Hi Patrick! Now I see the problem, thanks!
Just in case I'll be in a need of speeding up this kind of queries for
my storage: can I reduce this problem with indexing?
Sincerely yours,
-----
Yury Katkov
On Mon, Mar 12, 2012 at 2:38 AM, Patrick van Kleef
<pkl...@openlinksw.com> wrote:
> Hi Yury,
>
>
>> I find extremely useful to execute some SPARQL queries to get familiar
>> with the dataset. One of these queries is default for Virtuoso (get
>> all classes), I also like this one:
>> ############
>> select distinct ?y WHERE
>> {
>> ?x ?y ?z
>> }
>> LIMIT 100
>> ###########
>> It was very surprising when I found that such a simple query will lead
>> to "Virtuoso S1T00 Error SR171: Transaction timed out" on
>> http://dbpedia.org/sparql . Is it possible to speed it up a little?
>
>
> You are basically doing a table scan, on all of the triples in the db to
> make a hash of all the unique values of P, then return the first 100. The
> full execution time for this type of query is larger than we currently allow
> any query to run for on dbpedia.org, depending on how many other substantial
> queries are running at the same time.
>
> What you can do on any Virtuoso /sparql endpoint, is to use an ANYTIME
> query, by filling in the Execution Timeout field on the form:
>
> http://dbpedia.org/c/XQOOXC
>
> If you click on the above link and press the "Run Query" button, Virtuoso
> will collect as many unique values of P it can find in 50 seconds and then
> return up to a maximum of 100 results, depending on how many unique values
> it collected in that time period.
>
>
>
> Patrick
> ---
> OpenLink Software
------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
