- From: Story Henry <henry.story@bblfish.net>
- Date: Fri, 28 May 2010 11:15:52 +0100
- To: Mischa Tuffield <mmt04r@ecs.soton.ac.uk>
- Cc: adasal <adam.saltiel@gmail.com>, "K. Krasnow Waterman" <kkw@mit.edu>, Dan Brickley <danbri@danbri.org>, Sampo Syreeni <decoy@iki.fi>, Stephane Corlosquet <scorlosquet@gmail.com>, Melvin Carvalho <melvincarvalho@gmail.com>, Matthew Rowe <m.rowe@dcs.shef.ac.uk>, semantic-web <semantic-web@w3.org>
On 28 May 2010, at 10:35, Mischa Tuffield wrote:
>> I wonder though, how much more is actually needed than having everything decentralised, or running through one of many hubs?
>> Are semantic technologies needed beyond, maybe, the semantic representation of a profile, or maybe that is just it?
>
> Well, without being too controversial, and I am a big fan of FOAF/RDF/SPARQL (being messing around with this stuff since 04 [when i was all young and stuff], and have built a lot FOAF related thingies [http://qdos.com/apps] at my current job at Garlik - hopefully will build more) but I don't really think that a semantic representations of a profile (FOAF doc) is where the SW community could make the most impact in the space of distributed social networks.
It's the linked data piece that is the most important contribution of the semantic web, and there it certainly does help to have standard representations (plural: rdf/xml, rdf, turtle,...) to represent the same distributed graphs. It just avoids having to re-invent the wheel.
> From my POV the SW community can make the most impact by illustrating a standard way which nodes in this distributed social network can communicate with each other. By that I mean, e.g. how we do a "friend request" (which people are talking about on the foaf-protocols list [ack. henry, nathan, melvin et al {sorry if I missed anyone out}]),
> something like a SPARQL UPDATE request, or my favourite would be a simple ping of FOAF document (this would require the FOAF representation I guess).
Yes, it is extremely easy to do. We have started testing this
http://markmail.org/message/zf2e2yrtpykklfo3
http://markmail.org/message/idx5zknj4zua5dgc
> If the protocol between the nodes in a distributed social network worked, them people could implement their "bit" however they wished, personally I would have it run off of a triplestore, something like what I did in foaf.qdos.com, but again, if there was a standard protocol (stressing that this "protocol" should be built on top of existing standards RDF, Atom, SPARQL, XMPP, etc, etc), then you could choose to not use any SW tech in your node in this distributed social network. Danbri, mentioned libraries, libraries which would allow for nodes to talk to each, this would be ideal, and here is where I think our community could have the most impact.
Agreed. Not every node needs initially to run off a triple/quad store.
I just retrofitted xwiki, a java wiki to produce webids. I put a screencast together
just today to show how it works from the users perspective.
http://www.youtube.com/watch?v=8iZPJBpI2Po
Henry
Received on Friday, 28 May 2010 10:19:58 UTC