2_webSemantique - Centre d'information topographique ...

Earth Sciences Sector

Semantic Web

…vers l’interopérabilité sur le Web

Jean Brodeur

Journée INNOVATION en Géomatique
-

6e Édition

Centre d’information topographique
-

Sherbrooke

8 novembre 2007

2

Déroulement de la présentation

•
Contexte

•
Description

•
Ontologie

•
Technologies du W3C

•
Conclusion


Earth Sciences Sector

Web Sémantique

Contexte

4

Interoperability of information

•
Concerns the
understanding and usage of information

•
Increases the
availability
,
access
,
integration
, and
sharing

of information

•
Concerns the establishment of
data infrastructures

at
local, regional and global level

5

…between people

•
Is based on

–
the communication process;

–
People knowledge and the commonness.

6

… through the communication paradigm

<Factory>


<name>FactoryA</name>


…


User’s request with his own
concepts in memory

(e.g. Factory, Mill,

Plant, etc.)

(Communication channel)

“Factories
within

Kyoto?”

<Factory>


<name>FactoryA</name>


…


5. Data encoding

(message production)

6. Data
transmission

7. Data
reception

8. Data decoding
(message recognition)

R

R’’’’

R’’

R’’’

R’

2. Request
transmission

1. Request encoding
(message production)

Cr =
f
(C)

Provider

User

Administrative area
(Kyoto)

Building (factory)

(Communication channel)

<Factory>


<name>FactoryA</name>


<location>


<GPL_CoordinateTuple>



<tuple


CrsName="urn:EPSG::21418">


1259753 18503245


…

Interoperability =

correspondence of received data with
the initial request.

= T

|
S
|

-
FactoryA

-
EPSG:21418

-
1259753, 18503245

-
Factory

-
Kyoto

-
Factory

-
Kyoto

4. Request decoding
(message recognition)

3. Request
reception

-
Building (factory)

-
Factory

-
Administrative

-
Kyoto


area (Kyoto)

|
S
|

= T

Request recognition from database’s
geographic concepts




then search of corresponding
geographic information.

Recognition =
f

(
{
C
1
, ... ,C
n
}
, Cr)

7

Heterogeneity of information

•
A major barrier to interoperability

•
Types of heterogeneity

–
System

(i.e. interaction between computers of
different OS and databases of
different
DBMS
)

–
Syntactic

(i.e.
differences between formats

such as a GML document and a Shapefile)

–
Schematic

(i.e.
differences in conceptual
schemas

such as
street

may be defined as a
class or as a value of an attribute of a
road

class)

–
Semantic

(i.e.
difference of meaning

given to
a signal, e.g.
chair

means either a seat or a
position of authority, or the various signal
that have a similar meaning, e.g.
watercourse

vs.
river/stream
)

8

Current Web

•
Information is mainly based on
Web documents

•
A Google search lists Web documents that correspond to
keywords
–

e.g. “
Semantic Web
”





•
Web documents are intended to
human beings
, which
have to figure out the nature and usefulness of their
contents

•
It is not designed for the use of information by software

Earth Sciences Sector

Web Sémantique

Une description

10

Semantic Web

•
An idea introduced by

T. Burners
-
Lee

•
From a Web of
documents

for
humans to a Web of
data and
information

processable by
computers

•
Published the first time in
2001

–
T. Berners
-
Lee, J. Hendler, and O.
Lassila, “The Semantic Web,” Scientific
Am., May 2001, pp. 34
–
43.

11

Semantic Web

•
Is about
a Web that answers questions

instead of
returning Web pages about topics of interests

•
Is about
data

that is
application independent
,
composeable
,
classified
, and
part of a larger information
structure

•
Is about
data

that is
understandable and processable by
machines

–
Needs to make the data smarter

Text and DB records

XML with

mixed vocabularies

XML and single

domain vocabularies

Ontologies and rules

12

Data, information, and knowledge pyramid

from semanticweb.org

13

Semantic Web

•
Is seen as a
solution

to

–
information overload

specially with the propagation of
the Internet

–
breaking stovepipe systems and allowing
sharing
information

–
aggregating information

from multiple sources

–
enabling users to
retrieve the data they need

more
efficiently
based on their own vocabulary (concepts) and
data specific vocabulary (concepts)

14

Semantic Web deals with…

•
Common formats

–
XML is the syntactic foundation
(RDF, RDF
-
S, OWL,
RIF
,
SPARQL)

–
Oriented toward integration and
combination of data from various
sources (Web)

–
As opposed to the original Web
that is oriented toward the
interchange of documents

•
Language

–
Capturing how the data relates to
real world objects (RDF
-
S and
OWL).

Berners
-
Lee, T., 2006.
Artificial Intelligence and the Semantic Web
,

AAAI Conference keynote,
2006
-
07
-
18.

http://www.w3.org/2006/Talks/0718
-
aaai
-
tbl/Overview.html

15

Semantic Web… What is needed?

•
Logical assertions

–
connect subject to an object with a verb

–
RDF

•
Classification of concepts

–
Taxonomies/ontologies

•
Formal models

–
Concepts, their properties, and relationships

–
OWL

–
For reasoning

•
Rules

–
Inference rules to derive conclusion

–
RIF

•
Trust

–
Provide access to resources only to trusted agents. An agent can be asserted
“trusted” from another via a digital signature

16

Web services and Semantic Web

•
Based on URI

•
XML

•
Smart data

•
Semantic Web to discover Web services (Semantic Web
-
enabled Web services)

•
Semantic Web to support interaction between Web
services

17

Geospatial Semantic Web

•
Developed by

–
Max J. Egenhofer
, 2002.
Toward the Semantic Geospatial Web
, Proceedings of the
10th ACM international symposium on Advances in geographic information systems,
p.1
-
4, November 08
-
09, 2002, McLean, Virginia, USA


–
Frederico Fonseca and Amit Sheth
, 2002.
The Geospatial Semantic Web
, UCGIS
White Paper, 2002. http://www.ucgis4.org/priorities/research/

2002researchagenda.htm

•
Challenges

–
Ontologies of spatial concepts use across disciplines


geospatial
-
relations ontology


Geospatial feature ontology

–
Ontology management: designing, developing, storing, registering,
discovering, browsing, maintaining and querying

–
Canonical form for geospatial data queries

–
Matching concepts to ontologies

–
Ontology integration

Earth Sciences Sector

Web Sémantique

Ontologie

19

Ontology

•
What is an ontology?

–
Taxonomy?


XML schema?

–
Thesaurus?


Conceptual model?

–
UML, RDF/S, OWL?


Description logic?

–
Logical theory?

•
What is the purpose or role of an ontology?


20

Ontology

•
A
foundation

for the success of the
Semantic Web

•
Meaning of data

in a format that machine can
understand

•
Data derived its
semantics

from ontology

•
To support
integration of heterogeneous data

across
communities

21

Semantics

Concept

•
Thoughts that give
meaning

to signs and
phenomena;

referent

signifier

signified

•
Links between signs and
real world phenomena.

(Frege, Peirce, Ogden & Richards, Eco)

Phenomenon


Colosseum,

Rome,

N41
°
53'25" Latitude

E12
°
29'32" Longitude

Sign

22

Ontology

•
Philosophy

•
Artificial intelligence

23

Ontology… A philosophical account

•
Study or science of being (or existence)

•
Description of the world in itself

•
Type of entities, properties, categories, and relationships
that compose the reality

•
Philosophy consider that there is only one ontology

24

Ontology… An artificial intelligence account

•
“An explicit specification of a conceptualisation”
(Gruber 1993)

•
“A logical theory accounting for the intended meaning
of a vocabulary” (Guarino 1995)

•
A layer enabling the definition of concepts of reality

•
Meaning of a subject area or an area of knowledge

•
A
formal representation of phenomena

with an
underlying
vocabulary

including
definitions

and
axioms

that make the intended meaning explicit and
describe
phenomena and their interrelationships
(Brodeur 2003)

25

Ontology… An artificial intelligence account


•
Represented by classes, relations, properties, attributes,
and values

•
AI considers that
reality may be abstracted differently
depending on the context

from which “things” are
perceived

•
AI recognizes that
multiple ontologies

about the same
part of reality
may exist



26

Ontology… an example

•
Common conceptualization

•
Living structure

–
Static

–
Volatile

•
Explicit commitment to shared meaning among an
interested community

•
Can be re
-
used and extended

27

Ontology

Spectrum

Daconta, M. et al., 2003.
The semantic web
, Wiley.

28

Multiple ontology levels

•
Global or top
-
level ontology:
general concepts independent
of a specific domain (e.g. space,
time, …)

•
Domain ontology: concepts
specific to a domain (e.g.
transportation, geology, land
cover, …)

•
Application ontology: concepts
that are specialised in a given
context and use (e.g. parcel
delivery, ambulance
dispatching, rescue, …)

29

Role of ontology

•
Knowledge base that supports interpretation, reasoning,
and inference

–
Description logic:
river/stream

watercourse

–
Notion of similarity/proximity: the concept
watercourse

contains

the concept
river/stream

–
Joe is passenger of Train 1234; Train 1234 goes to
Rome; Joe goes to Rome

–
…

30

Reasoning and inference

•
Possible through the relation that exist between concepts

–
Subsumption:
isA
,
isSuperclassOf

–
Meronymy:
part of

–
GeoSemantic Proximity: Based on a 4 intersection matrix between intrinsic
and extrinsic properties of two concepts.


intrinsic properties provide the literal meaning of the concept


extrinsic properties provide meaning through the influence that other concepts
have on a concept (e.g. behaviours and relationships)

–
Matching distance: a distance between concepts in a graph

–
…

31

Subsumption relations

32

GeoSemantic Proximity


Intrinsic Properties

(C
K
°
)

Extrinsic
properties

(

C
K
)

C
K

33

Geosemantic Proximity

Common
extrinsic
properties

Common
intrinsic
properties

No common
intrinsic
properties

No common
extrinsic
properties

The geosemantic proximity of

Road

with
Street

is then
GsP_fftt

ou
contains

Road

vs.
Street
:

•
Street

participates in a relationship with other types of
Road

•
Then, the intersection of extrinsic properties of
Street

with intrinsic properties of
Road

is not empty

Road

vs.
Street
:

•
Street

corresponds to a value of the attribute
classification

of
Road

•
Both have the same geometry

•
Then, the intersection of intrinsic properties of
Road

and
Street

is not empty

34

Context

•
Provides concepts with
real
-
world semantics

•
About
how phenomena are perceived and abstracted

resulting in various classes, properties (thematic, spatial,
temporal), and relationships

•
About
how data is captured

in databases including
constraints such as on object dimension

•
Provide details on:

–
Use:

user ID, user profile, user location, type of uses

–
Data:

source, geospatial entities, meaning, scale, date of validity, etc.

–
Association:

relationships (spatial, semantic, etc.)

–
Procedure:

process steps to capture the data, query to get the data, etc.

•
Metadata

constitutes a valuable source of contextual
details

•
Can be captured by the way of intrinsic and extrinsic
properties

35

Interoperability, Semantics, and Ontologies

<Factory>


<name>FactoryA</name>


…


(Communication channel)

“Factories
within

Kyoto?”

<Factory>


<name>FactoryA</name>


…


R

R’’’’

R’’

R’’’

R’

Provider

User

(Communication channel)

<Factory>


<name>FactoryA</name>


<location>


<GPL_CoordinateTuple>



<tuple


CrsName="urn:EPSG::21418">


1259753 18503245


…

-
Factory

-
Kyoto

-
Factory

-
Kyoto

Ontologies

Earth Sciences Sector

Web Sémantique

Technologies du W3C

37

W3C Technologies

•
Resource Description Framework (RDF)

–
http://www.w3.org/RDF/

•
Resource Description Framework Schema (RDF
-
S)

–
http://www.w3.org/TR/rdf
-
schema/

•
Web Ontology Language (OWL)

–
http://www.w3.org/2004/OWL/


38

RDF

•
Is based on the triple:
Subject

-

Predicate

–

Object

•
Subject:

the resource, the
thing

about which
something is asserted

•
Predicate:

the
relation

that
binds the
subject

to the
object

•
Object:

either a
literal
value

or a

resource

referred to the
subject

by
the
predicate

Subject

Object

Literal Value

Predicate

Predicate

Example:

<rdf:Description rdf:about="
#colosseum
">


<ex:
isLocatedIn
>


<rdf:Description rdf:about="
#Rome
">


</rdf:Description>


</ex:
isLocatedIn
>

</rdf:Description>

39

RDF
-
S

•
Based on
RDF

•
Set of
standard RDF resources

to create application/user
community
specific RDF vocabularies

•
Allows to create
classes

of data

•
Class instances are then created in RDF

•
Relations are introduces as property

40

RDF
-
S, an example

CI_Address

CitationAndResponsibleParty

+ addressAdministrativeArea

+ addressCity

41

OWL

•
Language

for knowledge representation

•
Initiated in November 2001

•
Is an evolution of DAML+OIL

–
DAML: DARPA Agent Markup Language

–
DARPA: Defence Advanced Research Projects Agency

–
OIL: Ontology Inference Layer

•
Three levels from low to high expressivity

–
Lite:

intended mainly for the description of classification hierarchy with
attributes, cardinalities are limited to 0 or 1

–
DL:

stands for description logics, add knowledge representation that
improves reasoning, allows much flexibility on cardinality restrictions

–
Full:

allows

maximum expressiveness and the syntactic freedom of RDF.
As such a class may be either a collection of individuals or an individual in
itself

42

OWL , an example

CI_Address

CitationAndResponsibleParty

+ addressAdministrativeArea

+ addressCity

43

Tools

•
Jena 2 Toolkit:

–
RDF/OWL API

–
http://jena.sourceforge.net/

•
Protégé 2000

–
Editor for ontology

–
http://protege.stanford.edu/

•
Tools at Network Inference

–
http://www.networkinference.com/

•
OilEd:

–
http://oiled.man.ac.uk/

–
Editor for ontologies

–
Mostly for DAML+OIL, exports OWL but not a current representation

•
OWL Validator:

–
http://owl.bbn.com/validator/

–
Web
-
based or command
-
line utility

–
Performs basic validation of OWL file

•
OWL Ontology Validator:

–
http://phoebus.cs.man.ac.uk:9999/OWL/Validator

–
a "species validator" that checks use of OWL Lite, OWL DL, and OWL Full constructs

•
Euler:

–
http://www.agfa.com/w3c/euler/

–
an inference engine which has been used for a lot of the OWL Test Cases

•
Chimaera:

–
http://www.ksl.stanford.edu/software/chimaera/

–
Ontology evolution environment (diagnostics, merging, light editing)

–
Mostly for DAML+OIL, being updated to export and inport current OWL

•
Extensive list of tools,

–
http://www.w3.org/2001/sw/WebOnt/impls

Earth Sciences Sector

Web Sémantique

Conclusion

45

Conclusion

•
Semantic Web from T. Burners
-
Lee perspective is:


•
Data interoperability

across applications and
organizations (for IT)

•
A set of
interoperable standards for knowledge exchange


•
An
architecture for interconnected communities and
vocabularies



•
Importance of
URIs and ontologies

•
One URI denotes one concept

46

Conclusion

•
Similitudes importantes entre le Web Sémantique et les
travaux sur l’interopérabilité des données
géographiques

•
ISO/TC 211 amorce un réalignement de ses activités de
normalisation dans le but de profiter des effets du Web
Sémantique et par le fait même d’y contribuer

–
Revue du modèle de référence (ISO19101)

–
Description des modèles UML en OWL

–
Mise à jour du langage de schéma conceptuel (ISO/TS19103)

–
…

Earth Sciences Sector

Questions

Merci