Is Precision Too Precise?: 'Neo-Conventional' Functionality for User-Controlled Information Retrieval

[Gerry McKiernan]

             _Is Precision Too Precise?:
    'Neo-Conventional' Functionality for User-Controlled
                 Information Retrieval_

   As part of my never-ending review of Data Mining and Knowledge
Discovery in Databases, I initially focused on technologies
or approaches that would explicitly enhance identification of
relevant records by increasing the precision of results.

   Implicit in my review is a complementary belief that there is
also a need to provide opportunities not only for systems and
algorithms to mine data and discover knowledge, but also a need
to enable users to mine and discover data and knowledge within
a 'structured Information Space'. That we need to recognize
the inherent desire [shall we say the Human Need to Browse]
to discover and therefore need to create information systems that
facilitate such discovery in a user's Information Quest.

   [For a great review on the topic of browsing, folk are
referred to the excellent work of Chang and Rice as well
as the bibliography by Kurth and Petters noted below:

   Shan-ju Chang and Ronald E. Rice, "Browsing: a multidimensional
     framework," Annual review of information science and technology
     28:231-276 (1993).

     Martin Kurth and Thomas A Peters, "Browsing in information
     systems: an extensive annotated bibliography of the
     literature," Library hi tech bibliography 10: x, 275 1995]

    Belkin (at Rutgers) as well as others have clearly documented
the users desire to truly _interact_ with information systems in
the Information Seeking process.

    My belief in the value of structured browsing for Knowledge
Discovery was the prime reason why I adopted a standard library
classification scheme for my CyberStacks(sm) collection. Likewise
recognition of the inherent limitations of the use of one
structured Information Space, led me to not only consider
the benefits of a search engine [still pending], but the benefits
of enhancing the search process by the use of appropriate
algorithms that would increase the identification of the most
relevant set of records, thereby alleviating the use of the
the associated tedium.

    My recognition of the nature and value of browsing so well
documented by Chang and Rice in their seminal review paper
noted above, has led me to create two 'neo-conventional' mechanisms
with CyberStacks(sm) that offer users a new ['neo'=new] type of
browsing that had not been available in conventional public information
public information systems at the time [Since creating these
function, the experimental online catalog of the Library
of Congress has in fact incorporated features that facilitate
enhanced structured browsing. For a test drive of this experimental
OPAC, users are invited to visit my Onion Patch(sm) project at:

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

  Within CyberStacks(sm), I have created a Cross-Classification
Index. This index also users to browse resources classified
within an individual LC category from an alphabetical listing
of the Library of Congress topic headings associated with a
specific class number. It is an effort to provide an entry
vocabulary to a classification range in which users will
find related classification topics offered for their consideration.
This functionality has been provided in the belief that many
users do not know _exactly_ what they want, but they know it
when they see it [It's like when I go shopping with my wife
for a new shirt or pair of shoes - I can't always explain the
style or color of what I want, but when I find it, I (usually)
know it [and buy it] [:->]!]

   A second 'neo-conventional' functionality that has been greatly
expanded in CyberStacks(sm) is a Title Index icon-link that allows
users to enter the segment of the CyberStacks(sm) where an
individual resource has been described and classified along
with others of a similar classification. This feature allows
users to move from a possibly-relevant title into the subject
class where it has been classified and presents users with
resources that also may be relevant given their similar grouping.
It's the equivalent of allowing users to browse the shelves of
a library from a selected title, a feature found within the
Library of Congress experimental system, that appears to becoming
commonplace in more OPAC systems (e.g. Innovative Interfaces)

   My Cross-Classification Index is viewed as a precursor to
other kinds of browsing structures within CyberStacks(sm),
notably a 'hyper-thesaurus' that would allow users to browse
the syndectic structure (e.g. broader, narrower, and related)
terms, phrases and/or subject headings assigned to records
within a database (WWW, OPAC, etc.). I expect that the new
Scout Report Signpost that uses not only the Library of
Congress classification, but LC subject headings (LCSH)
and/or the ADAM project in the UK that uses the _Art and
Architecture Thesaurus_ will be among the first to provide
enhanced structured browsing to their respective collections
through the implementation of some form of 'hyper-thesaurus'.

    In addition to enhancing access to information through
the use of conventional structures, I believe that we will
also see the development of 'neo-conventional' vocabulary
structures that will allow users to browse through pathways
of linked subject headings derived from the associations
of subject headings and their associations with subject
headings of associated records [What?]. A precursor of
this functionality [and my inspiration] can be found
in the 'Related' subject heading option and the 'Sort'
option in the Lib of Congress experimental system. My
vision of such a function is still 'Under Construction'
[I hope that with the right inspiration  I will be able
to (more) fully articulate this functionality in the near
future]

    In the meantime, I would appreciate learning about
other inspired (or not-so-inspired) methods of structured
browsing in Web as well as OPAC databases. [Certainly,
the Information Visualization technologies in my _Big
Picture_ clearinghouse offer 'neo-conventional' forms
of structured browsing]. As always, any and all leads,
suggestions, recommendations, opinions, citations,etc.
would be most welcome!

    Regards,

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

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

P.S. For an elaboration of the Cross-Classification Index
and Title Index icon-link feature, interested folk are
invited to review my article in the _Journal of Internet
Cataloging_

    "The New/Old World (Wide Web) Order: The Application
of 'Neo-Conventional' Functionality to Facilitate Access
and Use of a WWW Database of Science and Technology Internet
Resources". _Journal of Internet Cataloging 1(1): 47-55 (1997).

                 "The (Dis)Plays the Thing"
                     with apologies to WS