Re: Call for comments on your implementation experience

From: Jim Hendler <hendler@cs.umd.edu>
Subject: Call for comments on your implementation experience
Date: Fri, 7 Nov 2003 10:09:40 -0500

> 
> it is the opinion of the chair of WOWG (me) that we have now met all 
> of our exit criteria (modulo potential change in RDF) and thus I am 
> going to begin preparing materials for our eventual move to Proposed 
> Recommendation -- one thing that would help is if the implementors of 
> OWL reasoners, ontologies, or etc. were to share with us their 
> experiences -- this is especially valuable (but not limited to) those 
> who have contributed their results to our test documents.
>   We are expecially interested in feedback with respect to whether you 
> were able to implement the langauge based on the documents (were we 
> clear), were there other people's tools, products, techniques you 
> found valuable, and how long did it take you and how hard was it (in 
> qualitative terms is fine).
>   Unclear to me exactly how I will use this information, but I would 
> prefer responses to this list, and in public, so that I am free to 
> forward what you write to the Working Group and/or other W3C lists.
>   thanks
>   Jim Hendler
> p.s. Please feel free (and encouraged) to circulate this message to 
> all and sundry who may have implemented in this space

I had no severe problems in implementing OWLP, either the parsing side or
the reasoning side.  I did not find the OWL documents difficult to read.
:-)

The biggest problems I encountered in implementing the parsing part of OWLP
had to do with the RDF syntax.  Implementing a parser for the abstract
syntax was much easier than implementing a parser for the RDF/XML syntax.
This had less to do with the OWL encoding in RDF than it had to do with RDF
itself.

The biggest problems I encountered in implementing the reasoning part of
OWLP had to do with the RDF embedding, particularly handling annotations
as real parts of the meaning of an OWL ontology.  The other parts of the
(partial) reasoner in OWLP were easy to handle by passing the reasoning off
to a description logic reasoner.

The total effort to implement OWLP was about three months.

Peter F. Patel-Schneider
Bell Labs Research

Received on Monday, 10 November 2003 14:37:12 UTC