Review of RIF in RDF

Review of RIF in RDF

The following is a brief review of the RIF in RDF document.

Overall comment:  This is overall a good document, but my sense is that it
is preliminary in several respects.

Good points:
1. Clear writing. Objective is clearly laid out, and for the most part, it
is clear what has and has not been achieved.

2. Use cases are very helpful in understanding why one might wish to have
this mapping. (On the other hand, as discussed below, some use cases could
benefit from added discussion/motivation/explanation.)

3. Detailed tables giving the mapping are very helpful.

Areas to modify/improve:

(1) Use cases could use some extra discussion/motivation. For example, in
UC3: "It may be desirable in some cases to process them as RDF triples or
as RIF frames."  Why is this so? You say UC4 is a special case of UC3, but
it's still not clear how and why this is useful.

(2) Examples are needed! Simple examples, complex examples, in all their
ugliness and detail. Show the RIF doc in presentation syntax, then show
the XML, then show how the mapping is used to get it to the RDF structure.
Then, if possible, show what the reverse mapping would do. Give examples
of where roundtripping causes difficulty. Which leads to the next point:

(3) The question that was foremost in my mind as I read this was: how are
general rules, especially rules with universal and existential
quantifiers, expressible as RDF triples? This is doubtless already known
to RDF experts. Nevertheless, it would be helpful to take some examples of
complex rules and explain the transformation, relative to the rules given.


Regards,
Leora

-- 
Leora Morgenstern, Ph.D.
http://www-formal.stanford.edu/leora
646.872.7269

Received on Tuesday, 14 September 2010 15:01:22 UTC