Review of rif-rdf-owl-compatibility

Hi all, here is my review of

http://www.w3.org/2005/rules/wg/draft/rif-rdf-owl/

First, it is a rather thoroughly written, informative, and substantial  
document. I have a couple of minor remarks and suggestions (see  
below), and 2 more important ones:

- the document focuses on OWL (full and DL), and leaves OWL 2 as an  
end note. I would think that OWL 2 is now sufficiently stable to be  
mentioned in a more detailed way, especially since the changes to OWL  
that are relevant for this document seem to be only minor.  Most  
importantly, this would allow to replace/extend the DLP embedding in  
the appendix with an OWL 2 RL embedding: this would make things far  
more transparent to the reader since it would link the embedding to a  
known profile.

- the document is very substantial, and this seems to be due to the  
fact that it (i) repeats relevant notions from other documents to make  
it self contained and (ii) mixes examples and discussions into the  
technical part. In my understanding, (i) is unavoidable, but (ii)  
could be streamlined: it would make a lot of sense to have all  
examples and discussions in one place -- so as to allow to start by  
reading through examples or to skip them if not interested.

Minor remarks:

- I would avoid calling 'ontologies' 'data models': this might give a  
wrong impression and confuse readers and is un-necessary.  The term  
'data models' occurs in various places as synonym for  
'ontology' (e.g., as "OWL data models" in the abstract)-- I would  
think that referring the reader who doesn't know what an ontology is  
to the OWL 2 primer would be more useful.

- in Section 1, you say 'RDF data and RDFS and OWL ontologies are  
represented using RDF graphs.' ...this is not always the case since  
there are alternative syntaxes, and this should be made clear, e.g.,  
'RDF data and RDFS and OWL ontologies can be represented using RDF  
graphs (for OWL and OWL 2, various alternative syntaxes).

- I would suggest to scope the sentence 'which can be seen as a guide  
to describing how a RIF processor could be turned into an RDF/OWL- 
aware RIF processor.' (and this scoping would be made easier if the  
OWL 2 RL  was used!

- you write "as well as the RDF triple :john :uncleOf :mary, can be  
derived." but this now sounds like a deductive statement (i.e, by  
which algorithm can they be derived). I would phrase this semantically  
"The RIF frame formula :john[:uncleOf -> :mary], as well as the RDF  
triple :john :uncleOf :mary,  are consequences of this combination,  
and should be derived by an appropriate reasoning engine."

- "This section specifies how a RIF document interacts with a set of  
RDF graphs in a RIF-RDF combination. In other words, how rules can  
"access" data in the RDF graphs and how additional conclusions that  
may be drawn from the RIF rules are reflected in the RDF graphs." and  
I have no idea how this 'reflection' should work or what it means? One  
doesn't *have to* materialize the consequences in the original graph?!

- Capitalisation of RDF, RDFS, and OWL doesn't seem to be uniformly  
handled in the document

- in chapter 4, I would stronly suggest to define the notion of a RIF- 
OWL-combination as one that consists of a RIF document and an ontology  
O?

- you write "This is a key property of Description Logic semantics  
and  ", whereas i see it as "This is a key property of the first order  
logic nature of Description Logic semantics and": wouldn't this be  
clearer?

- throughout the document, each "Definition" environment seems to be  
defining only a single term. I would suggest to group successive  
"Definitions" into a single one if they are related (I would think  
this enhances readability)

- why is 'safeness' at risk? One could discuss whether a "make the  
combination safe automatically" would be a useful option (by adding  
the usual HU(x) predicates), but I think that safety is an essential  
feature of this document that must be kept!

- "The above definition of DL-safeness is intended to identify a  
fragment of RIF-OWL DL combinations for which implementation is easier  
than full RIF-OWL DL. " -- I disagree: implementation doesn't become  
easier at all, this is only about decidability, i.e., "for which  
reasoning is still decidable".

-  "EC(c) = set of all objects k such that..." i would suggest to  
change this to "EC(c) is the set of all objects k such that" (because  
the "the" is important and the = confusing)

- in 5.2, "Let ..be such that ....are the two-ary import statements in  
R and all imported documents and" should be rephrased to  "Let ..be  
such that ....are two-ary import statements in R and R contains no  
other imported documents and" ?!

- the table in section 8.1.5 contains many occurrences of the same  
rule...why?

Received on Wednesday, 26 November 2008 16:03:06 UTC