You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jena.apache.org by Andy Seaborne <an...@apache.org> on 2020/05/01 10:21:05 UTC
Re: Lucene integration
(Sorry for the delay)
On 28/04/2020 16:47, Nouwt, B. (Barry) wrote:
> Hi all, I have a short question about the Lucene Index that can be integrated with an Apache Jena Dataset (https://jena.apache.org/documentation/query/text-query.html#).
>
> Would inferred triples (by a forward GenericRuleReasoner) also get indexed by the Lucene index?
I don't think so in the general case.
The inferred triples are kept separately from the base graph so the
configuration of the base as having a text index is not used. (Maybe
someone can correct me here - it would be good if it were otherwise.)
A special case might be if the inferred triples use literals already in
the base data, then they are indexes via the base graph so the pattern
## In the base graph
(?s ?score ?literal) text:query (rdfs:label 'word') ;
## In the inference graph
?x rdfs:label ?literal .
might ( I have not tried it) help.
It does the search then looks up the ?literal found which could drive
the rule's machinery.
For forward rules producing triples, the inferences could be
materialised and then indexed separately.
Andy
>
> Regards, Barry
Re: Lucene integration
Posted by Nicola Vitucci <ni...@gmail.com>.
This matches my results as well. If the TextDataset wraps a model, and
triples are added to such model causing an update to the base data, your
query works.
Nicola
Il giorno ven 1 mag 2020 alle ore 11:21 Andy Seaborne <an...@apache.org> ha
scritto:
> (Sorry for the delay)
>
> On 28/04/2020 16:47, Nouwt, B. (Barry) wrote:
> > Hi all, I have a short question about the Lucene Index that can be
> integrated with an Apache Jena Dataset (
> https://jena.apache.org/documentation/query/text-query.html#).
> >
> > Would inferred triples (by a forward GenericRuleReasoner) also get
> indexed by the Lucene index?
>
> I don't think so in the general case.
>
> The inferred triples are kept separately from the base graph so the
> configuration of the base as having a text index is not used. (Maybe
> someone can correct me here - it would be good if it were otherwise.)
>
> A special case might be if the inferred triples use literals already in
> the base data, then they are indexes via the base graph so the pattern
>
> ## In the base graph
> (?s ?score ?literal) text:query (rdfs:label 'word') ;
> ## In the inference graph
> ?x rdfs:label ?literal .
>
> might ( I have not tried it) help.
>
> It does the search then looks up the ?literal found which could drive
> the rule's machinery.
>
> For forward rules producing triples, the inferences could be
> materialised and then indexed separately.
>
> Andy
>
> >
> > Regards, Barry
>