[Lucene.Net] Porting SIREn to Lucene.Net

[Rob Vesse <rv...@dotnetrdf.org>]

Hi again
I got a response back from SIREn folks and it seems the only Lucene 3.x 
specific features they rely on are the new TokenStream APIs which was 
introduced in Lucene 2.9 afaict.
I don't have time to work on this immediately as I'm currently doing some 
work on integrating basic Lucene.Net functionality with dotNetRDF but as I 
said before I'm interested in helping with this at some stage so if anyone 
wants to set up a project site for this somewhere please let me know.
Regards,
Rob Vesse

----------------------------------------
From: "Rob Vesse" <rv...@dotnetrdf.org>
Sent: 16 August 2011 13:46
To: lucene-net-user@lucene.apache.org
Subject: Re: [Lucene.Net] Combining Lucene.Net and dotNetRDF? 

Hi Troy

Ok, I've got in touch with the SIREn guys to ask about whether it uses 
Lucene 3.x specific features or not so will get back to you on that front 
once I know more.  Full text SPARQL support is a feature that I'm planning 

on adding later this year and can be done purely with Lucene.Net afaict but 

having SIREn ported as well would obviously provide more useful indexes 
that could be used for other things besides full text search.  If you and 
anyone else who is familiar with Lucene.Net would be interesting in 
collaborating on this (assuming it is possible) then I'd be up for that.

Btw should it turn out that Lucene 3.x features are required is there any 
current planned timescale as to when Lucene.Net will reach version 3.0?

Interesting approach to the LINQ thing, haven't seen anyone else doing 
that.  Did see some guys in Oxford recently who are working on a new LINQ 
to SPARQL based on dotNetRDF but their approach is very much the 
translating the LINQ tree to the SPARQL string a la LinqToRdf and my own 
incomplete unpublished port of the same.  I'm not much of a LINQ expert 
hence why I haven't tried to do something in that area myself so not sure 
how helpful I can be but I'm always happy to answer questions, take feature 

requests, contributions, patches etc as necessary.

Best Regards,

Rob Vesse

----------------------------------------
From: "Troy Howard" 
Sent: 11 August 2011 23:46
To: lucene-net-user@lucene.apache.org, rvesse@dotnetrdf.org
Subject: Re: [Lucene.Net] Combining Lucene.Net and dotNetRDF? 

Hi Rob!

That was me (as well as Eric Sterling and Adron Hall). Unfortunately I
don't have anything ready to publish yet from that work.

Regarding SIREn, I've looked into it, and I'm considering porting it.
One possible barrier is that it was developed against Lucene 3.0, and
Lucene.Net is currently at 2.9.4 compatibility. Porting may be
difficult, depending on wether they used features that are unique to
Lucene 3.0, and wether there is a reasonable way to port code that
uses those 3.0 features back to 2.9.4 API.

I think it's a great idea to start a SIREn porting project, and if
you're interested in being involved in that, we could put together a
quick project site on bitbucket (my preference) or github.

One of my intentions was to make a network accessible triple store
(e.g. database server) using ZMQ (for comms) and Lucene.Net (for the
triple store) and dotNetRDF/LinqToRDF for the queries. My intention
was to start with LinqToRDF, serializing an expression tree, sending
that across the wire to the database service, as opposed to using a
typical database server interface using a string based query in a
human readable DSL.

The downside of that approach is that clients would have to be .NET
specific (or be able to serialize a native expression tree such that
it could be deserialized into a LinqToRDF tree)... So the
text-based-query approach is more portable. Porting SIREn could be a
big help towards that goal. Ultimately I'd like to support both
interfaces, but my intention was to start with the Linq interface, and
avoid the whole process that most Linq interfaces have (which is an
intermediate serialization of the query into some kind of DSL like SQL
or XML or whatever..).

I'm pretty flexible about that though.

Thanks,
Troy

On Thu, Aug 11, 2011 at 8:51 AM, Rob Vesse  wrote:
> Hi All
>
> I wondered if anyone had already tried to integrate dotNetRDF and
> Lucene.Net since I noticed one of your contributors was involved in a
> hackathon [1] related to my project [2] back in June.  So if anyone has 
any
> existing code for this that they are willing to share I'd be interested 
to
> see it.
>
> On a similar note has anyone looked at converting SIREn (a plugin for
> Lucene in pure Java) that is designed specifically for RDF indexing to
> .Net?  It's likely that this may get done anyway as part of adding 
Lucene
> integration to dotNetRDF but again don't want to reinvent the wheel if
> somebody is already looking at this.
>
> Regards,
>
> Rob Vesse
>
> [1] http://opensourcebridge.org/sessions/730
> [2] http://www.dotnetrdf.org
>
>
>