ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Ontology similarity and accurate communication

To: "'[ontolog-forum] '" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Patrick Cassidy" <pat@xxxxxxxxx>
Date: Wed, 19 Mar 2008 13:43:17 -0400
Message-id: <0a8d01c889e8$b76078b0$26216a10$@com>
Matthew,
   Perhaps adding more detail will help clarify:
[PC] > >   Perhaps the non-intuitive element is that an instance of a Role
(a
> > TimeSLice) can also be an instance of a Person(a
> > dimension-neutral Type) -
> 
[MW] > MW: You introduce dimension neutral type with no definition or
> explanation.
A dimension-neutral type is a type that can have instances that are 3D and
instances that are (in effect) 4D such as Roles.  In the ontology I am
using, an instance of such a type is assumed to be 3D unless it is also an
instance of an explicitly 4D type such as 'Role", in which case it is
(effectively) 4D.
    Any given instance of a 3D type will be unique.  There may be many 4D
instance of Person, related to the corresponding 3D Person by some relation
(fillsRole, isaTimeSliceOf)  Thus there is only one 3D Matthew, but many
TimeSlices such as Roles related to that unique Matthew.
   In this representation, the 3D entity serves much the same purpose as a
"whole-life" individual in a 4D ontology.  The main logical difference (not
addressing the different mental models that the two conjure) is that the 3D
entity can enter into time-indexed relations that a 4D entity cannot.  This
provides (for me) the ability to use a syntax that is closer to the English
linguistic usage, which I expect to be convenient for language-understanding
applications.  Thus we can say, in one logical notation:    (01)

Variant Notation: {a R y} translates to (R x y) in SKIF
  1. {Matthew isanInstanceOf Person}
// not explicit yet as to 3D or 4D
  2. {Matthew filledTheRoleOf EmployeeOfShell from Jan_1_1990 to
Dec_31_2007}   // here Matthew *must* be 3D, by the
                                                               //
restriction on the arguments of the 'filledtheRoleOf' relation 
  3. {Matthew filledTheRoleOf EmployeeOfLeeds from Jan_1_1980 to
Dec_31_2004}   // here Matthew *must* be 3D, by the
                                                               //
restriction on the arguments of the 'filledtheRoleOf' relation
  4. {EmployeeOfShell isaSubtypeOf HumanRole}
  5. {EmployeeOfLeeds isaSubtypeOf HumanRole}
  6. {MatthewAsEmployeeOfShellAndLeeds isanInstanceOf EmployeeOfShell}
  7. {MatthewAsEmployeeOfShellAndLeeds isanInstanceOf EmployeeOfLeeds}
  8. {MatthewAsEmployeeOfShellAndLeeds hasStartingTimePoint Jan_1_1990}
  9. {MatthewAsEmployeeOfShellAndLeeds hasStartingTimePoint Dec_31_2004}
  10. {MatthewAsEmployeeOfShellAndLeeds hasRoleFiller Matthew}
  ------------------------------     (02)

   Summary:
      there is one Matthew, who is 3D (this is clear from statements 2 and
3)
      there is a Role named MatthewAsEmployeeOfShellAndLeeds which is
related to the 3D Matthew by the relation 'hasRoleFiller' whose value must
be a 3D Object.    (03)

    The instances of Employee - which are TimeSlices that are similar to 4D
TimeSlices - are all instances of Person, and there can be many of them, for
each 3D person that fills those Roles.      (04)

    The Role 'Employee' is a subtype of Role and a subtype of Person.  But
an instance of Employee cannot be an instance of a 3D Person.  It is an
instance of a TimeSlice of a Person.    (05)

    One potentially confusing point is that I have not defined an explicit
'3D' object type.  An object is recognized as being 3D when it is
constrained by being used in a relation that is meaningful only for 3D types
and not for 4D types such as Roles.  I could add an explicit 'Object3D' (I
experimented with one once) but did not see a real need for it.  It could be
added in if that would make the representation more understandable.  In that
case the additional assertion would be added to the above:
    0. {Matthew isanInstanceOf Object3D}    (06)

    I hope that is clearer.  
    If you see any logical inconsistencies, please show them in some logical
notation.    (07)

Pat    (08)

Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx    (09)


> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of matthew.west@xxxxxxxxx
> Sent: Wednesday, March 19, 2008 8:30 AM
> To: ontolog-forum@xxxxxxxxxxxxxxxx
> Subject: Re: [ontolog-forum] Ontology similarity and accurate
> communication
> 
> Dear Pat,
> 
> >    If one were to attempt to create instances of a 3D Person
> > that were also
> > instances of a role, the multiplicity you refer to would be a
> > problem.  In
> > the illustration below, Matthew is an instance of 3D Person,
> > not an instance
> > of Employee - if Matthew were an instance of Employee, the problem
> you
> > describe would then present itself.  But we can create
> > instances of Role
> > such as an Employee that have a Person as a filler.
> 
> MW: So far so good. A person is a filler for a role.
> 
> > Nevertheless the *Role*
> > "Employee" is also a subtype of Person,
> 
> MW: Now there you go and spoil it. If employee is a subtype of person,
> it means that each instance of employee is a separate instance of
> person, i.e. there are two of me. There is a relationship between
> employee and person of course, but you need to let go of the popular
> myth that it is subtype/supertype.
> 
> > therefore an instance
> > of an Employee
> > is always an instance of a Person, implicitly though not
> > explicitly.
> 
> MW: This is not an option. You are or you are not a person, and it is
> very explicit.
> 
> > But
> > the instances of a Role such as Employee are more like time
> > slices and are
> > not identical to the instance of Person that fills the role.
> 
> MW: This is exactly right, i.e. an employee is not a person, but a
> person for some period of time.
> 
> > The
> > multiplicity is in the multiplicity of TimeSlices, as there
> > are in a 4D
> > representation; every different TimeSlice is a different
> > entity.
> 
> MW: That would be a stage theory approach to 4-dimensionalism. I
> prefer what Catherine Hawley calls perdurantism, where states are
> extended in time as well as space. So there is one object that is
> the person, that is an extent from birth (say) to death, and other
> ones for the employee, which are from start to end of the
> employment.
> 
> > The way
> > that can work is illustrated below using OWL.
> >    One comment I made was not phrased properly:
> > [PC] > >   (3) a person can be an instance of multiple roles
> > in any given
> > time
> > > > interval
> >    It would more accurately have been phrased that:
> >    (3a) an instance of Person can fill multiple roles in any
> > given time
> > frame
> 
> MW: That is OK.
> 
> >    (3b) a timeslice of Person can be an instance of multiple roles
> 
> MW: So is that.
> 
> >    (3c) an instance of Role in any given time interval can be
> > an instance of
> > Person - because Person and Role are not disjoint.
> 
> MW: This is not OK. The (3D) mistake you are making here is that
> a person-at-a-point-in-time is not a person, but a person-at-a-point-
> in-time. The person is able to pass through time and is present at
> a large number of points in time. The person-at-a-point-in-time only
> refers to that point in time, not the whole person. This is all a lot
> more straightforward in 4D of course.
> 
> > The illustration below should clarify what that means.
> >
> >   Perhaps the non-intuitive element is that an instance of a Role (a
> > TimeSLice) can also be an instance of a Person(a
> > dimension-neutral Type) -
> 
> MW: You introduce dimension neutral type with no definition or
> explanation.
> 
> > because the type Person is not disjoint with the type
> > TimeSlice.  In the
> > illustration below, Matthew is a Person, but the instance
> > MatthewAsEmployeeOfShellAndLeeds is a Role - and also an
> > instance of Person
> > - but is not identical to Matthew.
> 
> MW: Now I have a counting problem. How many persons are there?
> 
> > It is more like a
> > TimeSlice of Matthew.
> 
> MW: It is precisely a timeslice of Matthew, i.e. Matthew for a period
> of time.
> 
> > Even So, Matthew is not necessarily 3D or 4D, Matthew is a
> > dimension-neutral
> > entity.
> 
> MW: Well I'm afraid this is a problem. For a 4-dimensionalist the
> only real world objects are spatio-temporal-extents. If it isn't one
> of those, it just doesn't exist. Of course a 3 Dimensionalist would
> have to say that I'm not extended in time, but pass through it. Sigh.
> 
> >   If one were to use a formalism that permitted relation
> > arities higher than
> > two, it would be possible to specify roles and their time
> > limits without
> > using time slices.  TimeSlices (among which are Roles) are
> > syntactically
> > convenient when using OWL, they aren't logically necessary.
> 
> MW: I'm glad you have noticed how practical what I prefer to call
> states are.
> 
> > They have the
> > same logical effect as an explicit time-indexed assertion on a
> > dimension-neutral entity.
> 
> MW: Not really. You still have to decide whether it is or is not
> a temporal part of the person.
> 
> > In that case, there would be no
> > TimeSlices, only
> > time-indexed assertions (which are logically equivalent,
> > after translation,
> > to assertions on TimeSlices).  The axioms to translate the
> > two formalisms
> > are not here because this is a pure OWL representation.
> >
> >   All instance of Role should have start and end times
> > specified - they are
> > time slices.
> >   Person is in this case neither exclusively 3D nor 4D - time
> > slices can be
> > generated by making a person an instance of 'TimeSlice', or
> > an instance of
> > Role, which is a subtype of TimeSlice.
> 
> MW: Only 4D objects can have states (timeslices). I agree role
> is a subtype of state (timeslice). So is person.
> 
> > Every TimeSlice has a
> > start time and
> > end time.
> 
> MW: So does person, which is why person is a state (timeslice).
> 
> >   But an instance of Person can also have attributes and
> > relations specified
> > by explicit time-indexed relations, without using TimeSlices.
> 
> MW: Not if you are a 4-dimensionalist. only spatio-temporal extents
> can have times, relations must be timeless. You can't point to the
> spatial extent of a relation.
> >
> >   In the following note that an Employee is a Role, and
> > something can be an
> > instance of an Object as well as a Role.
> 
> MW: But it can't... except accidentally.
> 
> >   (definitions - some parts are from Cyc - are abbreviated).
> >
> >     HumanRole is a subtype of Person and Role
> 
> MW: No. Both person and role are subtype of stateOfPerson. A role
> is a temporal part of a person, or if you prefer a role consists of
> a person.
> 
> MW: You will have a counting problem here, because each instance
> of HumanRole will be counted when you count how many persons there
> are.
> 
> >     Every HumanRole is a Role whose RoleFiller is a Person
> > (restriction)
> 
> MW: That is OK. This is equivalent to what I said above about a
> role consisting of a person.
> 
> >     PersonWithOccupation is a subtype of HumanRole
> 
> MW: That is OK.
> 
> MW: I don't read OWL in XML form, so I'll take your word that what
> follows matches what is above.
> 
> Regards
> 
> Matthew West
> Reference Data Architecture and Standards Manager
> Shell International Petroleum Company Limited
> Registered in England and Wales
> Registered number: 621148
> Registered office: Shell Centre, London SE1 7NA, United Kingdom
> 
> Tel: +44 20 7934 4490 Mobile: +44 7796 336538
> Email: matthew.west@xxxxxxxxx
> http://www.shell.com
> http://www.matthew-west.org.uk/
> 
> >
> >     <owl:Class rdf:ID="Employee">
> >         <rdfs:comment>A Person who was employed by another
> > Agent in some
> > hiring event.</rdfs:comment>
> >         <rdf:type rdf:resource="#PersonType"/>
> >         <rdf:type rdf:resource="#RoleType"/>
> >         <rdfs:subClassOf
> rdf:resource="#TemporaryRoleCreatedByEvent"/>
> >         <rdfs:subClassOf rdf:resource="#PersonWithOccupation"/>
> >        <rdfs:subClassOf>
> >          <owl:Restriction>
> >            <owl:onProperty>
> >              <owl:ObjectProperty rdf:about="#isEmployedBy"/>
> >            </owl:onProperty>
> >            <owl:someValuesFrom rdf:resource="#IntelligentAgent"/>
> >          </owl:Restriction>
> >        </rdfs:subClassOf>
> >     </owl:Class>
> >
> > **** NOTE that 'employee' is a subtype of 'Person'  ****
> >
> >   <owl:Class rdf:ID="EmployeeOfShell">
> >     <rdfs:subClassOf rdf:resource="#Employee "/>
> >     <rdfs:comment>A person who is an employee of
> Shell.</rdfs:comment>
> >   </owl:Class>
> >
> >   <owl:Class rdf:ID="EmployeeOfLeeds">
> >     <rdfs:subClassOf rdf:resource="#Employee "/>
> >     <rdfs:comment>A person who is an employee of
> Leeds.</rdfs:comment>
> >   </owl:Class>
> >
> > <!-- Matthew is employee of Shell from 1990 to 2008
> >      Matthew is employee of Leeds from 1980 to 2004
> > -->
> >
> >   <Person rdf:ID="Matthew"/>
> >
> >   <EmployeeOfShell rdf:ID="MatthewAsEmployeeOfLeedsAndShell">
> >         <rdf:type rdf:resource="#EmployeeOfLeeds"/>
> >         <hasRoleFiller rdf:resource="#Matthew"/>
> >         <rdfs:comment>Matthew is an Employee of Both Shell
> > and Leeds from
> > the beginning of
> >            1990 to the end of 2004.</rdfs:comment>
> >         <hasStartingTimePoint rdf:ID="DTEG19900101T0000"/>
> >         <hasEndingTimePoint rdf:ID="DTEG20041212T2400"/>
> >   </EmployeeOfShell >
> >
> > Pat
> >
> > Patrick Cassidy
> > MICRA, Inc.
> > 908-561-3416
> > cell: 908-565-4053
> > cassidy@xxxxxxxxx
> >
> >
> > > -----Original Message-----
> > > From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> > > bounces@xxxxxxxxxxxxxxxx] On Behalf Of matthew.west@xxxxxxxxx
> > > Sent: Tuesday, March 18, 2008 2:42 PM
> > > To: ontolog-forum@xxxxxxxxxxxxxxxx
> > > Subject: Re: [ontolog-forum] Ontology similarity and accurate
> > > communication
> > >
> > > Dear Pat,
> > >
> > > >    This is a good issue, but I think it has a resolution
> > > > without the problem
> > > > you envision, because in 3D the types of an entity can change
> > > > with time.
> > > >
> > > >   (1) employee is a role, which means that it must be
> > time indexed.
> > > >   (2) every instance of employee (in some time interval) is
> > > > an instance of
> > > > person  (in that time interval)
> > > >   (3) a person can be an instance of multiple roles in
> > any given time
> > > > interval
> > > >   (4) in some time interval Matthew can be an instance of
> > "Employee
> > > of
> > > > Shell" and "Employee of Leeds U"
> > >
> > > MW: What you are missing is that EACH instance of employee MUST be
> a
> > > separate person if employee is a subtype of person. To put that the
> > > other way round, if I am just one person and there is an employee
> > > subtype of person, then I either am or am not an instance
> > of employee,
> > > but I can only be an instance of employee once.
> > >
> > > MW: If you think that a Person can be more than one
> > employee, then the
> > > relationship between person and employee is something other than
> > > subtype/supertype.
> > >
> > > MW: If I were a 3D-ist then I would suggest somthing like a
> > consists of
> > > relation.
> > >
> > > >
> > > >   In 4D, I believe that the 4D worms will intersect, and that
> > > > is another way
> > > > of viewing the same thing, but it is only inconsistent if one
> > > > assigns the
> > > > same type "Person" to a 4d object in one ontology and a 3D
> > > > object in the
> > > > other, and then tries to use the same term to represent the
> > > > two different
> > > > types.
> > >
> > > MW: In 4D it is quite clear, and employee is a state of a person,
> > > and the relationship between person and employee is
> > temporal part of.
> > > Both employee and person are subtypes of state_of_person.
> > >
> > > MW: The interesting thing about temporal part of is that most
> > > properties are inherited by substates (except for example being
> > > a person for the whole of their life).
> > >
> > >
> > > Regards
> > >
> > > Matthew West
> > > Reference Data Architecture and Standards Manager
> > > Shell International Petroleum Company Limited
> > > Registered in England and Wales
> > > Registered number: 621148
> > > Registered office: Shell Centre, London SE1 7NA, United Kingdom
> > >
> > > Tel: +44 20 7934 4490 Mobile: +44 7796 336538
> > > Email: matthew.west@xxxxxxxxx
> > > http://www.shell.com
> > > http://www.matthew-west.org.uk/
> > >
> > > >
> > > >    I need to be more specific.  Another day or two for my
> > > > more detailed
> > > > reply to PatH.
> > > >
> > > > Pat
> > > >
> > > > Patrick Cassidy
> > > > MICRA, Inc.
> > > > 908-561-3416
> > > > cell: 908-565-4053
> > > > cassidy@xxxxxxxxx
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
> > [mailto:ontolog-forum-
> > > > > bounces@xxxxxxxxxxxxxxxx] On Behalf Of matthew.west@xxxxxxxxx
> > > > > Sent: Tuesday, March 18, 2008 5:18 AM
> > > > > To: ontolog-forum@xxxxxxxxxxxxxxxx
> > > > > Subject: Re: [ontolog-forum] Ontology similarity and accurate
> > > > > communication
> > > > >
> > > > > Dear John,
> > > > >
> > > > > > MW> So for example, there are ontologies where you will
> > > > find employee
> > > > > >  > as a subtype of person, and others that understand
> > it is not.
> > > > > >
> > > > > > I don't know which way you are advocating, but I'll
> > summarize my
> > > > > > position:
> > > > > >
> > > > > >   1. There is a fundamental distinction between natural
> > > > types, such
> > > > > >      as Cat or HumanBeing, and role types, such as Pet or
> > > > Employee.
> > > > > >
> > > > > >   2. Every instance of a role type is a subytpe of some
> > > > natural type,
> > > > > >      but it may also be a subtype of other role types.
> > > > > > HeartSpecialist
> > > > > >      is a subtype of Physician, which is a subtype of
> > HumanBeing.
> > > > > >
> > > > > MW: I do mean that employee is not a subtype of person (or
> human
> > > > > being).
> > > > >
> > > > > MW: If we remind ourselves of what being a subtype means,
> > > > it means that
> > > > > each instance of a subtype is an instance of the supertype.
> > > > Now ler us
> > > > > look at an example. I am an employee of both Shell and Leeds
> > > > > University.
> > > > > I have different employee numbers, very different salaries, and
> > > > > different start dates. Now if employee is a subtype of
> > person then
> > > > > each of these is a person, i.e. there are two of me.
> > > > >
> > > > > MW: This kind of situation is true of roles generally,
> > you can play
> > > > > multiple roles at the same time and the same role multiple
> times
> > > > > (and at the same time). These do not all generate new people.
> > > > >
> > > > > MW: So the question is: what is the relationship between a role
> > > > > and the person who plays is. Fortunately, as a
> > > > 4-dimensionalist, there
> > > > > is a simple answer. The role is a temporal part of the
> > person that
> > > > > plays the role, or if you prefer,  the person for a
> > period of time,
> > > > > rather than for the whole of their life.
> > > > >
> > > > > MW: This is not so different from the question of the
> > vase and the
> > > > > piece of clay. Are pots subtypes of clay? Or is the pot
> > a different
> > > > > object than the piece of clay it is made from?
> > > > >
> > > > > Regards
> > > > >
> > > > > Matthew West
> > > > > Reference Data Architecture and Standards Manager
> > > > > Shell International Petroleum Company Limited
> > > > > Registered in England and Wales
> > > > > Registered number: 621148
> > > > > Registered office: Shell Centre, London SE1 7NA, United Kingdom
> > > > >
> > > > > Tel: +44 20 7934 4490 Mobile: +44 7796 336538
> > > > > Email: matthew.west@xxxxxxxxx
> > > > > http://www.shell.com
> > > > > http://www.matthew-west.org.uk/
> > > > >
> > > > >
> > > > >
> > _________________________________________________________________
> > > > > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > > > > Subscribe/Config:
> > http://ontolog.cim3.net/mailman/listinfo/ontolog-
> > > > > forum/
> > > > > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > > > > Shared Files: http://ontolog.cim3.net/file/
> > > > > Community Wiki: http://ontolog.cim3.net/wiki/
> > > > > To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> > > > >
> > > >
> > > >
> > > > _________________________________________________________________
> > > > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > > > Subscribe/Config:
> > > > http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> > > > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > > > Shared Files: http://ontolog.cim3.net/file/
> > > > Community Wiki: http://ontolog.cim3.net/wiki/
> > > > To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> > > >
> > > >
> > > >
> > >
> > >
> > > _________________________________________________________________
> > > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > > Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
> > > forum/
> > > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > > Shared Files: http://ontolog.cim3.net/file/
> > > Community Wiki: http://ontolog.cim3.net/wiki/
> > > To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> > >
> >
> >
> > _________________________________________________________________
> > Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> > Subscribe/Config:
> > http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> > Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> > Shared Files: http://ontolog.cim3.net/file/
> > Community Wiki: http://ontolog.cim3.net/wiki/
> > To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> >
> >
> >
> 
> 
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-
> forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>     (010)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (011)

<Prev in Thread] Current Thread [Next in Thread>