- From: Hassan Aït-Kaci <hak@ilog.com>
- Date: Mon, 21 Jan 2008 17:40:35 +0100
- To: axel@polleres.net
- CC: Christian de Sainte Marie <csma@ilog.fr>, Jos de Bruijn <debruijn@inf.unibz.it>, RIF <public-rif-wg@w3.org>
Axel Polleres wrote: > whereas, on the contrary in frames, any slot can appear 0 or several > times, i.e. > > person1[firstname -> "Christian", lastname -> "de Sainte Marie"] > > and > > person1[firstname -> "Christian"] AND person1[lastname -> "de Sainte > Marie"] > > say the exactly same thing, i.e. mutually entail each other. What? You mean that there can be only one "person1" object named "Christian"? IMHO, what you are describing corresponds to the notion of database record key(s) - i.e., the attribute(s) that characterize individual records. Furthermore, even if you do have a notion of keys (which we have not discussed within RIF as far as I recall), there is still another situation to take into consideration: just like algebraic (i.e., positional) terms are built of symbols that may have (or not) well-formedness constraints (e.g., signatures, typing, annotations, etc...). To make things even more interesting, not everyone agrees with one specific semantics for the very same frame syntax (i.e., do we allow repeated "slots" - and in this case, is this an error or is it that such a slot has as value the set (or some kind of aggregate) of all the values for this slot. Some also allow mandatory as well as non-mandatory attributes - some even allow any symbol as a slot (i.e., unconstrained signatures). While the RIF BLD masterminds (i.e., the authors of the document: Harold and Michael) give one possible semantics for these syntactic constructs; but this semantics is not necessarily compatible with many extant rule languages, several of which would wish to use the RIF. -hak -- Hassan Aït-Kaci * ILOG, Inc. - Product Division R&D http://koala.ilog.fr/wiki/bin/view/Main/HassanAitKaci
Received on Monday, 21 January 2008 16:42:06 UTC