Knowledge Content Objects: a grounded ontology for

blabbedharborΤεχνίτη Νοημοσύνη και Ρομποτική

23 Φεβ 2014 (πριν από 3 χρόνια και 3 μήνες)

48 εμφανίσεις

Knowledge Content Objects: a grounded ontology for
content management over the semantic web

Aldo Gangemi

Laboratory for Applied Ontology

Istitute of Cognitive Sciences and Technology

CNR, Rome

aldo.gangemi@istc.cnr.it

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Ontology application: the state of play

|
Ontologies used for any purpose, from domain to task to
application (service) modelling

|
Ontology building easier when limited to small communities for
small tasks in simple applications, and as a controlled terminology
(cf. FOAF)

|
Real world applications are more difficult: medium to large
organizations and communities, complex tasks (otherwise why
bothering?), not only controlled terminologies, distributed and
networked communities, need to manage inconsistency, conflicts,
evolution, provenance, severe integration problems

|
Ontologies for humanities pose specific problems, due to typical
problem solving (or problem dropping) procedures
---

interpretation
being a major challenge

|
Just like in human cognition, we need
patterns

emerging out of
complexity

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Related projects

|
DOLCE

and its extensions:
http://dolce.semanticweb.org

|
Reusable reference ontologies, in various logical languages

|
CIDOC2DOLCE: alignment and extension of CIDOC, ask me

|
OntoWordNet
:
http://www.w3.org/2001/sw/BestPractices/WNET

|
Migration of WordNet to formal ontology

|
Metokis
:
http://metokis.salzburgresearch.at

|
6th FP project, just finished, see this presentation

|
W3C

Semantic Web Best Practices and Deployment

Working
Group:
http://www.w3.org/2001/sw/BestPractices/

|
CODePs
: see my ISWC2005 paper: Ontology Design Patterns for
Semantic Web Content, web page with a repository asap

|
NeOn
:
http://www.neon
-
project.org

|
6th FP IP project: networked, collaborative ontologies, starts in March

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

How Metokis fits in the OE state of play

|
Application and content modelling are getting closer

|
Ontology patterns used in conceptual modelling

|
Case tools used in ontology engineering

|
KOS being reengineered or at least ported to semantic models

|
Abstraction step: going “outside the computational box”

|
Social tasks and knowledge: support for
knowledge collectives

|
Computational tasks should become a(n executable) part of a social
process of knowledge elicitation, negotiation, and sharing

|
Metokis ontologies are an attempt to fill the modelling gap between
social and computational tasks and knowledge

|
Plan Ontology

|
IO and KCO Ontology

|
Ontology Design Patterns

|
Everything based on the DOLCE foundational layer

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Modelli di progettazione ontologica

|
Design pattern
:, esempi di risoluzione di problemi generati da
analisi dei requisiti

|
In architettura (Alexander): modelli di progettazione da “buoni esempi”

|
Nel software engineering (Gamma et al.): componenti software
sperimentati

|
In ingegneria

ontologia?


|
Content Ontology Design Pattern
: modelli concettuali, esempi di
risoluzione di problemi generati da analisi dei requisiti

|
CODeP
logici
: soluzioni a partire da vincoli logici

|
CODeP
concettuali
: frammenti di ontologie, ricchi di relazioni e
costruitii per soddisfare le necessità degli esperti/utenti di un dominio

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Modularità: Stratificazione dei moduli

Domain

ontology

{Sculpture,Restoration, Mythical being, Caryatid, Doric order, Armilla, Fresco, …}


Core

ontology (specific domain
-
independent)


{Work of art, Painting technique, Author, Artistic period, Plastic art, Interpretation, …}


Foundational

ontology (domain
-
independent)

{Object, Process, Part, Time, Location, Representation, Plan, …}

Inherits from

Inherits from

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Criteri di costruzione:
Sfingi

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Quante (e
quali
) sfingi?

monument

plays_role

stoney object

to be restored

plays_role

instance
-
of

Greek mythical creature

instance
-
of

statue

subclass
-
of

refers_to

represents

English

expressed_

according_to

class

individual

word

Aegyptian mythical creature

instance
-
of

“sphinx”

The Sphinx

sphinx

Sphinx

sphinx

Sphinx

sphinx

metaphorical

blending

symbolic figure

instance
-
of

pharaoh

metaphorical

blending

*refers_to

* in naïve iconography

“Oedipus and the Sphinx”

represented_by

lexicalizes

lexicalizes

lexicalizes

lexicalizes

lexicalizes

lexicalizes

interpersonal role

instance
-
of

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Quali modelli di progettazione ho usato?

|
Classe


Individuo,
istanza_di
(individuo, classe)

|
Parola


Entità,
lessicalizza
(parola, entità)

|
Oggetto informativo


Individuo,
rappresenta
(
oggetto informativo
, individuo)

|
Oggetto informativo


Concettualizzazione,
esprime
(
Oggetto informativo
,
concettualizzazione)

|
Oggetto


Ruolo,
gioca
(oggetto, ruolo)

|
Individuo@spazio1


Individuo@spazio2,
fusione_metaforica
(
individuo@spazio1
,
individuo@spazio2
)


|
!!! Informalmente parlando … in realtà rappresentati in linguaggi logici
computabili (KIF, OWL(DL), CL, etc.)

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Cosa ci faccio?

Ex:
query

su dati
semistrutturati

(ex. documenti in un Semantic Web)


«tell me what works of art from ancient Egypt are related to European
works of art that also involve Greek mythology in a same cultural
unit»


(retrieve (?x1 ?x2)


(exists (?y ?z ?k)


(and


(
creation_3

?x1) (
creation_3

?x2)


(
non
-
physical
-
object

?y)


(or (and (
represents

?x1 ?y)


(
origin

?y
Classical_Greece
))


(exists ?w


(and (
non
-
physical
-
object

?w)


(
represents

?x1 ?w)


(
refers
-
to

?w ?y)


(
origin

?w
Classical_Greece
))))


(or (and (
represents

?x2 ?y)


(
origin

?y
Classical_Greece
))


(exists ?w1


(and (
non
-
physical
-
object

?w1)


(
represents

?x2 ?w1)


(
refers
-
to

?w1 ?y)


(
origin

?w1
Classical_Greece
))))


(
origin

?x1
EGYPTIAN_EMPIRE$EGYPT
)


(
origin

?x2
EUROPE
))))

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Altri modelli

|
Oggetto informativo


Oggetto fisico,
realizzato_da
(oggetto
informativo, oggetto fisico)

|
Oggetto informativo


Dati concreti,
codificato_da
(oggetto
informativo, dati concreti)

|
Oggetto fisico


Dati concreti,
digitalizzato_da
(oggetto fisico, dati
concreti)

|
Regolamento


situazione,
soddisfatto_da
(regolamento,
situazione)

|
sse
: tutti o alcuni dei ruoli, compiti, parametri, etc. definiti nel
regolamento hanno una corrispondenza nella situazione

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Altri enigmi per le ontologie relative ai beni
culturali

|
Artefatti object
-
oriented e process
-
oriented, ex.
scultura

vs.
musica

|
Periodi artistici: intervalli di tempo, eventi, o descrizioni di stili?

|
Autenticità: valutazione basata su criteri locali

|
Interpretazione:

|
entità rappresentate vs. figure simboliche, allusioni, rimandi
“intertestuali”

|
significato implicito ed esplicito, della critica e del pubblico

|
Metaphorical blending

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Riutilizzare l’esistente!

|
“Magazzini” culturali

|
Reingegnerizzazione di
legacy design patterns

|
Tassonomie e campioni materiali

|


|
Elicitazione dei modelli di razionalità degli esperti e degli
utenti reali

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

The basic DDPO Design Pattern

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Information objects: a foundation for content
description and KCO

[a
semiotic

ontology design pattern]

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Formally …


Class(DnS
:
information
-
object

complete




intersectionOf(



DOLCE
:
social
-
object




restriction(DnS
:
about

allValuesFrom(DOLCE
:
particular))




restriction(DnS
:
realized
-
by

someValuesFrom(DOLCE
:
information
-
realization))



restriction(DnS
:
interpreted
-
by

allValuesFrom(Actions
:
agent))



restriction(DnS
:
expresses

allValuesFrom(DnS
:
description))



restriction(DnS
:
ordered
-
by

someValuesFrom(DnS
:
information
-
encoding
-
system))))


|
The

definition

says

that

information

objects

(IO)
:

|
are

ordered

by

some

information

encoding

system

|
are

realized

by

some

particular

(an

“information

realization”)

|
can
express

a description [and, if that description is
satisfied by

a situation,]

|
can be
about
that situation, or some entity in its
setting

|
can be
interpreted

by agents that
internally represent

the description
expressed by

IOs

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

IO pattern for Dante’s Comedy

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Content objects

|
“Content” is ambiguous. WordNet (v2.0):

|
Depicted object

|
Sum of what is perceived, discovered, or learned

|
What a communication that is about something is about

|
Everything that is included in a collection

|
In KCO ontology, COs
realize

information objects

|
MP3 file, eBook, JPEG image, QT movie, etc.

|
Either
digital

(computational objects) or
analog
(non
-
digital)

|
Therefore, COs are
information realizations

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Semantic facets of (paid) content objects

|
Content

descriptions:
provides the propositional content that is announced
by a digital content on an abstract level.

|
Community

descriptions
: information about the organisational structure
(roles, rights and obligations) by which a content product can be used and
information that influences trust such as certificates and brand name
information.

|
Business

descriptions
: describes the business and legal requirements
during information and negotiation phases. After a purchase, contracts will
be enforced according to mutually agreed rights and obligations.

|
Presentation

descriptions
: describes the presentation modes to which the
information of a content product can be adapted to by rendering and other
application
-
specific means.

|
Trust and Security

descriptions
: content must be associated with some
measure of trust for the end user, and for the content provider, there must
be some security features which guarantee that the content will not be
illegally copied or otherwise misappropriated.

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Aligning Knowledge Content

Objects with a Foundational Ontology

|
Are

physical

data

structures

(
realize

a

frame

-

a

data

structure

-

for

content

objects
),

hence

they

are

information

realizations

|
Generic

KCO
:

realizes

a

maximally

generic

frame

(a

-
unique
-

ontological

schema)

|
Domain
-
anchored KCO
:
specializes

a generic KCO by realizing an ontological schema
about the community and business elements. Furthermore, it commits its instances to a
domain specific ontology which can be as broad as desired

|
KCO Prototype

(Master Copy):
instantiates

a domain
-
anchored KCO (by filling at least
some of the schema slots) and contains, at a certain point in time, a particular digital
content including all semantic annotations required by the corresponding domain
-
anchored KCO

|
KCO

Instance

(Copy)

:

it

also

instantiates

a

domain
-
anchored

KCO,

but

is

treated

as

a

“copy”

of

a

prototypical

KCO,

e
.
g
.

generated

when

a

user

is

granted

access

to

a

content

object
.

Depending

on

the

contract,

this

copy

can

change

over

time,

through

alterations

made

by

the

owner
.

For

example,

somebody

may

buy

a

MIDI

backing

track

for

a

pop

song,

in

order

to

modify

it,

e
.
g
.

by

adding

her

own

voice

to

it,

or

changing

some

parts

A kind of information
object in DOLCE!

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

KCO (master or copy layer) formalization


Class(KCO
:
KCO

complete



KCO
:
physical
-
data
-
structure



restriction(DnS
:
realizes

someValuesFrom(intersectionOf(



KCO
:
data
-
structure




restriction(KCO
:
instantiates

someValuesFrom(intersectionOf(



KCO
:
data
-
structure




restriction(DnS
:
realized
-
by

someValuesFrom(KCO
:
domain
-
KCO))))))))



restriction(KCO
:
realizes
-
frame
-
for

someValuesFrom(KCO
:
content
-
object
))



restriction(KCO
:
content

someValuesFrom(DOLCE
:
particular))




restriction(KCO
:
time
-
based
-
rendition

someValuesFrom(KCO
:
script))



restriction(KCO
:
interaction
-
based
-
rendition

someValuesFrom(KCO
:
script))



restriction(KCO
:
usage
-
context

someValuesFrom(DnS
:
plan))



restriction(KCO
:
user
-
task

someValuesFrom(DnS
:
task))



restriction(KCO
:
content
-
user

someValuesFrom(unionOf(



DnS
:
organization

Collectives
:
collective

DnS
:
role)))



restriction(KCO
:
contract
-
semantics



someValuesFrom(CommonSenseMapping
:
contract))



restriction(KCO
:
negotiation
-
semantics

someValuesFrom(KCO
:
negotiation
-
protocol))



restriction(KCO
:
pricing
-
semantics

someValuesFrom(KCO
:
pricing
-
scheme))



restriction(KCO
:
trust
-
value

someValuesFrom(KCO
:
trust
-
region))



restriction(KCO
:
mappable
-
to

allValuesFrom(KCO
:
ForeignClass)))

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Complementary axioms for KCO facets
(examples)


SubClassOf(



restriction(KCO
:
realizes
-
frame
-
for

someValuesFrom(



restriction(DnS
:
realizes

someValuesFrom(



restriction(DnS
:
about

someValuesFrom(DOLCE
:
particular)))))



restriction(KCO
:
content

someValuesFrom(DOLCE
:
particular)))



SubClassOf(



restriction(KCO
:
realizes
-
frame
-
for

someValuesFrom(



restriction(DnS
:
realizes

someValuesFrom(



restriction(DnS
:
expresses

someValuesFrom(DOLCE
:
particular)))))



restriction(KCO
:
content

someValuesFrom(DOLCE
:
particular)))



EquivalentClasses(



restriction(KCO
:
user
-
task

someValuesFrom(DnS
:
task))



restriction(KCO
:
usage
-
context

someValuesFrom(intersectionOf(



DnS
:
Plan



restriction(DnS
:
defines

someValuesFrom(DnS
:
task
)))))

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

CD: Content Description

CO: Community Semantics

BS: Business Semantics

PR: Presentation Semantics

SD: KCO Self
-
Description

AC: Access Semantics

BS

CD

CO

PR

AC

SD

KCO Structure

realises

information object

Foundational

Ontology

is
-
defined
-
in

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

CD: Content Description

CO: Community Semantics

BS: Business Semantics

PR: Presentation Semantics

SD: KCO Self
-
Description

AC: Access Semantics

BS

CD

CO

PR

AC

SD

KCO Structure

realises

information object

Foundational

Ontology

is
-
defined
-
in

Domain
-
Model

Domain

Ontology

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

CD: Content Description

CO: Community Semantics

BS: Business Semantics

PR: Presentation Semantics

SD: KCO Self
-
Description

AC: Access Semantics

BS

CD

CO

PR

AC

SD

KCO Structure

realises

information object

Foundational

Ontology

is
-
defined
-
in

Domain
-
Model

Domain

Ontology

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

CD: Content Description

CO: Community Semantics

BS: Business Semantics

PR: Presentation Semantics

SD: KCO Self
-
Description

AC: Access Semantics

BS

CD

CO

PR

AC

SD

KCO Structure

realises

information object

Foundational

Ontology

is
-
defined
-
in

Domain
-
Model

Domain

Ontology

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento


Resource
Integration S.


KCTP
Services


Task Services


Reasoning
Services


V


V


C


C


W


Ser


DSer


Send


Rec


Task

Execution

Env.


Reasoning

Engine


KCCA

Registry


KCCA

Manager


Domain
-
specific
Services


Applic
-
spec.

Services

KCO Operations

Propositional
content

...




KCO API

API extensions


Env. API

Applications / Interaction layer

GUI

KCCA

Network

Business
semantics

Community
description

Presentation
description

KCCA Services

KCCA architecture

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

Metokis service taxonomy

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

APPLICATION LAYER

Local System Registry

HTTP Message

HTTP Message

HTTP Sender/Receiver Servlet

HTTP Sender/Receiver

KCTP Router

De
-
Serializer

Serializer

HTTP Message

KCTP Message RDF
Model

KCTP Message
RDF Model

Check/ Get / Register

KCCA Service

KCCA Service

COMMUNICATING WITH

OTHER KCCA NODES

Response KCTP Message

Request KCTP Message

PRESENTATION LAYER

|_S
1
_| |_S
2
_| |_S
3
_|…………………...|_S
N
_|

SERVICES CONTAINER

Repository Services

KCO Services

Registry Services

METOKIS REPOSITORY API

WRAPPER

DATABASE

DATABASE

METOKIS WORLD

HTTP Message

KCTP Message

KCTP Message

KCTP

Message

Components in KCCA/KCTP.


KCTP Handler consists of KCTP Router, HTTP Sender Receiver, HTTP Sender
Receiver Servlet, De
-
Serializer and Serializer.

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

KCO in the real World

|
Three layers

|
Content

|
The actual
content

of the KCO

|
Human readable (e.g. contract.pdf, salzburg.jpg)

|
Machine readable (e.g. config.xml, rss_newsfeed.xml, ...)

|
Analog content (e.g. A book located at urn:room3:shelf2:loc43)

|
Content Description Layer

|
Access
,
format

and
encoding

information (MM
-
Facet)

|
Provides a referable
placeholder

to any content or content segment

|
Semantic Layer

|
The

actual
knowledge

of the KCO

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

KCO in the real World

|
This layering holds for all facets

|

! That means that
each facet may include content
, e.g.:

|
The
contract

facet may include the human readable contract.pdf as
content in addition to the semantic statements in the semantic layer
of the facet

|
The
propositional

content

facet provides a semantic description of
the domain, but the current and approved XML
-
Document can still
be used

|
Even the
MM
-
Metadata

facet may include a MPEG7 or Dublin Core
File in addition to the semantic statements part of this facet

Laboratorio di Ontologia Applicata, CNR
-
ISTC, Roma/Trento

KCO in the real World

|
Advantages of this approach

|
KCOs do not break existing Applications

|
Content of the KCO provides compatibility to existing components
(Even tasks provided by humans)

|
Newly developed XML
-
Standards can be used in combination with
KCOs

|
Seamless migration to Semantic Web Technologies

|
It’s not necessary to provide a “complete” ontological description of
the problem domain

|
Concentration on some key requirements is possible

|
Steadily improvement of the semantic description based on
feedback and future requirements