On Thu, 12 Apr 2012 23:33:45 +0200, Paul A. Houle <[email protected]>
wrote:
> On 4/12/2012 7:59 AM, baran_H wrote:
>> If ther isn't yet such a SPARQL-endpoint-link for :BaseKB ('comparable
>> to DBpedia in scope'), is it planned?
> I'll say two things:
>
> (1) It will cost about $7000 a year to provide a public SPARQL
> endpoint. As I don't have venture capital financing or an endowed chair
> at a university, I need to know this is investment that pays off.
>
> (2) I've been studying the barriers to semantic web adoption and
> identified the public SPARQL endpoint as one of them. Surveys show that
> the average public SPARQL endpoint listed in the LOD cloud is down --
> it's difficult and expensive to provide a SPARQL endpoint that is
> reliable. The DBpedia endpoint, for instance, lacks some of the most
> interesting data sets in DBpedia (to keep costs down) and is down quite
> often. answers.semanticweb.com has a drumbeat of questions by people
> who are asking why their queries time out.
>
> An obvious thing is to charge people for API calls, but that's
> difficult. The trouble is the cost of a SPARQL call can vary by a
> factor of 10,000 or more. I can't set a price which is fair for
> everybody.
>
> Hypothetically, a system could try to account for the cost of queries
> but it's a harder problem than it looks. If multiple people are using
> the store, it's not always possible to figure out who is responsible
> for what. For instance, you query might be delayed 20 seconds because
> some other guy filled up the RAM with other stuff with the queries he's
> doing. Superficially it looks like you ran an expensive query but it
> wasn't your fault.
>
> If query cost was accounted for accurately I think users would have
> anxiety about their costs not being predictable. I'd also feel bad too
> if people got punished for bugs in the query optimizer.
>
>> Or was
>>
>> 'you can query Freebase with the powerful SPARQL 1.1 language and get
>> back as many query results as you want and not be afraid that your
>> queries will time out.'
>>
>> only an ironical statement?
>>
>>
> Not at all. Install :BaseKB on your own hardware, pay for what
> you need, provide your own SLA, and don't have to deal with some other
> guy making queries that slow you down. If I give you a SPARQL endpoint
> I also have to give you timeouts to protect the other users. If you're
> the only user you're in control -- you can load :BaseKB together with
> any other data you want, only load part of :BaseKB, etc.
--------------------------------------------------------------------------
Thank you very much for your detailed and very informative reply.
Because you write you have been studying the barriers to semantic
web adoption and identified the public SPARQL endpoint as one of
them, I have some questions to you.
Anticipating the serious problems of public SPARQL endpoints you
seem to say: Everybody can download the freebase-RDF-datasets,
install his/her SPARQL endpoint on her/his harddisk and work with
it 'locally' as single user.
a.) Do you see local installations only as a temporary solution
until public SPARQL endpoints get more powerful and cheaper in
the future?
b.)Or do you seriously think on a general Linked Data concept
basing more and more on downloaded RDF-datasets and locally
installed SPARQL points with all consequences, e.g. conceptually
and heavily constraining the potential of web-wide querying
crowds?
c.)Would public SPARQL endpoints 'in the final analysis' much
more powerful and cheaper if they are (forget SQL completely)
implemented with a small subset of SPARQL which allows only a
class of fundamental and relative simple queries, algorithmic
optimized for high performance, supported by various indexing
methods etc.?
Thanks, Baran.
PS:Because i am alone not so important for such a detailed mail,
i assume(!) Paul sent his mail accidently only to me and not to
the list. Therefore i reply to the list with his complete mail.
Sorry if he thought, such a fool doesn't deserve a list-reply,
in this case he can perhaps reply again only to me, no problem.
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2
_______________________________________________
Dbpedia-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
