INFOMINE "Hyper-Thesaurus"

[Steve Mitchell]

Hi Gerry,



>>Hi Steve/
>>
>>On Friday, September 6, Steve Mitchell wrote to Web4Lib
>
>>What do you mean by a hyper-thesaurus? I'm thinking you
>> could examine  the subject headings (hyper-thes.?) list
>> for any of the INFOMINEs. Hundreds of terms, with item
>>counts, for each general set of disciplines will be All are
>>linked to search the INFOMINE to which they apply.
>>
>>The application of LC headings within the INFOMINE collection
>>represents an outstanding example of the value-added benefit of
>>adopting/adapting standard library practices to the management
>>of Web resources.
>>
>>  My concept of the 'hyper-thesaurus' takes this approach to the
>>next logical step. Instead of having _only_ an alphabetical listing
>>of headings with the associated number of entries within the
>>appropriate INFOMINE group, these subject headings are structured in
>>a way that is identical to their arrangement within the Library of
>>Congress Subject Headings (LCSH). For each entry within LCSH, the
>>Broader Term or phrase is indicated as are Narrower Terms and
>>Related Terms. This structure presents conceptual relationships
>>among terms and phrases.
>>
>


This is something we've thought about and, to some degree done, in
INFOMINE, at least with the broader terms, through what we call
Discipline level LCSH (see http://lib-www.ucr.edu/infomine
/disciplines/bioagmed.html for an example). This is a controlled
domain of mostly broad LCSH which
we apply in addition to other and/or more specific LCSH. To add the 
narrower/broader relationality (above and beyond what we have)
would be ideal but also brings up questions re the efficacy to
users vs. the time and money to do so.


Note that the indexing for each record is visible upon clicking
on a box for "related resources" that will be found under each
record's title. This allows the user to view and then click on
other subjects or kw which are then auto-searched. The broader
terms and narrower terms, while not specified as such,
are usually pretty obvious at this point. 


>> With the collection of headings within INFOMINE, I can see how
>>a hand-made LCSH subset could be easily created [Assuming of course
>>that neither of us has anything thing to do for the next six weeks
>>[:-)].



The fact is that our cooperators have already been, through their
contributions, implicitly formulating an INFOMINE LCSH subset
which has been working pretty well in practice.
We can extend/fine-tune this as the need develops...some of
the ways you suggest are interesting, depending as mentioned
on the time/$ required to employ them.

One concern we have about structures with limited/rigid
ways of representing the content of a resource, as Bill Drew
pointed out, is that you can actually end up misrepresenting
or under describing a resource fairly easily. And this occurs
regardless of whether your approach is more image based or
text based. But as you've mentioned all indexing schemes
have their drawbacks...and, most, including ours,
involve compromises.  


A main concern with many of these schemes is that of the time involved in
indexing/inputting a record. There are a lot of good organizing schemes
out there but many of them are too involved in that they ultimately are
requiring indexers/catalogers to spend up to an hour to adhere to the
conventions required for creating each record. This is a major drawback
because it means that, given the momentum of growth of academically
valuable resources on the Internet, a lot of good resources will not be
included due to time and labor/$ constraints. This is one of the
challenges of the Internet: there will be more good info
coming at us than has ever been the case before and approaches
that worked well in print often wont/dont work well economically in
handling this new mass of materials.

On the other hand, machine based descriptive schemes, while promising,
still generally miss hitting the barn door by half a mile in terms of 
accuracy/providing information in the right context... 
Similarly, organizing schemes which rely on author conventions for
generating indexing/descriptive infomation aren't exactly taking off...
though in time they may...lets hope.

Its a question as always of determining the right mix of ideal site
organization vs the time and costs of achieving and maintaining
this mix. Our feeling has been that it has been important to
get going, to get concrete experience with this environment/type
of work and, through this experience, help create practical solutions
to organizing it.  This experience is guiding the evolution of INFOMINE.





>>The key of course would be to automate the whole process as
>>has been done in the general INFOMINE environment, a truly original
>>and innovative approach!
>>
>>The INFOMINE collection and its associated collection of links to
>>specific Net resources would certainly be a most appropriate test-bed
>>for such an experiment.
>>


We're certainly open to many of the ideas you've raised and are
ourselves a cooperative project relying on both on and off campus
participants. If you have specific ideas about how to modify
INFOMINE to do what you need then let us know. And, because
our project is an ongoing one, we'll try and get a spin on
what costs are involved in addition to how much increased time
would be involved in adding records/describing resources with
the augmentations you're suggesting. In the meantime, we'd like
to invite you and others interested in agriculture
(or any other academic subject) to participate in INFOMINE.



Cordially,

Steve Mitchell
INFOMINE



>>  The initiatives at the Library of Congress, particularly
>>that with the thesaurus of graphic materials, is one of the
>>major efforts of which I am aware (and one of the best BTW).
>>[LC is of course also involved with larger projects in this
>>area with LCC and the general LCSH as a CD-ROM service]
>>
>>  The URL for a representative presentation of what I have in mine
>>is:
>>
>>http://lcweb.loc.gov/lexico/tgm1/h/Homing_pigeons.html
>>
>>Homing pigeons
>>
>>--[country or state]
>>
>>Used For
>>     Carrier pigeons
>>     Homers (Birds)
>>     Racing pigeons
>>Broader Term
>>     >Pigeons
>>     >>Birds
>>     >>>Animals
>>Date Approved
>>     24-Feb-89
>>
>>   Click through the hierarchy of Broader, Narrower and Related
>>terms for an idea of what I have in mind for this type of navigation.
>>
>>   I can see this approach being modified to present the Web
>>resources (or even non-Web resources in an IPAC - an integrated
>>Internet OPAC [:->]) and used to provide additinal navigation
>>through a CyberCollection.
>>
>>  Regards,
>>
>>Gerry McKiernan
>>Curator, CyberStacks9sm)
>>Iowa State University
>>Ames IA 50011
>>
>>gerrymck at iastate.edu
>>
>>http://www.public.iastate.edu/~CYBERSTACKS/
>>
>>                     "Imagine"
>
>