Re: extended inference use case

From: Deborah McGuinness (
Date: 02/04/03

  • Next message: pat hayes: "Re: extended inference use case"
    sorry - i could not get anything out sooner on use cases.  More will
    My use cases center around the a few scenarios.  I include one here.
    Topic - configuration.
    PROSE/QUESTAR used a description logic-based system to take a set of
    constraints and generate a complete and correct parts list along with a
    price.  PROSE/QUESTAR configured telecommunications equipment.  Because
    the domain was very complicated and non-intuitive, we also did a stereo
    configurator for pedagogical reasons.  We also did a wines and food
    configurator but some people liked the stereo configurator better for
    naturally occuring business problems.
    The rules were used in a number of ways.
     - once a set of constraints was recognized to be configuring a system
    of type x (as a result of recognition in the dl reasoner) then a set of
    rules fired adding more constraints onto the parts list.  This can be
    done without  a separate rule form in OWL today but in CLASSIC it was
    done by rules.
    - rules were used in order to compute values.  Once the system knew that
    it had x as a value for property y, then it might be able to calculate a
    new value for a property.
    -rules were also used as filters before proceeding.  Once something was
    recognized to be an instance of a class, then if it also passed another
    filter (possibly an unnamed concept description), then an action was
    [PROSE] Deborah L. McGuinness and Jon Wright. ``An Industrial Strength
    Description Logic-based Configurator Platform''. IEEE Intelligent
    Systems, Vol. 13, No. 4, July/August 1998, pp. 69-77.
    I will add more to this in another email and also add other cases in
    data archaeology for a use case.
    Mike Dean wrote:
    > I think we want to include something like this, although it
    > would probably be better as part of a larger Web Services or
    > other application use case.
    >   OWL places some limitations on expressivity to retain
    >   tractability.  A frequently cited limitation is "property
    >   chaining", the ability to express constraints among
    >   multiple properties.  We can augment an OWL ontology with
    >   additional inference rules.
    >   Several examples:
    >     2 siblings have the same father, i.e.
    >       sibling(S1, S2)
    >       father(S1, F)
    >       =>
    >       father(S2, F)
    >     a Debtor is a Person whose (cumulative) liabilities
    >     exceed his (cumulative) assets
    >         Mike
     Deborah L. McGuinness
     Knowledge Systems Laboratory
     Gates Computer Science Building, 2A Room 241
     Stanford University, Stanford, CA 94305-9020
     (voice) 650 723 9770    (stanford fax) 650 725 5850   (computer fax)
    801 705 0941

    This archive was generated by hypermail 2.1.4 : 02/04/03 EST