Re: [Fwd: [Deri.ie-research] OWL2 XML Parser Library in C++]

On 27 Aug 2009, at 11:46, Antoine Zimmermann wrote:

> Dear all,
>
> FYI, implementations in DERI are progressing.

Awesome.

[snip]
> From: Gergely Lukacsy <gergely.lukacsy@deri.org>
> Date: 27 August 2009 09:45:08 BDT
> To: deri Research <deri.ie-research@lists.deri.org>
> Subject: [Deri.ie-research] OWL2 XML Parser Library in C++
>
>
> Dear All,
>
> As part of our efforts to re-implement an existing Prolog based DL
> reasoner to C++ we made an OWL 2 DL parser for the OWL XML syntax  
> in C+
> + from scratch.
> This does the job: transforms an input knowledge base into an internal
> Object-Oriented representation suitable for application dependent
> further processing.
>
> The implementation aims to be fast (atom tables, custom memory
> allocator, Google hash-map implementation, etc.), memory efficient
> (SAX parsing, etc) and robust (non-parseable elements are  simply
> discarded, error handling, etc.).

This is interesting. If we could settle on some common error handling  
conventions that would be good.

> And it is "home cooked", we are here
> to fix/extend it if needed.

This sounds very very cool.

> The parser is practically feature complete, but a comprehensive
> testing phase is missing. We plan to run it  on official W3C test
> cases to make it as conform as possible,  but  unfortunately, at this
> moment no such test cases exist for OWL2 XML.

What sort of test cases do you want? I presume you have serialization  
as well, so you could try roundtripping. We could also compare with  
the OWL API parser

> If anybody is interested in helping with the testing or would like  to
> use the parser in some project/research do not hesitate to let us  
> know.
>
> The parser is available in source or as pre-compiled static binary
> libraries for Windows and Max OS platforms. Example program is  
> included.
> The implementation, together with the developer API documentation
> generated by Doxygen, can be accessed on http://www.dlog-reasoner.org/

I'm looking forward to playing with it.

It would be interesting if it would serialize to RDF (which is the  
vastly simpler task compared to going the other way).

Cheers,
Bijan.

Received on Thursday, 27 August 2009 11:09:39 UTC