Hi Mark!

The DESCRIBE queries are actually graph queries -- the result of DESCRIBE 
query evaluation is a set of statements. That is, you can use the repository 
connection method for preparing a graph query. Actually you might use the 
more general prepareQuery() method which will return an instance of the 
GraphQuery interface.


Cheers,
Ivan

On Tuesday 29 December 2009 01:51:20 Mark Larkin wrote:
> How does one execute a DESCRIBE query. I see examples of Tuple and Graph
> queries in the user guide, but I don't see how to prepare/evaluate a
> DESCRIBE query.
>
>
>
> Mark Larkin
>
>
>
>
>
> From: Atanas Kiryakov [mailto:[email protected]]
> Sent: Friday, December 25, 2009 7:10 AM
> To: [email protected]; Mark Larkin
> Subject: Re: [Owlim-discussion] Walking the Ontology
>
>
>
> Dear Mark
>
>
>
> first: Merry Christmas to all OWLIM fans!
>
>
>
> I am afraid that there is no support for such functionality. Generally,
> one can do SPARQL DESCRIBE queries, which return connected graphs, which
> include the URIs bound in the process of query evaluation plus their
> "surrounding" - sort of molecules, which are likely to differ from one
> engine to another, because the specifications leaves it quite
> undetermined.
>
>
>
> SPARQL DESCRIBE is more limited than the functionality that you are
> sking for as long as:
>
> - you get the "surrounding" sub-graphs with respect to *any* properties
>
> - it is not recursive, i.e. you are getting the immediate neighbouhood,
> where the individual(s) of intrest appear as subjects.
>
>
>
> It is not clear from your email, but I assume you expect some sort of
> "recursive closure". Please, confirm
>
>
>
> After all, I would recommend you to implement the algorithm on your own,
> accessing the repository through the Sesame APIs. Working directly with
> the SAIL methods (e.g. getStatements...) should be really fast if you
> are working with embedded Sesame/OWLIM (the standard option). This way
> you get maximum flexibility and good speed. You will be able to decide
> on your own:
>
> - wether you want to traverse over inferred statements
>
> - do you handle blank nodes in some special way?
>
> - do you have time and volume constraints and what should happen when
> you reach them
>
>
>
> On the negative side, you will have to take care to avoid loops and few
> other things of the same sort, but for relatively small result sets,
> this is doable with a simple hash-set
>
>
>
> Regards
>
> Naso
>
>
>
> ----------------------------------------------------------
> Atanas Kiryakov
> Executive Director of Ontotext AD, http://www.ontotext.com
> Sirma Group, http://www.sirma.bg
> Phone: (+359 2) 974 61 44; Fax: 975 3226
> ----------------------------------------------------------
> Fortes fortuna adiuvat.
>
>       ----- Original Message -----
>
>       From: Mark Larkin <mailto:[email protected]>
>
>       To: [email protected]
>
>       Sent: Wednesday, December 23, 2009 5:09 PM
>
>       Subject: [Owlim-discussion] Walking the Ontology
>
>
>
>       Are there methods that allow one to walk a populated Ontology?
> Utilmately I would like to extract a submodel associated with a
> particular node or individual.
>
>
>
>       For example, given a populated Ontology (Schema+Data), I would
> like to find all nodes that connect to a particular individual via a
> series of properties.
>
>
>
>       I know I can write algorithms for this, but I can't help to
> think that there may already be methods in the Seame API that do this.
>
>
>
>       I understand there are Tuple and Graph Queries, but these don't
> actually return a connected subgraph.
>
>
>
>       Mark Larkin
>
>
>
>
>
>
>
>
>
>
>
>
> ________________________________
>
>
>       _______________________________________________
>       OWLIM-discussion mailing list
>       [email protected]
>       http://ontotext.com/mailman/listinfo/owlim-discussion


_______________________________________________
OWLIM-discussion mailing list
[email protected]
http://ontotext.com/mailman/listinfo/owlim-discussion

Reply via email to