[seek-kr-sms] RE: growl+icons
ludaesch at sdsc.edu
Wed Apr 7 09:20:45 PDT 2004
I'm cc-ing to SEEK-KR-SMS because of general interest.
>>>>> "FV" == Ferdinando Villa <ferdinando.villa at uvm.edu> writes:
FV> I think it looks a whole lot better, although it may be less intuitive
FV> with these icons. A lot will depend on who is browsing - I'm sure it
FV> looks great and intuitive to Bertram. I'm not sure about some of my
FV> ecologist colleagues.
Indeed there is something inherently problematic with the
visualization of ontologies that are more than just labeled directed
graphs (a la RDF or OWL lite) but that contain formulas/axioms.
Of course you can visualize an OWL statement such as
owl employee eqv person and worksfor some employeer.
(here I use what we call the "Sparrow syntax" of OWL; Shawn and I are
working on some Sparrow tools -- more soon ;-)
But the problem is that when you visualize a formula, what you
effectively do is visualize the parse tree of the formula, but not its
"meaning" in terms of the facts (say of the form C1--R-->C2). So I
think there is some utility in visualizing formulas in graphs, but it
has to be taken with a grain of salt.
To improve the situation one needs a reasoning that can derive some
"consequences graph" from a set of axioms, most notably the class
hierarchy. Then that one would be visualized.
Visualizing axioms is not bad, but doesn't tell the whole story.
I think that for description logic formulas the textual representation
is still one that should be there as well.
Of course the "ontology graphs" are more suitably visualized as graphs
(while the jury is till out for "ontology formulas"; for those, as
mentioned we should often just visualize some or all of the "graph
FV> I think we should strive to design a set of icons that reflect the
FV> meaning clearly and allow us to preserve (and augment) the
FV> intuitiveness. This is NOT our job. I'll start interviewing scientist
FV> pals, and we should circulate a request for comments in the broader SEEK
FV> community. Then we should involve someone who knows about visual
FV> communication and get help for the actual icons. If you can compile a
FV> list of the required icons and their meaning, we can start from there.
FV> As far as your work is concerned, if the icons are configurable, you've
FV> done a great job and I don't think you need to come up with the icons
FV> yourself. Congratulations! I look forward to see all developing SEEK
FV> ontologies in a browser on the seek site (with a button to edit them and
FV> submit a documented change, of course!)
Serguei and Kai:
I think both of you aim at similar things for SEEK and GEON,
In GEON we're looking at two tools for ontology entry: Codex from PSU
and the Cmap tools...
time to compare notes =B-)
FV> On Wed, 2004-04-07 at 11:36, Serguei Krivov wrote:
>> Hi there,
>> Have a look at icon based version of ob:
>> and let me know if it is any better then it was.
>> I made intersection/union/complement notations as in Bertram's domain
>> maps which seems to be natural for dl based things.
>> Do you thing icons for subclass relations is better optin then label
>> "subclass-of"; personally I doubt this.
>> Rich -any idea how to make the background of gif icon transparent?
More information about the Seek-kr-sms