Bernhard Eversberg wrote:
> Notations need to be human-understandable, learnable, and compact,
> shelf-order or no.
I'm not sure this is so, if you are suggesting that the need for a
shelf-order has little imapct on the requirements for notation. What do
you think of LCSH's 'notation'? I suppose it does have one, the terms
themselves and the
sub-division structure.
Is it human-readable? Yeah, more so than notations of vocabularies meant
for shelf-order. Learnable? Arguably the same. Compact? Well, yeah, it
IS actually fairly compact, but not NEARLY as compact as notations meant
for shelf-order.
And it's defintiely not a series of numbers and letters that some of the
discussion assumed was needed for any classification vocabulary. It's so
different that we aren't used to considering it a 'notation' at all. But
in fact humans are certainly able to apply it and allocate it's terms to
documents. Despite the fact that there is certainly NOT "an outline of
at most two pages showing the top level of the hiearchy." Now, to be
sure, we could in fact _use_ more hiearchy in LCSH, to present certain
interfaces to users we want to present--NOT becuase it's neccesary for
humans to apply it, since it's being applied fine now.
I'm aware of course that LCSH isn't considered a 'classification', so
perhaps you think I'm cheating. I think that if we put shelf-order
aside, the differences between 'classification' and 'alpha subject
index' become a continuum of grey area, rather than a clear binary. As
evidenced by the fact that many implementors DO want more hieararchy in
LCSH, as evidenced by the fact that we use LCC/DDC _and_ LCSH for
facetted navigation and other interface features, etc.
We need to link features of our vocabularies to the interfaces we want
to provide. A 'numeric' style notation is not neccesarily needed for the
features we want to provide (and certainly sin't needed to make it
workable to assign, as LCSH shows).
Of course, LCSH too was designed for the physical world---it's
'notation' was in fact designed to put like things with like things when
filed alphabetically. In that sense, similar to shelf-order 'numeric'
notations, in fact. To what extent do we need this feature? To what
extent does it get in the way of interfaces we want to provide? The
nature of subdivisions are toward this end--why have the facetted
navigation interfaces we have seen disassembled the sub-divisions
destroying this feature (and taking some semantic content with them in
the process)?
Jonathan
>
> Because humans will have to apply the classification and allocate
> notations to documents. Yes, software could aid these processes,
> but don't get carried away on that idea! Those who work with it
> will not all have the appropriate software to their service, at
> all times and on all occasions. So, it must be easy and quick for the
> unaided human to understand, learn, and apply the notations.
> Which doesn't mean there shouldn't be a complex authority database
> network to keep things together and facilitate the maintenance
> of the classification, including multi-lingual thesaurus features.
>
> There has to be an outline of at most 2 pages showing the top
> level of the hierarchy. My initial question was just what that top
> level should look like. What division of knowledge or resources
> would be useful and plausible for our time and age, and open for
> future extensions as well? Perhaps lets compare BSO with Dewey and LC
> and pick those sections that we think will cover today's world and
> in their sum make a good first level. Only after that, think about good,
> brief, plausible, learnable notations for that list.
>
> B. Eversberg
>
--
Jonathan Rochkind
Sr. Programmer/Analyst
The Sheridan Libraries
Johns Hopkins University
410.516.8886
rochkind (at) jhu.edu
Received on Thu Jun 07 2007 - 07:51:37 EDT