Re: Requirements Document -was- Re: GUIDE: examples

Evan Wallace wrote:
> I assume that this is a reference to the Multimedia collections use
> case in the requirements document.  Actually, I had some problems with
> this use case.  I found it to be quite ambiguous regarding the properties
> that were being discussed.  The result was that the kinds of searches that
> might be expected to be based on the indexing information inferred from
> these properties would return lots of false hits (or wrong answers) when
> applying this pattern just slightly more generally.  Maybe that was part
> of the point of the use case, but if so that should be made clear, and the
> implied requirements for user control of assumptions or acceptable
> inferences should be made explicit.
> 
> What am I going on about?  Well in the case of styles, date.created for
> the style says nothing about date.created for an individual having that
> style (that is: the range of dates associated with the creation of the
> style is much smaller than the range of dates associated with the creation
> of instances of the style).  Take the example of a multiple listing search
> in the mid-Atlantic of the US for houses built in our Colonial era. Making
> the inference that anything in the Colonial style would likely be of that
> vintage, would lead to hundreds of false hits and quite possible no true
> hits!  This is not a special case, this is typical of many styles of
> antiques and collectables as well.
> 
> What this seems to say is that someone making these searches needs to
> be able to view the inferences (and default knowledge) to be used in
> the search and either reject or refine them to achieve the desired search
> quality.  Is that it?  If so, can this elaboration be included in english
> text in the Guideline example?  Can we also include it in the next iteration
> of the requirements document?  While we are at it, can we also either define
> or delete the "culture" relation/attribute?
> 
A problem of the use-case description in the requirements document is
that it is a summary of the descriptions in the original use cases which
explain them in more detail. For example, most properties (like
"culture") are defined by the VRA 3.0 standard for describing visual
(art) resources, a refinement of Dublin Core. 
(see http://www.swi.psy.uva.nl/usr/Schreiber/docs/collect-020107.html)

I am not sure I get your point about style and creation date, but it may
be a confusion between two types of concepts used by art historians (I
am not one, by the way): 

1. "Late Georgian" is a style that is used for objects actually crafted
during this period.
2. "Late Georgian style" is a style that refers to objects constructed
in the Late Georgian style, but at some later point in time. One could
view themn as "copies". 

With respect to the type of search: semantic annotations open a whole
new spectrum of search possibilities. Category
generalization/specialization can be used to broaden/limit search scope.
If a search term matches several possible concepts, interaction may be
needed to find out which concept the user is referring to. For example,
we found a need for this in cases where AAT and WordNet overlap. It is
really a new research area, of which we are only scratching the surface. 

Guus


-- 
A. Th. Schreiber, SWI, University of Amsterdam, Roetersstraat 15
NL-1018 WB Amsterdam, The Netherlands, Tel: +31 20 525 6793 
Fax: +31 20 525 6896; E-mail: schreiber@swi.psy.uva.nl
WWW: http://www.swi.psy.uva.nl/usr/Schreiber/home.html

Received on Thursday, 14 March 2002 08:35:52 UTC