[Fwd: Re: [seek-kr-sms] ontology/folder UI design in Kepler]

[Serguei Krivov]

Before discussing the ontology related GUI in Kepler it would be useful
to have:
1. A list of ontology related operation that domain user of Kepler will
do
2. A list of ontology related operations that Knowledge Engineers
associated with Kepler will   do.
It seems that some of you already  have implicit knowledge of these
operation, but does anyone have that  written down? If yes, please send
it around because I missed it, if not then I would suggest that "..
before discussing GUI people should have a clear answer to the question:
GUI to what? (F. Villa, Communications, 2001, 2004)"
 
The     ontology support in Kepler should be adequate to the tasks and
should not introduce the extra complexity if it is not necessary (agree
with Laura). E.g.: Suppose we want just to classify the actors according
to their domain and use  this classification for simple  search - then
we even  do not need ontologies  at all - simple thesaurus would do and
of course the tree control in this case is the most appropriate.  But
suppose we use ontologies for semantic typing and during the design of
workflow user uses ontologies to make decisions - "if this output is
comparable with that input". In this task what will  be the use of tree
based classification and of tree control interface to ontologies? I do
not see how it could be of use in this case. Perhaps the 2 dimensional
picture that shows actor classes as boxes and possible input output
relations as links will be most meaningful. (Note,  I do not say it
should be as complex as GrOWL, but as Ferdinando mentioned  in a few
months we shall incorporate in GrOWL graphic maps that will allow
generating  domain oriented diagrams from OWL ontologies and we may
create graphic maps for various views on actors hierarchy) On the other
hand in a Spartan  minimalist approach  one could manage with simple
search-one would click on IO port and get a list of compatible agents. (
no need for tree control in this approach either) The problem  with
minimalist approach is that user would not get immediate picture for all
possible connection and may need to run the search several times.
 
serguei
 
 
 
 
 
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/seek-kr-sms/attachments/20050301/b503b202/attachment.htm