RE: SWRL builtinAtom/List inconsistency

From: Benjamin Grosof (bgrosof@mit.edu)
Date: 11/16/04

  • Next message: Mike Dean: "RE: SWRL sameAs/differentFrom inconsistency"
    Hi Mike and all,
    
    At 08:30 AM 11/16/2004 -0500, Mike Dean wrote:
    > > 1. what was the design rationale for why the List builtins currently
    > > require use of i- ?
    >
    >A list is an instance of rdf:List, so any property used to refer to a list
    >must be an ObjectProperty, which requires use of an iObject.
    >
    >I think we just missed this subtlety in the builtin discussions.  Support
    >for lists is important for SWRL.
    >
    > > 2. perhaps we should define another kind of list that has d- members or
    > > that mixes i- with d- members?
    >
    >That doesn't really help (I explored it as a test case).  Consider a Lock
    >with a combination property that's a list of xsd:integers.  combination
    >still needs to be an ObjectProperty.
    
    Hmmm.  I see the issue more clearly now.
    Let's discuss on a JC call soon.
    We might want to change the current design to have each builtin's arguments 
    specified one-by-one as data or object properties.  This is somewhat clumsy 
    in that it requires in effect more detailed signature specifications for 
    the builtins.  But that's probably easier than defining whole new kinds of 
    lists and then new kinds of properties that have such lists as their arguments.
    
    Benjamin
    
    
    >         Mike
    
    ________________________________________________________________________________________________
    Prof. Benjamin Grosof
    Web Technologies for E-Commerce, Business Policies, E-Contracting, Rules, 
    XML, Agents, Semantic Web Services
    MIT Sloan School of Management, Information Technology group
    http://ebusiness.mit.edu/bgrosof or http://www.mit.edu/~bgrosof
    


    This archive was generated by hypermail 2.1.4 : 11/16/04 EST