(08) Semantic Web Technologies - RDF(S) Semantics
-
Upload
harald-sack -
Category
Education
-
view
1.270 -
download
2
description
Transcript of (08) Semantic Web Technologies - RDF(S) Semantics
Semantic Web Technologies
LectureDr. Harald Sack
Hasso-Plattner-Institut für IT Systems EngineeringUniversity of Potsdam
Winter Semester 2012/13
Lecture Blog: http://semweb2013.blogspot.com/This file is licensed under the Creative Commons Attribution-NonCommercial 3.0 (CC BY-NC 3.0)
Dienstag, 4. Dezember 12
Semantic Web Technologies , Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
2 1. Introduction 2. Semantic Web - Basic Architecture
Languages of the Semantic Web - Part 1
3. Knowledge Representation and LogicsLanguages of the Semantic Web - Part 2
4. Applications in the ,Web of Data‘
Semantic Web Technologies Content
Dienstag, 4. Dezember 12
Lecture: Semantic Web Technologies, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam, WS 2012/13
3
Descri
ption
Logics
last lecture
Dienstag, 4. Dezember 12
Lecture: Semantic Web Technologies, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam, WS 2012/13
4 3. Knowledge Representation and LogicsThe Languages of the Semantic Web - Part 2
• Excursion: Ontologies in Philosophy and Computer Science
• Recapitulation: Popositional Logic and First Order Logic
• Description Logics
• RDF(S) Semantics• OWL and OWL-Semantics• OWL 2 and Rules
Semantic Web Technologies Content
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam Rembrandt van Rijn, Die Anatomie des Dr. Tulp, 1632
5
Why do we need formal semantics for RDF(S)?
Formal
Seman
tics
for RD
F(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
63.4 RDF(S) Semantics
3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
3. Knowledge Representation & Logic3.4 RDF(S) Semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
7
Gregor Reisch: Typus Logicae aus „Margarita Philosophica“ (1503/1508)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
8
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
9
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
10
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
11
RDF(S) Spezification from W3C in 1999 did not provide a formal definition of RDF(S) Semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
12
Why do we need semantics for RDF(S)?
■RDF(S) Spezification from W3C in 1999 did not provide a formal definition of RDF(S) Semantics
■Tool developer complain about Incompatibilities□esp. for Triple Stores, e.g. same query at different
Triple Stores (same RDF data, same SPARQL query) delivers different results□Reason:
different Interpretations of RDF data and RDF queries■Consequence:
Definition of a formal semantic is mandatory!
Semantic Web requires a shareable, declarative and computable semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
13
Why do we need semantics for RDF(S)?
■ Reasoning about RDF(S) facts would be „nice“...■ Example:■ ex:SemanticWeb rdf:type ex:Lecture.ex:Lecture rdfs:subClassOf ex:Course.
therefore we may deduce
ex:SemanticWeb rdf:type ex:Course.
■ Which statements are logical consequences is governed by the formal semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität PotsdamTurmbau zu Babel, Pieter Brueghel, 1563
14
What are the Requirements?
■Mathematical Logic is applied to formalize correct deductions and inferences.
■ Therefore, the following is required:□Set of statements about which inferences and deductions
can be made (=statements S)
□ Entailment relation ⊨ ⊆ 2S × S
□to make inferences, as e.g. {s1,s2,s3} ⊨ s
□ Logic L = (S, ⊨)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
15
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
16
Model-theoretic Semantics
■ Basic Idea: Find a relation / mapping for statements of a logic with interpretations
■ Interpretation (ΔI, I)
□ ΔI … Domain of Discourse, ΔI ≠ ∅
□ Interpretationsfunktion I I :A→AI ⊆ ΔI , A … atomic ConceptI :R→RI ⊆ ΔI x ΔI , R … atomic Relation
■ Definition of criteria to decide, if a given Interpretation I „satisfies“ a statement s∈S (Model relation)
□ I is Model of s, I ⊨ s
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
17 ■Definition of the Entailment Relation ⊨ :
□A statement s∈S is entailed from a set of
statements S⊆S (i.e. S ⊨ s) exactly if
□ all Interpretations I satisfying all statements s‘∈S(i.e. I ⊨ s‘, for all s‘∈S) are also a model for s (i.e. I ⊨ s)
Model-theoretic Semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
18
s1 s2 s⊨
Modelof s1
Modelof s2
Modelof s
⊨ ⊨ ⊨
Model-theoretic Semantics
statements
interpretations
entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
19
Model-theoretic Semantics for RDF(S)
•What are statements in RDF(S)?
•every triple (s,p,o) is a statement
•Triple can be represented with a vocabulary V•URIs, bnodes and literals
• (s,p,o) ∈ (URI ∪ bnode) × URI × (URI ∪ bnode ∪ literal)
•An (RDF-)Graph is a finite set of triples •Every (RDF-)Graph is a statement
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
20 •Entailment Relation ⊨•⊨ denotes that an RDF(S)-Graph G
entails an RDF(S)-Graph G‘ (G‘ is the logical consequence of G) ,
• i.e. G ⊨ G‘
•To define a model-theoretic semantic for RDF(S) we define a set of interpretations and denote which interpretation is a model of a given graph
Model-theoretic Semantics for RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
21 •Step-by-step definition
•Goal: formally correct mapping of the intention behind RDF(S)
RDF-Interpretation
RDFS-Interpretation
simple Interpretation
Model-theoretic Semantics for RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
22
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
23
Simple Interpretation
A simple Interpretation I of a vocabulary V consists of• IR, a non-empty set of Resources, alternatively called.
the Domain or Universe of Discurse of I,• IP, the set of Properties of I, • IEXT, a function assigning each property to a set of pairs
from I, i.e. IEXT: IP!2IR×IR, where IEXT(p) is called Extension of the Property p,
• IS, a function mapping URIs from V to the union of the sets IR and IP, i.e. IS: V!IR∪IP,
• IL, a function from the (typed) Literals of V into the set IR of resources, i.e. IL: V!IR and
• LV ⊆ IR, a particular set of Literal Values containing (at least) all untyped Literals from V
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
24 •We define a simple Interpretation Function .I mapping all Literals and URIs from V to Resources and Properties.
•every untyped Literal “a“ is mapped to a: (“a“)I=a
•every untyped Literal with language tag “a“@t is mapped to the pair〈a,t〉i.e. (“a@t“)I=〈a,t〉
•every typed Literal l is mapped to IL(l):lI= IL(l)
•every URI u is mapped to IS(u): uI=IS(u)
Simple Interpretation
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
25
Simple Interpretation (schematics)
Dienstag, 4. Dezember 12
...if the interpretation is a model for all triples of the graph
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
26
When is an Interpretation a Model of the Graph?
Semantic Web
2
http://hpi-web.de/WS0809/semanticweb/
http://hpi-web.de/LehrVeranstaltung#SWS
http://hpi-web.de/LehrVeranstaltung#Name
⊨Semantic Web
http://hpi-web.de/WS0809/semanticweb/
http://hpi-web.de/LehrVeranstaltung#Name
2
http://hpi-web.de/WS0809/semanticweb/
http://hpi-web.de/LehrVeranstaltung#SWS
⊨
⊨
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
27
When is an Interpretation a Model of a Triple?
•iff s,p,o ∈ V und 〈sI,oI〉∈ IEXT (pI)
⊨Semantic Web
http://hpi-web.de/WS0809/semanticweb/
http://hpi-web.de/LehrVeranstaltung#Name
s
p o
* this holds for grounded triples only,i.e. triples that do not contain blank nodes.
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
28The Interpretation .I assigns a truth value tothe Graph G
GI=true iff TI=truefor all triples T∈G
Simple Interpretation of a Triple (schematics)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
29
What about Blank Nodes?
•Let A be a function assigning all bnodes to elements of IR
•for the interpretation I let I+A be defined as I where in addition for all bnodes b it holds that
•bI+A = A(b)•an Interpretation I is a model of an RDF-Graph G, if there
exists an A such as all triples wrt. I+A are true
•Conclusion: A Graph G1 simply entails a Graph G2, if everysimple Interpretation that is a model of G1 is alsoa model of G2.
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
30
http://hpi-web.de/Pizza http://hpi-web.de/Mozarella
125ghttp://hpi-web.de/hasIngredient
http://hpi-web.de/ingredient
http://hpi-web.de/amount
IR = {χ,υ,τ,ν,ε,ι,125g}IP = {τ, ν, ι}LV = {125g}IEXT = τ→{〈χ,ε〉} ν→{〈ε,υ〉} ι→{〈ε,125g〉}
IS = hpi:Pizza → χ hpi:Mozarella → υ hpi:hasIngredient → τ hpi:ingredient → ν hpi:amount → ιIL = empty, no typed literalsA = _:id1 → ε
_:id1
Example
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
31
http://hpi-web.de/Pizza http://hpi-web.de/Mozarella
125ghttp://hpi-web.de/hasIngredient
http://hpi-web.de/ingredient
http://hpi-web.de/amount_:id1
Example
If you chose A: _:id1 → ε then the result is
〈hpi:PizzaI+A,_:id1I+A〉 = 〈χ,ε〉 ∈ IEXT(τ) = IEXT(hpi:hasIngredientI+A)〈_:id1I+A,hpi:MozarellaI+A〉 = 〈ε,υ〉 ∈ IEXT(ν) = IEXT(hpi:ingredientI+A)〈_:id1I+A,“125g“I+A〉 = 〈ε,125g〉 ∈ IEXT(ι) = IEXT(hpi:amountI+A)
Therefore also the full graph will become true.I is a model of the Graph (wrt. simple Interpretation)
It‘s really simple!
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
32
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
33
RDF-Interpretations
•Simple Interpretations treat all URIs the same•For the understanding of RDF Vocabulary additional
requirements regarding the set of correct interpretations are necessary
•RDF Vocabulary VRDF:
rdf:type rdf:Property rdf:XMLLiteral rdf:nil rdf:List rdf:Statement rdf:subject rdf:predicate rdf:object rdf:first rdf:rest rdf:Seq rdf:Bag rdf:Alt rdf:_1 rdf:_2 ...
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
34
•rdf:type•assigns a type to an URI •class membership of the resource denoted
by the URI•rdf:Property•denotes a specific type of resource•characterizes all URIs that in RDF triples occur as
Property•rdf:XMLLiteral•predefined datatype (XML-Fragment)•distinguish well-typed / illtyped Literals
Semantics of the RDF Vocabulary
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
35• An RDF-Interpretation for a Vocabulary V is a simple
Interpretation for the Vocabulary V∪VRDF that additionally satisfies the following conditions:
(1) x ∈ IP exactly if 〈x, rdf:PropertyI〉∈ IEXT(rdf:typeI)
• x is a Property exactly if it is connected to the resource denoted by rdf:Property via the rdf:type-Property
• (this automatically causes IP ⊆ IR for any RDF-Interpretation).
RDF-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
36(2) if "s"^^rdf:XMLLiteral is contained in V and s is a
well-typed XML-Literal, then•IL("s"^^rdf:XMLLiteral) is the XML-value of s •IL("s"^^rdf:XMLLiteral)∈LV •〈IL("s"^^rdf:XMLLiteral),rdf:XMLLiteralI〉 ∈ IEXT(rdf:typeI)
(3) if "s"^^rdf:XMLLiteral is contained in V and s is an ill-typed XML-Literal, then•IL("s"^^rdf:XMLLiteral)∉LV
•〈IL("s"^^rdf:XMLLiteral),rdf:XMLLiteralI〉 ∉ IEXT(rdf:typeI)
RDF-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
37
•Additional semantic restrictions for RDF-Interpretations: all subsequent „axiomatic“ triples must be true
•a Graph G1 RDF entails a Graph G2, if all RDF-Interpretations that are a model of G1 are also a model of G2 .
rdf:type rdf:type rdf:Property .rdf:subject rdf:type rdf:Property .rdf:predicate rdf:type rdf:Property .rdf:object rdf:type rdf:Property .rdf:first rdf:type rdf:Property .rdf:rest rdf:type rdf:Property .rdf:value rdf:type rdf:Property .rdf:_1 rdf:type rdf:Property .rdf:_2 rdf:type rdf:Property .... ... ...rdf:nil rdf:type rdf:List .
RDF-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
38
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
39
RDFS-Interpretations
•For understanging RDFS Vocabulary additional requirements regarding the set of correct interpretations for RDF-Interpretations are necessary
•RDF Vocabulary VRDFS:
rdfs:domain rdfs:range rdfs:Resource rdfs:Literal rdfs:Datatype rdfs:Class rdfs:subClassOf rdfs:subPropertyOf rdfs:member rdfs:Container rdfs:ContainerMembershipPropertyrdfs:comment rdfs:seeAlso rdfs:isDefinedBy rdfs:label
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
40•To simplify the representation:
•Class extension function ICEXT: IR # 2IR
•ICEXT(y) contains exactly those elements x for which〈x,y〉∈ IEXT(rdf:typeI)
•IC = ICEXT(rdfs:ClassI)IC is the extension of the special URI rdfs:Class
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
41•An RDFS-Interpretation for a Vocabulary V is an RDF-
Interpretation for the vocabulary V∪VRDFS that additionally satisfies the following conditions:
(1) IR = ICEXT(rdfs:ResourceI) Every resource has the type rdfs:Resource
(2) LV = ICEXT(rdfs:LiteralI) Every well-typed or untyped Literal has the type rdfs:Literal
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
42•An RDFS-Interpretation for a Vocabulary V is an RDF-
Interpretation for the vocabulary V∪VRDFS that additionally satisfies the following conditions (contd.):
(3) If 〈x,y〉∈ IEXT(rdfs:domainI) and
〈u,v〉∈ IEXT(x), then u∈ICEXT(y)
(4) If 〈x,y〉∈ IEXT(rdfs:rangeI) and
〈u,v〉∈ IEXT(x), then v∈ ICEXT(y)
If x and y are connected via property rdfs:domain/rdfs:range and the Property x connects resources u and v, then u/v is of type y
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
43•An RDFS-Interpretation for a Vocabulary V is an RDF-
Interpretation for the vocabulary V∪VRDFS that additionally satisfies the following conditions (contd.):
(5) IEXT(rdfs:subPropertyOfI) is reflexive and transitive on IP
(6) If〈x,y〉∈ IEXT(rdfs:subPropertyOfI),
then x,y ∈ IP and IEXT(x) ⊆ IEXT(y)(7) If x ∈ IC,
then 〈x,rdfs:ResourceI〉∈ IEXT(rdfs:subClassOfI)
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
44•An RDFS-Interpretation for a Vocabulary V is an RDF-
Interpretation for the vocabulary V∪VRDFS that additionally satisfies the following conditions (contd.):
(8) If 〈x,y〉∈ IEXT(rdfs:subClassOfI), then x,y ∈ IC and ICEXT(x) ⊆ ICEXT(y)
(9) IEXT(rdfs:subClassOfI) is reflexive and transitive on IC
(10) If x∈ICEXT(rdfs:ContainerMembershipPropertyI), then 〈x,rdfs:memberI ∈IEXT(rdfs:subPropertyOfI)
(11) If x ∈ ICEXT(rdfs:DatatypeI), then 〈x,rdfs:LiteralI ∈IEXT(rdfs:subClassOfI)
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
45
In addition there are numerous axiomatic triples (1)
RDFS-Interpretations
xy has domain zy
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
46
In addition there are numerous axiomatic triples (2)
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
47
In addition there are numerous axiomatic triples (3)
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
48 •a Graph G1 RDFS entails a Graph G2, if all RDFS-Interpretations that are a model of G1 are also a model of G2 .
RDFS-Interpretations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
49
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
50
•Model-theoretic Semantics describes the behaviour of a Logic wrt. correct Entailments, but is not well suited algorithmic implemen-tations
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
51
•Model-theoretic Semantics describes the behaviour of a Logic wrt. correct Entailments, but is not well suited algorithmic implemen-tations
•To proof G1 ⊨ G2 via model-theoretic semantics ALL (RDFS)-Interpretations would have to be considered
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
52 •Therefore Algorithms are developed, to decide the validity of reasoning in a syntactic way(Algorithms only apply assertions of logic without using the interpretation)
•Proof of correctness is mandatory (!), i.e.operational (proof-theoretic) semantics and model-theoretic Semantics are congruent(operational Semantics = Results of an algorithm)
•Proof theory reduces model-theoretic semantics to symbol manipulation.
Syntactic Reasoning with Deduction Rules
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
53•General Form of Deduction Rules:
•If statements s1,...,sn are contained in the set of known valid assertions (are valid), then also the statement s can be added (is also valid).
•The whole set of deduction rules for a logic is called Deduction calculus
s1 ... sns
Syntactic Reasoning with Deduction Rules
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
54
General Notation for RDF(S) Deductiuon Rules
•a and b refer to arbitrary URIs (anything admissible for the Property of a triple)
•_:n will be used for the ID of a bnode
•u snd v refer to arbitrary URIs or IDs of blank nodes(any possible Subject of a triple)
• l may be any Literal
•x and y refer to arbitrary URIs, IDs of blank nodes or literals (any possible Object of a triple)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
55
Deduction Rules for Simple Entailment
•all URIs are treated likewise
•Theorem:A Graph G1 simply entails a Graph G2, if G1 via deduction rules se1 and se2 can be transformed to a Graph G1‘ in a way that G2 is contained in G1‘.
u a x .u a _:n . se1
u a x ._:n a x . se2
Attention:_:n must not be contained in the graph the rule is applied to!
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
56•Rules se1 and se2 „weaken“ specific Subject/Object
relations via undetermined bnode
•Example.:ex:PizzaFunghi ex:isDeliveredBy ex:PizzaToGo_:id1 ex: isDeliveredBy ex:PizzaToGo
se1
ex:Mushrooms ex:isToppingOf ex:PizzaFunghiex:Mushrooms ex:isToppingOf _:id1
se2
...problematic, if _:id1 exists, e.g._:id1 ex:isHeadOfStateOf ex:Kanada
Deduction Rules for Simple Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
57
u a x . rdfax every axiomatic tripel „u a x .“ canalways be deduced
u a l .
u a _:n .lg literals can be substituted by bnodes
that do not already occur in the graph
u a y .
a rdf:type rdf:Property rdf1for all properties in a triple can be deduced that it is an entity of the class of properties
u a l .
_:n rdf:type rdf:XMLLiteralrdf2 where _:n does not yet occur in the graph
unless it has been introduced by a preceeding application of the lg rule
Deduction Rules for RDF Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
58 •Theorem:A Graph G1 RDF entails a Graph G2 exactly if there exists a Graph G1‘ that can be derived from G1 via the rules rdfax, lg, rdf1 und rdf2 and that simply entails G2.
Deduction Rules for RDF Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
59 •Property Restrictions
a rdfs:domain x . u a y .u rdf:type x .
rdfs2
rdfs3a rdfs:range x . u a y .
y rdf:type x .
Deduction Rules for RDFS Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
60 •Everything is a resource
u a x.u rdf:type rdfs:Resource . rdfs4a
rdfs4bu a x.
x rdf:type rdfs:Resource .
Deduction Rules for RDFS Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
61 •Subproperties
u rdfs:subPropertyOf v . v rdfs:subPropertyOf x .u rdfs:subPropertyOf x .
rdfs5
rdfs6u rdfs:subPropertyOf u . u rdf:type rdf:Property .
a rdfs:subPropertyOf b . u a y .u b y .
rdfs7
Deduction Rules for RDFS Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
62•Subclasses
u rdf:type rdfs:Class .u rdfs:subClassOf rdfs:Resource .
rdfs8
rdfs9u rdfs:subClassOf x . v rdf:type u .
v rdf:type x .
u rdfs:subClassOf u . u rdf:type rdfs:Class .
rdfs10
rdfs11u rdfs:subClassOf v . v rdfs:subClassOf x .
u rdfs:subClassOf x .
Deduction Rules for RDFS Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
63•Container
•Literals
rdfs12u rdf:type rdfs:ContainerMembershipProperty .
u rdfs:subPropertyOf rdfs:member .
u rdfs:subClassOf rdfs:Literal . u rdf:type rdfs:Datatype .
rdfs13
u a l . u a _:n .
gl
Deduction Rules for RDFS Entailment
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
64
RDFS Entailment and Inconsistencies
•An inconsistent Graph G entails every arbitrary Graph
• Inconsistency: there is no interpretation I with GI=true
•But with RDF(S) there are only restricted possibilities to create inconsistencies
•Example: „XML-Clash“:
ex:hasSmiley rdfs:range rdf:Literal .
rx:meanRemark ex:hasSmiley „>:->“^^XMLLiteral .
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
65
RDFS-Entailment (refined)
•Theorem:A Graph G1 RDFS entails a Graph G2 exactly if there exists a Graph G1‘ that can be derived from G1 via the rules rdfax, lg, rdf1, rdf2, rdfs1-rdfs13, gl, and rdfsax with
(1) G1‘ simply entails G2, or(2) G1‘ contains an XML-Clash (is inconsistent)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
66
Completeness of the Deduction Calculus
•simple entailment and RDF entailment is sound and complete
•RDF(S) entailment is sound and complete (at least according to the specification), but:
has the logical consequence
but this is not derivable using the deduction rules.
ex:hasTopping rdfs:subPropertyOf _:bnode ._:bnode rdfs:domain ex:Pizza .ex:PizzaFunghi ex:hasTopping ex:Mushrooms .
ex:PizzaFunghi rdf:type ex:Pizza .
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
67
Additional Rules for external Datatypes
•In RDFS external datatypes can be represented via rdfs:Datatype
•„Functionality“ of external datatypes cannot completely be represented via RDF(S) graph
•Additional deduction rules for general relations of external datatypes
d rdf:type rdfs:Datatype . u a “s“^^d ._:n rdf:type d.
rdfD1
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
68 •Range of some datatypes might overlap, e.g. “15“^^xsd:double and “15“^^xsd:Integer
•If s with datatype d represents the same value as t with datatype e, then
•If the range of datatype d in included within the range of datatype e, then
d rdf:type rdfs:Datatype .e rdf:type rdfs:Datatype .
u a “s“^^d .u a “t“^^e . rdfD2
d rdfs:subClassOf e . rdfDAx
Additional Rules for external Datatypes
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
69
3.4 RDF(S) Semantics3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
70
Intensional vs. Extensional Semantics
•Denoted semantics („standard semantics“, intentional semantics) is not the only „reasonable“ semantics for RDF(S)
•Other semantics might force stronger requirements to interpretations (extensional semantics)
•But: deduction rules of intensional semantics are easier to implement
•Problem: RDF(S) does not have the possibility of negation
•hpi:harald rdf:type hpi:NonSmoker .hpi:harald rdf:type hpi:Smoker .
--> does not automatically generate a contradiction....
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
71
Locality of global properties
Problem: Cows only eat vegetablesOther animals also eat meat.
Animal eats FoodDomain Range
Vegetables
Meat
subClassOf
subClassOf
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
72
Disjunctive Classes
Problem: Subclass relation cannot express disjunctive class (subclass) membership
Human
Woman
Man
subClassOf
subClassOf
≠
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
73
Class Combinations
Problem: Combination of classes define a new class.New class contains only members from given class combinations.
Motorist
Pedestrian
Motorcyclist
Cyclist
Road User
subClassOf
subClassOf
subClassOf
subClassOf
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
74
Cardinality Restrictions
Problem: Every human (usually) has two partents
Human ParenthasParent
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
75
Special Property Constraints
• Transitivity (e.g. „is greater than“)• Uniqueness (e.g. „is mother of“) • Inversiveness (e.g. „is parent of“ and „is child of“)
We need more se
mantic express
ivity!
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
763.4 RDF(S) Semantics
3.4.1 Why do we need semantics for RDF(S)?3.4.2 Model-theoretic semantics for RDF(S) 3.4.3 Simple Interpretation3.4.4 RDF Interpretation3.4.5 RDFS Interpretation3.4.6 RDF(S) Entailment3.4.7 Sematic Limitations of RDF(S)
3. Knowledge Representation & Logic3.4 RDF(S) Semantics
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
77 3. Knowledge Representation and LogicsThe Languages of the Semantic Web - Part 2
• Excursion: Ontologies in Philosophy and Computer Science
• Recapitulation: Popositional Logic and First Order Logic
• Description Logics
• RDF(S) Semantics• OWL and OWL-Semantics• OWL 2 and Rules
Semantic Web Technologies Content
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
OWL
Web On
tology
Langu
age78
next lecture
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
79
3. Knowledge Representation & Logic3.4 RDF(S) Semantics
• P. Hitzler, S. Roschke, Y. Sure: Semantic Web Grundlagen, Springer, 2007.
• P. Hitzler, M. Krötzsch, S. Rudolph:Foundations of Semantic Web Technologies,CRC Press, 2009.
Bibliography
Dienstag, 4. Dezember 12
Vorlesung Semantic Web, Dr. Harald Sack, Hasso-Plattner-Institut, Universität Potsdam
80
□Bloghttp://semweb2013.blogspot.com/
□Webseitehttp://www.hpi.uni-potsdam.de/studium/lehrangebot/itse/veranstaltung/semantic_web_technologien-3.html
□bibsonomy - Bookmarkshttp://www.bibsonomy.org/user/lysander07/swt1213_08
3. Knowledge Representation & Logic3.4 RDF(S) Semantics
Thank you very much to Pascal Hitzler from Kno.e.sis Center, Wright State University, Dayton, OH for his cool textbook and his slides for the lecture ,Knowledge Representation for the
Semantic Web‘ that have been the foundation for this lecture!
Dienstag, 4. Dezember 12