[Mulgara-dev] 100k models?

[Paul Gearon]

This may be problematic, as we have been systematically removing Jena
support.  We have a number of reasons for this (which I'm happy to go into),
but it really comes down to the horrible performance created by Jena.

What do you mean by "query a set of models, in sequence or in union"?  You
can explicitly state that you want your query resolved against a union of
models, or an intersection (or an intersection of unions, etc...).
Alternatively, you can even describe a set of properties of models in
another model (I suppose this would be a metamodel, but Mulgara won't know
the difference), and then issue a query which determines which models to
query against, and to union the results from each.

The combinations are pretty flexible.  So what exactly are you looking for?

Paul

On 4/5/07, Laurian Gridinoc <laurian at gmail.com> wrote:
>
> Hello,
>
> On 05/04/07, David Wood <dwood at softwarememetics.com> wrote:
>
> > Mulgara scales well in terms of model numbers.  We have seen people try
> > it with hundreds of thousands of models and get decent performance.
> >
>
> However, Jena+Mulgara is *not* recommended.  May I ask what you are trying
> > to accomplish?  Perhaps we could recommend a different approach.
> >
>
> Well, first we have developed  around the Jena API... OK, Watson is a
> search engined designed to crawl, index and search ontologies (in a way like
> Swoogle, but we have different goals).
>
> The crawling and the "validation" (ontology validation, consistency, etc.)
> parts use Jena, with some particular things:
>
> 1. the crawler, not only crawls and creates models, but provides it own
> SAX stream to Jena, such that I can get some serialization (rdf/xml) level
> metrics, such as xml:base, relative rdf:about, etc.
>
> 2. the validation reads the models and use Pellet as reasoner to measure
> interesting things.
>
> As I can easily remove the Jena dependency from the crawler, it would be
> harder from the validation procedure, which relies on reasoners and on Jena
> Ontology API.
>
> Once we validate the models, we create our own databased model and lucene
> indexes over the extracted information, we don't use further the underlying
> Jena model.
>
> BTW, is any way to query a set of models, in sequence or as an union?
>
> Cheers,
> --
> Laurian Gridinoc, purl.org/net/laur
> _______________________________________________
> Mulgara-dev mailing list
> Mulgara-dev at mulgara.org
> http://mulgara.org/mailman/listinfo/mulgara-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mulgara.org/pipermail/mulgara-dev/attachments/20070405/9ccef4e9/attachment.htm>