You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stanbol.apache.org by Alessandro Adamou <al...@studio.unibo.it> on 2014/06/10 17:01:13 UTC

Named graph support in Referenced Sites

Hi,

I've been looking into configuring a Referenced Site (fully live, 
non-cached and entirely SPARQL-queried) into a running EntityHub.

Problem is, the target RDF store is fully quad-based, all the data are 
in named graphs and the configuration of the default graph is not to 
compute the union. And since the default graph is queried, only 
quad-pattern queries are satisfied.

According to the RDF store log, the entity Hub queries are of the form, 
e.g. for a 'find':

CONSTRUCT {
     ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1 .
<http://stanbol.apache.org/ontology/entityhub/query#QueryResultSet> 
<http://stanbol.apache.org/ontology/entityhub/query#queryResult> ?id .
} WHERE {
     { ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1
         FILTER(regex(str(?v_1),"Paderb.*","i")) .
}  }

A query like this will never have any results in the RDF store I'm 
using. I'm wondering if there is any way to configure queries and/or 
Sites so that they can:
* supply named graphs as FROM [NAMED] clauses in the queries, and/or
* issue quad patterns a la "GRAPH ?g { }" in queries (and include them 
for every triple in the result set from the entity hub)

If these are not supported, it might be worth opening a ticket for them?

Cheers

Alessandro

Re: Named graph support in Referenced Sites

Posted by Rupert Westenthaler <ru...@gmail.com>.
Hi Alessandro

Quads are not supported.

>> If these are not supported, it might be worth opening a ticket for them?

Yep. Even better if you could provide examples on how to fix the
generated SPARQL queries. The SPARQL queries are generated by [1]. For
adding this feature one would need to

1. extends [1] to support named graphs
2. add a new property to Referenced Sites that allows to configure
named graphs URIs

best
Rupert

[1] http://svn.apache.org/repos/asf/stanbol/trunk/entityhub/query/sparql/src/main/java/org/apache/stanbol/entityhub/query/sparql/SparqlQueryUtils.java



On Tue, Jun 10, 2014 at 5:03 PM, Alessandro Adamou
<al...@studio.unibo.it> wrote:
> Sorry, I meant that my default graph is always *empty*
>
> Alesandro
>
>
>
> On 10/06/2014 16:01, Alessandro Adamou wrote:
>>
>> Hi,
>>
>> I've been looking into configuring a Referenced Site (fully live,
>> non-cached and entirely SPARQL-queried) into a running EntityHub.
>>
>> Problem is, the target RDF store is fully quad-based, all the data are in
>> named graphs and the configuration of the default graph is not to compute
>> the union. And since the default graph is queried, only quad-pattern queries
>> are satisfied.
>>
>> According to the RDF store log, the entity Hub queries are of the form,
>> e.g. for a 'find':
>>
>> CONSTRUCT {
>>     ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1 .
>> <http://stanbol.apache.org/ontology/entityhub/query#QueryResultSet>
>> <http://stanbol.apache.org/ontology/entityhub/query#queryResult> ?id .
>> } WHERE {
>>     { ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1
>>         FILTER(regex(str(?v_1),"Paderb.*","i")) .
>> }  }
>>
>> A query like this will never have any results in the RDF store I'm using.
>> I'm wondering if there is any way to configure queries and/or Sites so that
>> they can:
>> * supply named graphs as FROM [NAMED] clauses in the queries, and/or
>> * issue quad patterns a la "GRAPH ?g { }" in queries (and include them for
>> every triple in the result set from the entity hub)
>>
>> If these are not supported, it might be worth opening a ticket for them?
>>
>> Cheers
>>
>> Alessandro
>> .
>>
>



-- 
| Rupert Westenthaler             rupert.westenthaler@gmail.com
| Bodenlehenstraße 11                              ++43-699-11108907
| A-5500 Bischofshofen
| REDLINK.CO ..........................................................................
| http://redlink.co/

Re: Named graph support in Referenced Sites

Posted by Alessandro Adamou <al...@studio.unibo.it>.
Sorry, I meant that my default graph is always *empty*

Alesandro


On 10/06/2014 16:01, Alessandro Adamou wrote:
> Hi,
>
> I've been looking into configuring a Referenced Site (fully live, 
> non-cached and entirely SPARQL-queried) into a running EntityHub.
>
> Problem is, the target RDF store is fully quad-based, all the data are 
> in named graphs and the configuration of the default graph is not to 
> compute the union. And since the default graph is queried, only 
> quad-pattern queries are satisfied.
>
> According to the RDF store log, the entity Hub queries are of the 
> form, e.g. for a 'find':
>
> CONSTRUCT {
>     ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1 .
> <http://stanbol.apache.org/ontology/entityhub/query#QueryResultSet> 
> <http://stanbol.apache.org/ontology/entityhub/query#queryResult> ?id .
> } WHERE {
>     { ?id <http://www.w3.org/2000/01/rdf-schema#label> ?v_1
>         FILTER(regex(str(?v_1),"Paderb.*","i")) .
> }  }
>
> A query like this will never have any results in the RDF store I'm 
> using. I'm wondering if there is any way to configure queries and/or 
> Sites so that they can:
> * supply named graphs as FROM [NAMED] clauses in the queries, and/or
> * issue quad patterns a la "GRAPH ?g { }" in queries (and include them 
> for every triple in the result set from the entity hub)
>
> If these are not supported, it might be worth opening a ticket for them?
>
> Cheers
>
> Alessandro
> .
>