- From: Michael Kifer <kifer@cs.sunysb.edu>
- Date: Tue, 29 Jul 2008 12:26:44 -0400
- To: Christian de Sainte Marie <csma@ilog.fr>
- Cc: RIF WG Public list <public-rif-wg@w3.org>
Christian,
Here is my basic line of thinking about the use of DTB "as is"
for the purposes of PRD.
PRD's semantics can be thought of as incremental computation of certain
semantic structures. Bodies of the rules in a PRD document are grounded in the
current semantic structure ("matched" in the PRD terminology) and new facts are
produced and ssome old ones are deleted.
Builtins are used to determine truth values of the rule bodies at the
grounding/matching stage. The grounding process is entirely captured by the
semantics of the builtins in DTB, since at this point one just needs to know
whether a builtin predicate is true in the current semantic structure or what
is the value of a builtin function in the current structure. So, DTB can be
used directly in PRD, if the language in PRD is adapted to match the above way
of looking at things.
In fact, I think the semantics of the condition formulas in BLD could be used
in the same way as above to make a stronger connection between PRD and BLD. The
only major thing in BLD that is not applicable to PRD is the semantics for the
PRD rules.
--michael
Received on Tuesday, 29 July 2008 16:27:26 UTC