W3C home > Mailing lists > Public > public-semweb-ui@w3.org > May 2006

Re: SWUI workshop accepted at ISWC 2006

From: Sören Auer <auer@seas.upenn.edu>
Date: Tue, 23 May 2006 13:37:37 -0400
Message-ID: <44734861.7000103@seas.upenn.edu>
To: public-semweb-ui@w3.org

Lloyd Rutledge wrote:
> To achieve these four goals, submit a SWUI06 paper on them. And
> build them up in this forum, of course. How can you frame these
> questions from the perspective of an improved user experience? And
> what does SW in particular make possible that wasn't before?

Ok, I will have a look if I can contribute something, meanwhile I will
respond with some thoughts to your remarks and maybe someone else on the
list has some ideas, too...

> Your fourth bullet, minimizing user interaction to achieve certain
> goals, is interesting. Can you define the goals in terms of the
> knowledge structure? How do you measure "minimized interaction"? With
> number of user "clicks"? However, doesn't what you write about
> coordinated client/server speed up on the processing time, while the
> "clicks" remain the same?

I think there are three possibilities to optimize required user
interactions:

* minimize the reception time a user needs to realize what has to be
done and how
* minimize the number of clicks he needs
* minimize the latency between clicks

While strategies of the first category are probably more an art and
might be hard to evaluate. Second and third are easy to measure. The new
category of dynamic web applications (using technologies ala AJAX)
rapidly reduce latency between clicks by minimizing data transfer and
time for rendering.

> Bullets one and two mention "widgets" as general purpose, detailed
> components of interfaces. Are there certain types of interaction
> widgets that are specific to the Semantic Web?

I think widgets somehow represent elements of a data model. If you are
working with a relational database, you might have widgets to edit rows,
cells, data values in cells according to the datatypes provided by the
DBMS etc. With the Semantic Web I guess it's slightly more complicated,
since there is somewhat like a hierarchy of data models, i.e. RDF has
triples, RDFS arranges information in hierarchies of classes, properties
and instances, OWL on top adds constraints and logical axioms and
finally there are domain models. So eventually widgets for representing
and authoring data on the Semantic Web are also kind of nested:
A widget for adding a new gene in Gene Ontology for example could
combine an OWL instance creation widget with widgets for editing RDFS
annotation properties etc.
I guess it could be useful for application developers in the Java and
Web Application World to have reusable sets of generic widgets to create
domain specific Semantic Web applications, such as Desktop Application
programmers can rely on "file open", menu or tree widgets. However, the
challenging part may be the interfacing to the Semantic Web middleware
behind. What do you think? What could such an interface be, just RDF or
some query language? A standardized object model, i.e. something like
DOM for XML could be useful...


--Sören
Received on Tuesday, 23 May 2006 17:37:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:16:46 GMT