Re: bootstrapping decentralized sparql

2009/5/17 Giovanni Tummarello <g.tummarello@gmail.com>:
>> for graphs which use a (specific) FOAF term.  It's a bit like
>> PingTheSemanticWeb or Sindice, but decentralized based on the ontologies
>> used.
>
> [....]
>
> Isnt this like saying  "why set up  an infrastructure with
> professional developers and administrators,  in a backed up , UPSed
> datacenter..
> .. when you can ask each ontology creator to do the same."

Why spend resources on creation and upkeep of a triple store and
search service when an ontology creator can distribute the information
more efficiently using references to endpoints in their ontologies?

>> The result here will be that a query for a foaf:Person with a
>> foaf:firstName of "Sandro" can be *complete*, at least across all graphs
>> which choose to register themselves as having data about instances of
>> the foaf:Person class and triples using the foaf:firstName property.
>
> please explain how this that you describe is different from what's
> possible already
>
> http://sindice.com/search?q=foaf%3Aname&qv=Sandro+Hawke&qt=ifp

It doesn't rely on a single search engine. What happens when the rdf
web equivalents of google go down? ;) Granted that the ontology
authors being required for the loop with distributing query
information is not perfect, but it shows there are other possibilities
to solving the discovery and query federation problem than just brute
force and large data stores.

Another alternative to predicate based query federations are the URI
prefix solutions that I have been developing for the Bio2RDF server
(although they can be applicable to any domain). The set of providers
at [1] can be used in a similar way to the predicate and type
federations described by Sandro here. If you configure queries in [1]
without regard to the URI prefixes the result is similar to predicate
based distribution systems. The configuration information is also
lightweight enough to be distributed easily.

For almost any query that is resolvable with the Bio2RDF server you
can fetch the instructions using URI's like [2] and perform the query
resolution process yourself using the same algorithm that the server
would have used by finishing the resolution. The process is simply
performed for any RDF URI's that come back from [2] with the rdf:type
[3]. Being able to have several servers that enable RDF query planning
without needing to perform any resolution or data storage themselves
might be a good step towards common federation models in a lightweight
manner.

Cheers,

Peter

[1] http://qut.bio2rdf.org/admin/configuration/rdfxml
[2] http://qut.bio2rdf.org/queryplan/pageoffset1/links/geneid:11234
[3] http://bio2rdf.org/ns/querybundle#QueryBundle

Received on Sunday, 17 May 2009 02:09:04 UTC