigation' by NLP of LC subject
headings.

  As always, any leads, suggestions, citations, opinions,
suggestions, comments, criticisms, critiques, etc., etc., etc.
are most welcome.

   Regards,

Gerry McKiernan
Curator, CyberStacks(sm)
Iowa State University
Ames IA 50011

gerrymck at iastate.edu
http://www.public.iastate.edu/~CYBERSTACKS/

                    "Show Me the System"