Re: Metadata Store

["Enrico Minack" <minack l3s de>]

Hi Joe,

thanks for the extensive answer, now here is mine.

> > I put some effort in getting Sesame and Sesame2 [1] working under C# and 
> > it
> > worked quite well.
> Using IKVM or a source translation?
I used IKVM. The dll files can be found here:
   http://sourceforge.net/projects/dotsesame


> The more interesting memory benchmark for us is Sesame vs. no Sesame.
> What are the differences there?
I just mentioned that in order to address your concerns (in a previous mail) 
that the memory consumption might increase due to ikvm.

> Yes.  The Lucene index and the metadata store will be paired and (at
> least initially) be one each per backend.
The problem having a metadata store for each backend would lack the 
possibility to connect different types of documents through their metadata, 
e.g. a file (filesystem queryable) which is an attachment of an email 
(evolution queryable). Or at least these connections would be more 
difficuilt to exploit on query time.

> All text searches will be done through lucene.  All metadata searches
> will be done through the store.
I heared about a Lucene Sail, that is a layer for Sesame. This would mean 
that querying the RDF store could be based on Lucene. This could allow 
querying only one store, the RDF store, where keyword matching is 
implemented by Lucene. I think this sounds interesting enough for further 
investigation.

> Pluggable filters and backends make sense because people can drop in or
> remove the ones they want to use.   ... I don't see that with a pluggable 
> RDF
> store.
I see your point and I agree. But what I am having in mind is the following:
I need to get Beagle connected to the NEPOMUK architecture and I am thinking 
about how to get this in a nice way which is both nicely applicable and 
accepted by the Beagle developers as well. This NEPOMUK architecture (which 
is like a backbone of components and services) will, among others, provide a 
central RDF store, which is shared among all applications using that NEPOMUK 
architecture. The backbone also wants to use the desktop search services 
provided by Beagle and expose that to the components of NEPOMUK and to all 
other NEPOMUK-enabled applications. Right now I am thinking of two 
solutions:
1) Beagle's rdf store is connected through a API and is easy to be replaced 
by an other rdf store that complies to that API. Thus, in order to 
NEPOMUK-enable Beagle one just need to replace that rdf store. Then all 
metadata could be stored in that rdf store.
2) putting all our stuff into our own backend which uses the NEPOMUK 
architecture and thus that rdf store. Then all metadata generated by this 
backend would not be stored at Beagle but at the NEPOMUKs store. Beagle 
queries forwarded to the backend will then be used to query the NEPOMUK 
store. A problem will be the connection of those metadata to the documents 
of other backends.

best regards,
Enrico M.