W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 2004

Re: peer-to-peer schema models: "Users, not designers, create and communicate meaning."

From: Giovanni Tummarello <giovanni@wup.it>
Date: Tue, 30 Nov 2004 22:18:40 +0100
Message-ID: <41ACE3B0.7090201@wup.it>
To: Jonathan Chetwynd <j.chetwynd@btinternet.com>, www-rdf-interest@w3.org

Hello there :-) while i find the rest of the posting a bit unclear, i'll 
give a contribution to the first question.

We have been studying RDF P2P system for quite some time there are 
several proposals out there.. but i can confidently say "there is no one 
solution"., it depends on your scenario

do you distribute queries around and expect people to be nice and answer 
them or should you collect the info "about the things" you're interest 
and then do the querying yourself?

This in my opinion could be mapped to tree scenarios

a) Huge databases and or the partecipating peer doesnt really want to 
give them out (e.g. library, phone direcory, full geographical databaes) 
then the only thing you can do is distribute queries and hope the're 
nice enough to answer them. It is anyway a limited mode.. its impossible 
to answer specific join queries "give me the books that have an author 
in the faculty of .. " if you dont transfer it all to your self or the 
remote host doesnt agree to do the join for you.

b) real time queries "do you have this in stock?" ..  as before --> 
distributing queries using one of the proposed approache (see edutella 
and the subsequent papers, not aware about the availability and ease of 
integrating the code) is probably the only thing you can do.

c)  lets together annotate, music, images, ideas, annotations, etc etc 
et.. museums, reviews you name it. Generally speaking a "monotonous 
scenario" (information never gets deleted.. its reviewed/updated maybe)

In this last scenario we argue that a good idea might be just collect 
all the rdf infos that are somehow related to the things you're 
interested in and then enjoy a local full speed sustainable (you'r e not 
bothering anyone out) .  We believe this has some chances of actually 
working "in the wild" for "napsters like" applications, while others are 
in fact great  (if not the only way) for the other scenarios mentioned 
above.  For this, we developed "RDFGrowth" .  once you specify which 
aspect of knwledge you want to learn about.. you'll learn all about it 
(that is  "all"  "about" the URIs that some selecting mechanisms pics, 
for prcecise definitions refer to the paper.)

One good thing about rdf growth si that.. it works. Its available, its a 
small jar. .its used our DBin [1] p2p application (but its less than 
alpha right now.. but working on it)  and it sused in other projects as 
well  (hyperl.org) to syncronize rdf knowledge among websites.
Using RDFGrowth its a matter of .. importing a jar and and a few lines 
of code
Of course we're just counting seconds till the first semantic spammer 
comes. RDFGrowth 2 tries to address that.. :-) in fact ideas and 
suggestions are very welcome as well as how to integrate this approach 
with traditional query routing ones.
Please refer to this presentation [2] and this paper [3].

Ciao ciao :-)


[1] http://www.dbin.org
[2] http://www.dbin.org/twiki/pub/About/WebHome/RDFGrowth.ppt
[3] 
http://www.dbin.org/twiki/pub/About/WebHome/RDFGROWth_workshopISWC2004.pdf



Jonathan Chetwynd wrote:

>
> Does anyone know of current peer-to-peer RDF projects?
>
> By this I am referring to projects where the RDF metacontent or schema 
> is defined over time by users.
> (some people might consider the CD labelling project of this type.)
>
> I've proposed this model for a number of projects including:
> The WWAAC concept coding framework project
> The CEN metadata for accessibility  project
>
> There may well be significant hurdles in adopting this approach, 
> however was inspired to request evidence of current successes by the 
> excellent: "Where the Action Is"  The foundations of embodied 
> interaction by Paul Dourish.
> when after a particularly intractable and rambling passage he announces:
>
> "Principle: Users, not designers, create and communicate meaning."
>
> regards
>
> Jonathan Chetwynd
> http://www.peepo.co.uk     "It's easy to use"
> irc://freenode/accessibility
>
>
Received on Tuesday, 30 November 2004 21:21:52 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 07:14:57 UTC