Re: Showing the Semantic Web

Jeroen van der Ham wrote:
>> IsaViz [1] is not actually aimed at showing Semantic Web data to 
>> anybody in the "real world". Its representation is too low-level for 
>> that. You are right when you say GSS tries to improve this, but even 
>> with GSS [2], IsaViz is still aimed at people who want to get an 
>> understanding of the structure of their RDF model, and get a more 
>> graphical, node-link diagram-based representation of their data. That 
>> is mostly RDF *developers*.
> 
> I agree with you that currently these are too low level.
> It might be my application domain, but I feel that it is important that 
> the structure I am putting into an RDF graph can be viewed as a simple 
> graph as well. I have a feeling that anything that is put into an RDF 
> graph has an underlying abstract graph that can be used to display the 
> data in a good way and show people that you are really bringing order 
> into chaos.

I entirely agree with you, but as you say I believe that the graph 
structure by itself, I mean, the interest that one might have in seeing 
this structure, will vary depending on the application domain.

Bioinformatics is one such domain, see e.g. mIsaViz [1], a 
specialization of IsaViz implemented by a team in Brazil that is used to 
represent RDF models that encode metabolic and regulatory pathway data.

[1] http://www.intelab.ufsc.br/misaviz/


> Take for example FOAF, you would really want to do something similar as 
> foafnaut: if something is a foaf:Person, has a foaf:name and 
> foaf:depiction, then use that foaf:depiction as icon and use foaf:name 
> as label.
> If you then only show foaf:knows relationships as big fat arrows, then 
> you can easily explore the whole graph and show the underlying structure.

Yes indeed.


>> For applications that try to present Semantic Web data and reach a 
>> broader audience, take a look at the various tools developed within 
>> the Simile project [3] such as Longwell and Piggy bank.  Other 
>> interesting tools out there include Haystack [4], mSpace [5] and 
>> Noadster [6]. 
> 
> The problem with most of these applications are that they use the 
> structured knowledge in the background to provide a clear view of (parts 
> of) the data. Now this is very interesting for us as RDF developers, you 
> aren't going to win over Joe Smith who doesn't care how the data got 
> onto his screen, just that he can see it. And to him there is no 
> difference between some large database that uses complex queries and 
> some RDF that does the same thing in a simpler way.
> 
> RDF gives you underlying structure of the knowledge you are describing. 
> I firmly believe that we should leverage that underlying structure to 
> display this to the user and allow him to more intuitively explore the 
> data represented to him.

Again, I believe this is highly dependent on the application domain and 
on the user's task. But there are indeed cases where this is true.

Emmanuel

-- 
Emmanuel Pietriga
INRIA Futurs - Projet In Situ    tel : +33 1 69 15 34 66
Bat 490, Université Paris-Sud    fax : +33 1 69 15 65 86
91405 ORSAY Cedex FRANCE     http://www.lri.fr/~pietriga

Received on Thursday, 19 January 2006 18:30:54 UTC