You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jena.apache.org by Howard Burrows <ho...@gmail.com> on 2013/08/07 16:48:15 UTC
Text search and named graphs
Is the text search module named graph aware and capable of being configured (assembled) such that each named graph in a dataset can have its own text indexes configured separately with their own EntityMap definitions?
I have a TDB dataset composed of multiple named graphs (i.e. one per 'vendor') which I would like to index using the text search module and then search using named graph semantics. Some of my SPARQL queries reference all graphs and some reference just a single graph so splitting the data into multiple datasets each with a separate Fuseki service endpoint wouldn't work well.
Any advice on how to assemble such a solution would be welcomed and appreciated.
Thanks,
Howard Burrows
Re: Text search and named graphs
Posted by Andy Seaborne <an...@apache.org>.
On 07/08/13 15:48, Howard Burrows wrote:
> Is the text search module named graph aware and capable of being
> configured (assembled) such that each named graph in a dataset can
> have its own text indexes configured separately with their own
> EntityMap definitions?
The index is simply URI and text string, with no concept of which graph
it comes from.
What you may wish to do is:
SELECT ?s
{ ?s text:query (rdfs:label 'word' 10) ;
rdfs:label ?label
}
which looks the text query up and then checks to see if the URI returned
actually does have a triple of that property.
So:
SELECT ?s
{
GRAPH :graph {
?s text:query (rdfs:label 'word' 10) ;
rdfs:label ?label .
}
}
Not exactly the same but similar.
Otherwise, you'll need to create separate indexes and have separate
property functions to access each one.
Andy
>
> I have a TDB dataset composed of multiple named graphs (i.e. one per
> 'vendor') which I would like to index using the text search module
> and then search using named graph semantics. Some of my SPARQL
> queries reference all graphs and some reference just a single graph
> so splitting the data into multiple datasets each with a separate
> Fuseki service endpoint wouldn't work well.
>
> Any advice on how to assemble such a solution would be welcomed and
> appreciated.
>
> Thanks,
>
> Howard Burrows
>