From: Peter F. Patel-Schneider (pfps@research.bell-labs.com)
Date: 12/06/01
Here is the newest version of my new view. I've looked it over for
problems. I think that it is fairly complete.
peter
Bringing the new RDF Model Theory to DAML+OIL
Peter F. Patel-Schneider
Bell Labs Research
(6 December 2001)
This is a new way of looking at DAML+OIL that is more compatible with the
new RDF model theory. I have tried to identify the tricky points below. I
will identify this new view as SWOL (Semantic Web Ontology Language).
The basic idea is to stay with the RDF approach of having the syntax of
semantically-meaningful constructs like subclass also show up as
relationships in interpretations. This is extended to what I think is a
reasonable point that does not exhibit any paradoxes. However, the
class-forming operators of SWOL do not get this treatment, as doing so
for all of them, in particular complementOf, would result in paradoxes and
an ill-formed model theory.
The RDF constructs like subclass are also given a more-powerful meaning.
This meaning is upward compatible with the RDF meaning but is more in
keeping with DAML+OIL.
Datatypes:
A datatyping scheme is a collection of datatypes, DT.
A datatype d in DT is a triple <Ld,Vd,LVd>
where Ld is its lexical space
Vd is its value space
LVd : Ld -> Vd is its lexical-to-value mapping
Given a datatyping scheme, let L = union over d in DT of Ld, lexical values
V = union over d in DT of Vd, data values
LV = union over d in DT of LVd
This datatyping method works best if there is a collection of primitive
datatypes, like integer, and the range-restriction of LV to the value
spaces of each of these datatypes is functional. The presence of datatypes
where this is not true, like XML Schema union datatypes, does not cause
severe problems, as long as one realizes that the SWOL type theory only
restricts the result of the lexical-to-value map, not the actual map. Thus
stating that the range of a property is integer union string does not turn
sequences of digit characters into integers. However, the presence of
datatypes with different lexical-to-value maps for the primitive datatypes,
e.g., octal integers without any syntactic tag, causes severe problems.
Lexical Issues:
N is a collection of names (URIs).
E is a collection of identifiers (blank node ids), disjoint from N.
K is a collection of literal occurences, of the form
<l,d> where l is in L, d is in DT
l:n where l is in L, n is an integer (or some other tag)
The surface syntax for the second kind of literal occurence would be just
the lexical part, but a tag is needed here for distinguishing purposes.
Syntax:
A SWOL knowledge base is a collection of statements. Some of these
statements are RDF triples, some are not. It is possible to make all
statements be in the form of triples, just not RDF triples, and I will take
this approach here.
The non-RDF parts the syntax mostly have to do with description
constructors. I recommend that an XML syntax be developed for these
constructors, something along the lines of
<intersectionOf>
<swol:Class rdf:about="foo">
<unionOf> ... </unionOf>
</intersectionOf>
However, I leave the syntax of these non-RDF constructs vague for now.
Interpretations:
A SWOL interpretation, I, over a datatyping scheme DT is a
generalized simple RDFS interpretation,
consisting of R, nonempty the domain of resources,
disjoint from V
P <= R, nonempty properties
C <= R, nonempty classes
EXT : P -> 2^(Rx(RuV)) property extensions
CEXT : C -> 2^(RuV) class extensions
I : N -> R mapping from names to denotation
I : K -> V mapping from literal occurences to
denotation
An extended SWOL interpretation, I', is an interpretation with the
following additional component
A : E -> R mapping from blank nodes to denotation
I' is said to extend I if they agree on R, P, C, EXT, CEXT, and I.
(Blank nodes are only mapped onto resources for now. It might be possible
to relax this restriction. Resources and values are disjoint.)
Both have the following conditions.
>From RDF:
CEXT(I(rdf:Property)) = P
I(rdf:type) in P
for x in R x in CEXT(y) iff <x,y> in EXT(I(rdf:type))
[rdf:type only lines up with CEXT on resources, not data values
>From RDFS:
CEXT(I(rdfs:Resource)) = R
CEXT(I(rdfs:Class)) = C
CEXT(I(rdfs:Literal)) = V
I(rdfs:Resource), I(rdf:Property) in C\DT
I(rdfs:Class), I(rdfs:Literal) in C\DT
< I(rdfs:Class), I(rdfs:Resource) > in EXT(I(rdfs:subClassOf))
< I(rdfs:Property), I(rdfs:Resource) > in EXT(I(rdfs:subClassOf))
I(rdfs:subClassOf), I(rdfs:subPropertyOf) in P
I(rdfs:domain), I(rdfs:range) in P
<x,y> in EXT(I(rdfs:subClassOf)) implies CEXT(x) <= CEXT(y)
<r,s> in EXT(IS(rdfs:subPropertyOf)) implies EXT(r) <= EXT(s)
if <x,y> in EXT(p) and <p,c> in EXT(I(rdfs:domain)) then x in CEXT(c)
if <x,y> in EXT(p) and <p,c> in EXT(I(rdfs:range)) then y in CEXT(c)
< I(rdf:type), I(rdfs:Resource) > in EXT(I(rdfs:domain))
< I(rdf:type), I(rdfs:Class) > in EXT(I(rdfs:range))
< I(rdfs:subClassOf), I(rdfs:Class) > in EXT(I(rdfs:domain))
< I(rdfs:subClassOf), I(rdfs:Class) > in EXT(I(rdfs:range))
< I(rdfs:subPropertyOf), I(rdfs:Property) > in EXT(I(rdfs:domain))
< I(rdfs:subPropertyOf), I(rdfs:Property) > in EXT(I(rdfs:range))
< I(rdfs:domain), I(rdfs:Property) > in EXT(I(rdfs:domain))
< I(rdfs:domain), I(rdfs:Class) > in EXT(I(rdfs:range))
< I(rdfs:range), I(rdfs:Property) > in EXT(I(rdfs:domain))
< I(rdfs:range), I(rdfs:Class) > in EXT(I(rdfs:range))
For datatypes:
DT <= C
I(swol:Datatype) in C\DT
CEXT(I(swol:Datatype)) = DT
for d in DT CEXT(d) = Vd
I(<l,d>) = LVd(l)
I(<l:n>) in LV(l)
For SWOL:
for x,y in C\DT CEXT(x) <= CEXT(y) implies <x,y> in EXT(I(rdfs:subClassOf))
[rdf:subClassOf only lines up with CEXT on non-datatypes]
EXT(r) <= EXT(s) implies <r,s> in EXT(IS(rdfs:subPropertyOf))
I(swol:Class) in C\DT
I(swol:ObjectProperty), I(swol:DatatypeProperty) in C\DT
I(swol:UniqueProperty), I(swol:UnambiguousProperty) in C\DT
I(swol:TransitiveProperty) in C\DT
<I(swol:Class), I(rdfs:Class)> in EXT(I(rdfs:subClassOf)
<I(swol:ObjectProperty), I(rdf:Property)> in EXT(I(rdfs:subClassOf)
<I(swol:DatatypeProperty), I(rdf:Property)> in EXT(I(rdfs:subClassOf)
<I(swol:UniqueProperty), I(rdf:Property)> in EXT(I(rdfs:subClassOf)
<I(swol:UnambiguousProperty),I(swol:ObjectProperty)> in EXT(I(rdfs:subClassOf)
<I(swol:TransitiveProperty), I(swol:ObjectProperty)> in EXT(I(rdfs:subClassOf)
x in CEXT(I(swol:Class)) iff CEXT(x) <= R
x in CEXT(I(swol:ObjectProperty)) iff x in P and EXT(x) <= R x R
x in CEXT(I(swol:DatatypeProperty)) iff x in P and EXT(x) <= R x V
x in CEXT(I(swol:UniqueProperty)) iff x in P and EXT(x) is functional
x in CEXT(I(swol:UnambiguousProperty)) iff
x in CEXT(I(swol:ObjectProperty)) and converse EXT(x) is functional
x in CEXT(I(swol:TransitiveProperty)) iff
x in CEXT(I(swol:ObjectProperty)) and EXT(x) o EXT(x) <= EXT(x)
I(swol:sameClassAs), I(swol:disjointWith) in P
I(swol:samePropertyAs) in P
I(swol:sameIndividualAs), I(swol:differentIndividualFrom) in P
<I(swol:sameClassAs), I(rdfs:subClassOf)> in EXT(I(rdfs:subPropertyOf)
<I(swol:samePropertyAs), I(rdfs:subPropertyOf)> in EXT(I(rdfs:subPropertyOf)
<I(swol:sameClassAs),I(rdfs:Class)> in EXT(I(rdfs:domain))
<I(swol:sameClassAs),I(rdfs:Class)> in EXT(I(rdfs:range))
<I(swol:disjointWith),I(rdfs:Class)> in EXT(I(rdfs:domain))
<I(swol:disjointWith),I(rdfs:Class)> in EXT(I(rdfs:range))
<I(swol:samePropertyAs),I(rdf:Property)> in EXT(I(rdfs:domain))
<I(swol:samePropertyAs),I(rdf:Property)> in EXT(I(rdfs:range))
<I(swol:sameIndividualAs),I(rdfs:Resource)> in EXT(I(rdfs:domain))
<I(swol:sameIndividualAs),I(rdfs:Resource)> in EXT(I(rdfs:range))
<I(swol:differentIndividualFrom),I(rdfs:Resource)> in EXT(I(rdfs:domain))
<I(swol:differentIndividualFrom),I(rdfs:Resource)> in EXT(I(rdfs:range))
<x,y> in EXT(I(rdfs:subClassOf)) iff x,y in C\DT and CEXT(x) <= CEXT(y)
<x,y> in EXT(I(swol:sameClassAs)) iff x,y in C\DT and CEXT(x) = CEXT(y)
<x,y> in EXT(I(swol:disjointWith)) iff x,y in C\DT and CEXT(x)^CEXT(y) = {}
<x,y> in EXT(I(rdfs:subPropertyOf)) iff x,y in P and EXT(x) <= EXT(y)
<x,y> in EXT(I(swol:samePropertyAs)) iff x,y in P and EXT(x) = EXT(y)
<x,y> in EXT(I(swol:sameIndividualAs)) iff x,y in R and x=y
<x,y> in EXT(I(swol:differentIndividualFrom)) iff x,y in R and x/=y
Satisfaction:
An extended interpretation SWOL-satisfies statements as follows. Note
that some statements produce multiple conditions, e.g., S rdfs:subClassOf O.
Statement Condition
S P O [an RDF triple] < IO(S), IO(O) > in EXT(I(P)) [as in RDF MT]
S rdf:type D IO(S) in ID(D)
D1 rdfs:subClassOf D2 ID(D1) <= ID(D2)
D1 swol:sameClassAs D2 ID(D1) = ID(D2)
D1 swol:disjointWith D2 ID(D1) disjoint from ID(D2)
P1 rdfs:subPropertyOf P2 EXT(I(P1)) <= EXT(I(P2))
P1 swol:samePropertyAs P2 EXT(I(P1)) = EXT(I(P2))
P rdfs:domain D EXT(I(P)) <= ID(Q) x R
P rdfs:range D EXT(I(P)) <= R x ID(Q)
where S P O is an RDF triple
P, P1, P2 are names
S, S1, S2 are names or blank node identifiers
D, D1, D2 are descriptions (see below)
Q, Q1, Q2 are roles (see below)
O is a name or blank node identifier or literal occurence
and IO : O -> R v V as follows
Construction Mapping
S a name I(S)
l a literal occurence I(l)
b a blank node A(b)
and ID : D -> 2^R as follows [ID does not allow non-resources in extensions]
Construction Extension
S CEXT(I(S)) ^ R
Thing R
Nothing { }
unionOf D1...Dn ID(D1) v ... v ID(Dn)
intersectionOf D1...Dn ID(D1) ^ ... ^ ID(Dn)
complementOf D R \ ID(D)
oneOf S1 ... Sn { I(S1), ..., I(Sn) }
toClass Q D { x : <x,y> in IR(Q) implies y in IC(D) }
hasValue Q O { x : <x,I(O)> in IR(Q) }
hasClass Q D { x : exists y <x,y> in IR(Q) and y in IC(D) }
minCardinality n Q { x : >=n y <x,y> in IR(Q) }
maxCardinality n Q { x : <=n y <x,y> in IR(Q) }
cardinality n Q { x : exactly n <x,y> in IR(Q) }
minCardinalityQ n Q D { x : >=n y <x,y> in IR(Q) and y in IC(D) }
maxCardinalityQ n Q C { x : <=n y <x,y> in IR(Q) and y in IC(D) }
cardinalityQ n Q C { x : exactly n <x,y> in IR(Q) and y in IC(D) }
and IC : D -> 2^(RuV) as follows
Construction Extension
S CEXT(I(S))
D (except S) ID(D)
and IR : Q -> 2^(Rx(RuV)) as follows
Construction Extension
P EXT(I(P)) if I(P) in CEXT(swol:ObjectProperty)
EXT(I(P)) if I(P) in CEXT(swol:DatatypeProperty)
[otherwise the entire statement is not satisfied]
inverseOf P converse of I(P) if I(P) in CEXT(swol:ObjectProperty)
[otherwise the entire statement is not satisfied]
Models and entailment:
An extended interpretation SWOL-satisfies a knowledge base if it
SWOL-satisfies every statement in the knowledge base.
An interpretation is a model for a SWOL knowledge base if there is some
extension of the interpretation that satisfies the knowledge base.
A SWOL knowledge base, KB1, entails another, KB2, if all models of KB1
are also models of KB2.
Theorem (to be proved):
Let KB1 and KB2 be SWOL knowledge bases. Let KB1- and KB2- be the RDF
triples in them. If KB1- RDFS entails KB2- then KB1 entails KB2.
Status of all RDF, RDFS, and ``old'' DAML-OIL constructs not handled above:
Surface syntax - does not show up at this level
xmlns:* rdf:aboutEach rdf:aboutEachPrefix rdf:li rdf:parseType
rdf:RDF rdf:Description rdf:ID rdf:about rdf:resource
Ontology versionInfo imports
Obsolete surface syntax - not needed
rdf:parseType of daml:collection
daml:List daml:nil daml:first daml:rest daml:item
Constructs with no special treatment needed (more or less)
rdfs:label rdfs:comment rdf:value rdfs:seeAlso rdfs:isDefinedBy
Unneeded description syntax
daml:Restriction daml:onProperty daml:hasClassQ
Not handled (yet)
daml:disjointUnionOf
Problematic Constructs
RDF reification - rdf:subject, rdf:predicate, rdf:object, rdf:Statement
- rdf:bagID
- what does it mean?
RDF containers - rdfs:Container, rdf:Seq, rdf:Bag, rdf:Alt, rdf:_n
- what do they mean?
daml:equivalentTo - what does it mean?
This archive was generated by hypermail 2.1.4 : 04/02/02 EST