RE: Decentralized RDF Distribution

Re. the time taken for the cloud (yep, this term does make more sense) to
answer a query, one way of looking at it is that you have a persistent query
waiting for a match, but another would be for the cloud to be trying to find
minimums in its problem space, with your query as one (of billions?) of the
parameters. I don't think the cellular automata view is bad at all - just
the individual cells will be much more complex than in previous machines.

Or what about a narrower analogy - if I go to a search engine and look for
cloud9, it'll query its database (a line in a flattened cloud?). On the
other hand, I could pass the search term to a spider, and set that off to
crawl for the result - in other words an agent goes off and does some
discovery of its own, in this case getting the information from the source.
In SW terms this could mean a resource external to the cloud (i.e. not in a
semantic memory as such). Maybe even ask at the post office.


---
Danny Ayers
http://www.isacat.net

<- -----Original Message-----
<- From: www-rdf-interest-request@w3.org
<- [mailto:www-rdf-interest-request@w3.org]On Behalf Of Bill de hOra
<- Sent: 20 February 2001 01:10
<- To: Seth Russell
<- Cc: RDF Interest
<- Subject: RE: Decentralized RDF Distribution
<-
<-
<-
<-
<- : Seth Russell:
<- :
<- : Well you've certainly said a mouthful here and unfortunately i only
<- : understood about 5% of it.
<-
<- If you want to point out the 95% I'll be happy to clarify.
<-
<-
<- : But let me ask about your basic assumptions:  Do you
<- : assume that you will
<- : be able to query the entire semantic cloud and ever get a relevant
<- : answer back?
<-
<- With the right assumptions Seth, I can expect anything. That
<- doesn't make it so.
<- But if you're asking me can I expect an answer to a query in
<- reasonable time
<- then probably not, unless the request got lucky. But I didn't
<- just envisage
<- queries as one-shot requests, I wanted to be able to place them
<- in a tuple space
<- and persist, similar to the way a production rule or event
<- listener would work.
<- A tuple space isn't really best thought of as a data store: it's
<- more of a data
<- structure that doesn't draw a hard line between data and behaviour.
<-
<-
<- : Because if that is your assumption, and if you use the apparently
<- : traditional data
<- : processing approach you have scoped, may I extend my sympathies to
<- : your programmers in advance.
<-
<- You've lost me here.
<-
<-
<- : In my opinion, if that is our objective, we will need an almost
<- : biological approach ... perhaps like the one I have scoped with SCM.
<- :
<- : :SCM :acronymOf "Sticky Cyber Molecules"; seeUrl
<- : <http://robustai.net/MyNetwork/StickeyCyberMolecules.html>.
<-
<- You'd probably be interested in what Bill Joy has to say about
<- composing the
<- equivalent of multi-cellular life for
<- distributed systems, essentially that data isn't sufficient
<- (very much a snipe
<- at XML and SOAP, but he has a point):
<- <http://www.openp2p.com/pub/a/p2p/2001/02/13/joy.html>
<-
<- Bill de hOra
<-

Received on Monday, 19 February 2001 14:46:17 UTC