Delivering the Power of Net-Centric Data to Ensure Mission Success UCore Semantic Layer: A Logically...

Post on 04-Jan-2016

214 views 0 download

Transcript of Delivering the Power of Net-Centric Data to Ensure Mission Success UCore Semantic Layer: A Logically...

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore Semantic Layer: A Logically Enhanced (OWL) Version of the UCore

Taxonomy

Session 2

March 17, 2010

Delivering the Power of Net-Centric Data to Ensure Mission Success

Overview

• Overview of UCore 2.0 Taxonomy

• Overview of UCore SL

– UCore SL Taxonomy

– UCore SL Relations

– Equivalence Relations

– Disjointness Axioms

– Restriction Classes

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore 2.0 Taxonomy

• 55 classes

• Simple (OWL Lite) taxonomy

• Intentionally weak semantics

• Answers the “What” and the “Who” of UCore Messages

• Does not address the “When” and “Where” directly

• Adequate for indexing and metadata purposes

Delivering the Power of Net-Centric Data to Ensure Mission Success

Entity Branch

• 27 Classes

• Relatively flat taxonomy

Delivering the Power of Net-Centric Data to Ensure Mission Success

Event Branch

• 28 Classes

• Very flat taxonomy

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore 2.0 Taxonomy

Specifies where the term fits into the

taxonomy

Specifies where the term fits into the

taxonomy

Definition derived from the Oxford English

Dictionary

Definition derived from the Oxford English

Dictionary

• No relations• No rules / constraints• No disjointness axioms

• No relations• No rules / constraints• No disjointness axioms

NOTE: No label provided – i.e., no human friendly name.

http://ucore.gov/ucore/2.0/codespace#Person

NOTE: No label provided – i.e., no human friendly name.

http://ucore.gov/ucore/2.0/codespace#Person

Delivering the Power of Net-Centric Data to Ensure Mission Success

Additional (implicit) semantics contained in the XML schema

Additional (implicit) semantics contained in the XML schema

Delivering the Power of Net-Centric Data to Ensure Mission Success

Additional (implicit) semantics contained in the XML schema

Additional (implicit) semantics contained in the XML schema

Delivering the Power of Net-Centric Data to Ensure Mission Success

Summary

• UCore Taxonomy is a lightweight ontology

• Does not utilize OWL DL logical resources

• It is up to the user to understand the meaning of a UCore 2.0 term and apply it correctly

• No automated support to validate logical consistency of COI extensions

• When semantics matter use UCore SL

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore SL

• Developed in OWL DL 1.0 (Next release OWL 2?)

– beta version

– UCore SL 1.1 (minor fixes)

• 143 classes, 16 relations

• Equivalences between UCore SL classes and UCore classes (issue of mapping)

• Use of definitions in UCore SL

• OWL DL resources

– Disjointness axioms,

– OWL restrictions

– Domain and range declarations

Delivering the Power of Net-Centric Data to Ensure Mission Success

OWL DL 1.0

OWL DL is a

W3C Recommendation

One of three species: OWL Lite, OWL DL and OWL Full

Knowledge representation language for authoring ontologies

Revision of DAML+OIL (DARPA funded project)

Ontology Editors

Protégé OWL

TopBraid Composer

Knoodl

NeOn ToolKit

OntoStudio

Swoop

Standards BasedRDFS/OWL, SPARQL, RIF,

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore SL Taxonomy

• 155 Classes

• Logical Definition

• Equivalence Statements

• OWL Restriction Classes

• Disjointness Axioms

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore Equivalence Classes

Examples of equivalence statements

Examples of equivalence statements

Not all UCore SL classes have UCore

equivalents

Not all UCore SL classes have UCore

equivalentsA equivalent to B iff every instance of A is an instance of B.

A equivalent to B iff every instance of A is an instance of B.

Delivering the Power of Net-Centric Data to Ensure Mission Success

Disjointness Axioms

A B

empty

A disjoint with B iff no instance of A is an instance of B

Delivering the Power of Net-Centric Data to Ensure Mission Success

Logically speaking, UCore 2.0 is too weak to detect simple inconsistencies.

Using UCore SL as a supporting layer makes it possible to identify that something cannot be both a Person and an Organization

Chart by Barry Smith, Director, National Center for Ontological Research, phismith@buffalo.edu.

Provides Additional Logical Resources

Delivering the Power of Net-Centric Data to Ensure Mission Success

OWL Restrictions

Property is subclass of the set of all things that inhere in some PhysicalEntity.

Property is subclass of the set of all things that inhere in some PhysicalEntity.

Delivering the Power of Net-Centric Data to Ensure Mission Success

OWL Restrictions

Person rdfs:subClassOf [

owl:onProperty :has;

owl:someValuesFrom :SocialSecurityNumber .

]

A B

A = The set of all people

B = The set of all things that have a Social Security Number

Delivering the Power of Net-Centric Data to Ensure Mission Success

OWL Restrictions

SocialSecurityNumber rdfs:subClassOf [

owl:onProperty :denotes;

owl:someValuesFrom :Person .

]

A B

A = The set of all Social Security Numbers

B = The set of all people

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore SL Relations

• 16 Relations

• Derived from UCore XML schema

• ?x agent_in ?y => ?x involved_in ?y

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore SL Relations

Most UCore SL Relations derived from the UCore XML Schema

Most UCore SL Relations derived from the UCore XML Schema

Delivering the Power of Net-Centric Data to Ensure Mission Success

UCore SL Relations

involved_in involved_in

AgentAgent EventEvent

rdfs:domain rdfs:range

Delivering the Power of Net-Centric Data to Ensure Mission Success

Note on Reasoning with Domain and Ranges

involved_ininvolved_in

XX

slr:involved_in

a owl:ObjectProperty ;

rdfs:domain sl:Agent ;

rdfs:range sl:Event .

Delivering the Power of Net-Centric Data to Ensure Mission Success

Why disjointness statements matter

involved_ininvolved_in

XX

This mistake is only caught if Agent and

SnowIceStorm are disjoint with one another

This mistake is only caught if Agent and

SnowIceStorm are disjoint with one another

Delivering the Power of Net-Centric Data to Ensure Mission Success

Properties

• Sex, Age, Height, Weight, etc.

• Diseases and Symptoms (biosurveillance)

Delivering the Power of Net-Centric Data to Ensure Mission Success

Additional (implicit) semantics contained in the XML schema

Additional (implicit) semantics contained in the XML schema

Delivering the Power of Net-Centric Data to Ensure Mission Success

Data Type Properties

Johann a :Person ;

:name “Johann”^^xsd:string ;

:age “42”^^xsd:integer ;

:height “73”^^xsd:double ;

:weight “210”^^xsd:double ;

:sex “1”^^xsd:string .

Delivering the Power of Net-Centric Data to Ensure Mission Success

More robust treatment of properties

Person Person Weight Weight

JohannJohann _:b01_:b01

rdf:typerdf:type rdf:typerdf:type

inheres_ininheres_in

“210”^^xsd:double“210”^^xsd:double

“95.2543”^^xsd:double“95.2543”^^xsd:double

“2001-10-26T21:32:52+02:00”^^xsd:dateTime“2001-10-26T21:32:52+02:00”^^xsd:dateTime

in_poundsin_pounds

in_kilosin_kilos

measured_onmeasured_on

Delivering the Power of Net-Centric Data to Ensure Mission Success

Information Content Entity

• Classify Identifiers, Codes, Names, etc.

• Classify Information

Delivering the Power of Net-Centric Data to Ensure Mission Success

Roles

Equipment Equipment CargoRole CargoRole

e-1e-1 cr-1cr-1

rdf:typerdf:type rdf:typerdf:type

inheres_ininheres_inholds_overholds_over

Time-1Time-1

Unclassified 30

End of Session 2

30

Delivering the Power of Net-Centric Data to Ensure Mission Success

Developing Ontologies with UCore SL

Session 5

March 17, 2010

Delivering the Power of Net-Centric Data to Ensure Mission Success

Overview

• How to extend UCore SL

• How to validate extensions of UCore SL

• How to represent temporal qualification of relations

Delivering the Power of Net-Centric Data to Ensure Mission Success

Two Strategies for Extending UCore SL

1. Import entire ontology

• Direct import mechanism

• Necessary to ensure consistent extension of UCore SL

2. Import selective resources from ontology

• MIREOT - Minimal information to reference external ontology terms (see http://obi-ontology.org/page/MIREOT )

• Selects only those resources from UCore SL that are relevant to a domain ontology

Delivering the Power of Net-Centric Data to Ensure Mission Success

owl:imports

owl:imports

a rdf:Property , owl:OntologyProperty ;

rdfs:label "imports" ;

rdfs:domain owl:Ontology ;

rdfs:range owl:Ontology .

owl:imports

a rdf:Property , owl:OntologyProperty ;

rdfs:label "imports" ;

rdfs:domain owl:Ontology ;

rdfs:range owl:Ontology .

The owl:imports annotation imports the contents of another OWL ontology into the current ontology.

The owl:imports annotation imports the contents of another OWL ontology into the current ontology.

Delivering the Power of Net-Centric Data to Ensure Mission Success

Example of owl:imports mechanism

<http://UCore-SL-Relations-OWL-DL><http://UCore-SL-Taxonomy-OWL-DL>

<http://ucore.gov/ucore/2.0/codespace>

<http://UCore-SL-OWL-DL>

owl:imports owl:imports

owl:imports

Delivering the Power of Net-Centric Data to Ensure Mission Success

Extending UCore SL

• How does an domain ontology extend UCore SL using the direct import mechanism?

– Import UCore SL directly into the domain ontologyPro: All UCore SL content is available for development

Con: All UCore SL content is available for development

– Import UCore SL and the domain ontology into a third ontologyPro: Separation of ontology content

Con: Difficult to reuse UCore SL terms (e.g. my:Person owl:equivalentClass sl:Person)

Delivering the Power of Net-Centric Data to Ensure Mission Success

Problems with Simple Import Strategy

• Direct OWL imports are not practical for day-to-day development.

• Avoid the overhead of importing the complete ontology from which the terms derive.

• Reuse existing ontology resources, therefore avoiding duplication of effort and ensuring orthogonality.

• Different resources may have been constructed using different design principles, which may not align. Importing such ontologies as a whole could lead to inconsistencies or unintended inferences.

<http://Common-Core-Ontology-OWL-DL>

<http://COI-Domain-Ontology-OWL-DL>

<http://UCore-SL-OWL-DL>

owl:imports

owl:imports

Delivering the Power of Net-Centric Data to Ensure Mission Success

MIREOT

• MIREOT - Minimum information to reference an external ontology terms

– The Ontology for Biomedical Investigations (OBI)

– OBI is being built under the Basic Formal Ontology (BFO)

– Membership in the OBO Foundry 

Delivering the Power of Net-Centric Data to Ensure Mission Success

MIREOT

• Define the minimal information we need

– URI of the class

– URI of the source ontology

– Position in the target ontology

• => this minimal set allows to unambiguously identify a term

• Information stored in external.owl

Delivering the Power of Net-Centric Data to Ensure Mission Success

MIREOT

• Additional information

– Label

– Definition

– Other annotations

– OWL Restrictions etc.

• External information stored in externalDerived.owl

• Use external.owl to auto-generate externalDerived.owl

Delivering the Power of Net-Centric Data to Ensure Mission Success

MIREOT Architecture

Delivering the Power of Net-Centric Data to Ensure Mission Success

Example of MIREOT Methodology

Delivering the Power of Net-Centric Data to Ensure Mission Success

external.owl

Delivering the Power of Net-Centric Data to Ensure Mission Success

Generate externalDerived.owl

Delivering the Power of Net-Centric Data to Ensure Mission Success

externalDerived.owl

Delivering the Power of Net-Centric Data to Ensure Mission Success

COI Extension Revisited

Delivering the Power of Net-Centric Data to Ensure Mission Success

Updating

Delivering the Power of Net-Centric Data to Ensure Mission Success

MIREOT

• The MIREOT standard is a trade-off between complete consistency checking and heavyweight importing versus lightweight importing but partial consistency checking. We are aware of and accept that by copying only parts of an ontology there is the risk that inferences drawn may be incomplete or incorrect: correct inference using the external classes is only guaranteed if the full ontologies are imported.

Delivering the Power of Net-Centric Data to Ensure Mission Success

How to validate extensions of UCore SL

• Validate OWL species

• Validate Consistency

• Identify cases of multiple inheritance

• Containment

Delivering the Power of Net-Centric Data to Ensure Mission Success

Validate OWL Species

Delivering the Power of Net-Centric Data to Ensure Mission Success

Validate Consistency

• Run OWL DL reasoner to test consistency

• OWL DL Reasoners: Pellet, Fact ++, RacerPro

• Pellet can also operate upon an ontology written in the OWL Full dialect but only by ignoring those features of OWL Full which make it undecidable. Thus, any inconsistencies introduced by the use of those features would not be detected.

Delivering the Power of Net-Centric Data to Ensure Mission Success

Delivering the Power of Net-Centric Data to Ensure Mission Success

OWL Reasoners

Delivering the Power of Net-Centric Data to Ensure Mission Success

Identify cases of multiple inheritance

Use SPARQL queries to identify cases of multiple

inheritance

Use SPARQL queries to identify cases of multiple

inheritance

Delivering the Power of Net-Centric Data to Ensure Mission Success

Containment

• A domain ontology fully extends UCore SL if and only if every class in the domain ontology is an (improper) subclass of some class in UCore SL.

• Simply put, containment requires that the domain covered by a lower-level ontology be circumscribed by the domain covered by the higher-level ontology from which it extends.

Delivering the Power of Net-Centric Data to Ensure Mission Success

Containment

Use SPARQL queries to identify cases of where a domain ontology term is not subsumed under a

UCore SL term

Use SPARQL queries to identify cases of where a domain ontology term is not subsumed under a

UCore SL term

Delivering the Power of Net-Centric Data to Ensure Mission Success

Temporal Qualifications

Delivering the Power of Net-Centric Data to Ensure Mission Success

The Problem

• Wheel-1 cannot be part of Car-1 and Car-2 at the same time.

• Rim-1 cannot be part of Car-1 and Car-2 at the same time.

1) Rim_1 part_of Wheel_1

2) Wheel_1 part_of Car_1

3) Wheel_1 part_of Car_2

4) Car_1 ≠ Car_2

----------------

5) Rim_1 part_of Car_1

6) Rim_1 part_of Car_2

1) Rim_1 part_of Wheel_1

2) Wheel_1 part_of Car_1

3) Wheel_1 part_of Car_2

4) Car_1 ≠ Car_2

----------------

5) Rim_1 part_of Car_1

6) Rim_1 part_of Car_2

Delivering the Power of Net-Centric Data to Ensure Mission Success

A Solution: Reification

?statement a rdf:Statement ;

rdf:subject ex:Rim_1;

rdf:predicate slr:part_of ;

rdf:object ex:Wheel_1 .

?statement a rdf:Statement ;

rdf:subject ex:Rim_1;

rdf:predicate slr:part_of ;

rdf:object ex:Wheel_1 .

ex:Rim_1 slr:part_of ex:Wheel_1ex:Rim_1 slr:part_of ex:Wheel_1

Delivering the Power of Net-Centric Data to Ensure Mission Success

Reification

ex:holdsDuring

a owl:ObjectProject ;

rdfs:domain rdf:Statement ;

rdfs:range ex:TemporalInterval .

ex:holdsAt

a owl:ObjectProject ;

rdfs:domain rdf:Statement ;

rdfs:range ex:TemporalInstant .

ex:holdsDuring

a owl:ObjectProject ;

rdfs:domain rdf:Statement ;

rdfs:range ex:TemporalInterval .

ex:holdsAt

a owl:ObjectProject ;

rdfs:domain rdf:Statement ;

rdfs:range ex:TemporalInstant .

Delivering the Power of Net-Centric Data to Ensure Mission Success

Reification

?statement

a rdf:Statement ;

rdf:subject ex:Rim_1;

rdf:predicate slr:part_of ;

rdf:object ex:Wheel_1 ;

ex:holdsDuring ex:time_1 .

?statement

a rdf:Statement ;

rdf:subject ex:Rim_1;

rdf:predicate slr:part_of ;

rdf:object ex:Wheel_1 ;

ex:holdsDuring ex:time_1 .

ex:time_1

a ex:TemporalInterval

ex:hasBeginning “2008-03-15”^^xsd:date ;ex:hasEnd “2009-03-15”^^xsd:date .

ex:time_1

a ex:TemporalInterval

ex:hasBeginning “2008-03-15”^^xsd:date ;ex:hasEnd “2009-03-15”^^xsd:date .

Delivering the Power of Net-Centric Data to Ensure Mission Success

How to reason with part_of

All part_of triplesSubset of part_of

triples

Use SPARQL Query to select

subset of part_of triples to reason

over

1) Rim_1 part_of Wheel_1

2) Wheel_1 part_of Car_1

3) Wheel_1 part_of Car_2

----------------

4) Rim_1 part_of Car_1

5) Rim_1 part_of Car_2

1) Rim_1 part_of Wheel_1

2) Wheel_1 part_of Car_1

3) Wheel_1 part_of Car_2

----------------

4) Rim_1 part_of Car_1

5) Rim_1 part_of Car_2

Delivering the Power of Net-Centric Data to Ensure Mission Success

Problems with Reification

• Using rdf:Statement makes your ontology OWL Full– Store reified triples as instance data and not as part of your

ontology

– Use an OWL DL compliant method of reificationex:PartOf_1

a ex:PartOfState ;ex:arg_1 ex:Rim-1 ;ex:arg_2 slr:part_of ;ex:arg_3 ex:Wheel-1 ;ex:holdsDuring ex:time_1 .

– Does OWL 2 make this concern irrelevant?

• Reification results in too many triples

Unclassified 64

End of Session 5

64