Semantic Web Technologies for Real World Applications€¦ · Semantic Web Technologies, Japan 2006...

Ulrich Reimer, Univ. of Applied Sciences St. Gallen 1Semantic Web Technologies, Japan 2006

Semantic Web Technologies forReal World Applications

Ulrich Reimer

Institute for Information and Process Management

University of Applied Sciences St. Gallen

Switzerland

2Ulrich Reimer, Univ. of Applied Sciences St. GallenSemantic Web Technologies, Japan 2006

Overview

1. Introduction

2. Application Areas for Semantic Web Technology

2.1 Content-Oriented Retrieval

2.2 Reference Modeling & Model-Driven Development

2.3 Information & Process Integration

2.4 Flexible Business Transactions

3. Opportunities, Barriers, and Future Development


getinformation

provide information,affect the world

Information Retrieval Web Services

The Web: Information retrieval and web services


What characterizes a Semantic Web application?

An application employing Semantic Web technology contains a model of the application domain:

• an ontology, i.e. concepts, their properties, relationships, axioms

• rules representing regularities, constraints, norms of the domain


Overview

1. Introduction








Time-consuming information search (1)

Many studies show:

A large percentage of daily work deals with information

(search, obtain, transform, distribute, file):

about 6 weeks/year for administrative workers

60 – 80% for knowledge workers

Recent study of a DMS company:

80% of interviewed companies need a whole day to find relevant information (although the DMS offers a search engine).

Reason:In many cases there are no guidelines for how to file documents.


Consequences of inefficient and insufficient information provision:

higher costs

longer process durations

longer time to market

important information is not found:

inferior quality of decisions, processes, etc.

compliance violations

reinventing the wheel

Time-consuming information search (2)


Information search with search engines

“tanker accident” atlantic

. . .

words contained in the documents

On 19 July 1979,the Atlantic Empressand the AegeanCaptain collided witheach other in theCaribbean Sea, offTobago island. ...

The tanker accident…

A single oil tankeraccident coulddestroy the entirecoastline. ...

Two tankers passedthe coast everyweek, transportingoil across theAtlantic to the USA. ...

The leaking oiltanker Prestige sinks off Spain'snorth-western coast...


Contents:tanker accident, Atlantic Ocean,oil spill, environmental pollution

Author: -Language: EnglishDokument type: news text

Using meta data to describe document contents

words contained in meta data




Contents:tanker collision, Atlantic Ocean,oil spill, environmental pollution


Different terms in meta data and query formulation




Contents:tanker collision, Gulf of Biscay,oil spill, environmental pollution



Different vocabulary for indexing and querying



tanker accident

accidentis-a

ship accident

is-a

train accident

. . . is-awater

inland water

is-a is-a

sea

Pacific Ocean

is-a

Atlantic Ocean

is-a . . .

Contents:tanker accident, Atlantic Ocean

Ontology

Controlled vocabulary for indexing and querying

Ontology: controlled vocabulary



Using background knowledge to extend query

tanker accident

Atlantic Ocean

tanker collision

Carribean Sea Bermuda Sea

part-of part-of

synonymy

Gulf of Biscay

part-of

(tanker collision OR tanker accident) AND (Atlantic Ocean OR Carribean Sea OR Bermuda Sea OR ...)

Ontology

Ontology: background knowledge

automaticquery extension



tanker accident

Atlantic Ocean

tanker collision

part-of part-of

synonymy

part-of

OntologySemanticsof relations



Ontology: background knowledge

Using background knowledge to extend query

Carribean Sea Bermuda SeaGulf of Biscay



Required background knowledge can be complex



document stillnot found




Technical realisation

What is needed:

• assign meta data to information objects

• content description with concepts and relations between them

• provision of background knowledge

• provision of the semantics of relations for query extension, ontology integration, etc.

RDF

RDF Schema, OWL,Rules


Ontology-based skills management

descriptionof skills

Ontology

Ulrich Reimer, Univ. of Applied Sciences St. Gallen 18Semantic Web Technologies, Japan 2006



Architecture


An ontology itself does not guarantee good query resultsSemantic query extension:

descend the concept hierarchy:described skill: WLANqueried skill: LAN

ascend the concept hierarchy:described skills: LANqueried skills: WLAN

Query extension (1)

LAN

WLAN

is-a


include siblings:described skill: Linuxqueried skill: BSD Unix

include synonyms:described skill: Voice over IPqueried skill: VoIP

Query extension (2)

VoIP Voice over IPsynonymy

operating system

Linux

is-a

BSD Unix

is-a


• an ontology is absolutely needed

• ontology development requires strong guidance by ontology experts

• keep ontology as small as possible

• query extension very important

Conclusion to ontology-based skills management


Overview

1. Introduction








Closure

Tool for Project Management

LogoutAdministrationControlling P-OfficeSearchMonitoringPlanning ExecutionInitialisation

Welcome Mr. John Q. Public 11/07/200311/07/2003Welcome Mr. John Q. Public 2004/05/18

PortfolioE-Gov-Project

Process analysisSOA

Web ServicesRegistryOrchestration

OntologyGUI

Forecast

∑7,850.00∑7,850.00∑4,250.00

Project information system

Save Edit Cancel


Modelling vs. Programming

Projectinformation system

Contractinformation system

etc.

hard-wired:

generic:

Genericinformation system

Projectontology

Contractsontology



Projectontology

Modelling vs. Programming

Portfolio

Project

Sub-project

Workpackage

MilestoneDocumentIssueClient

status

Responsible

Person

is-a

contains

has-part

has-part


Closure





Process analysisSOA


OntologyGUI

Forecast

∑7,850.00∑7,850.00∑4,250.00

How to present objects at the user interface?

Save Edit Cancel


Closure





Process analysisSOA


OntologyGUI

Forecast

∑7,850.00∑7,850.00∑4,250.00

How to present objects at the user interface?

Save Edit Cancel

Navigation

Functions


ClosureControllingPlanning Execution


ClientsIssuesMilestones Documents

Welcome Mr. John Q. Public 11/07/200311/07/2003

MonitoringSub-Projects

Project: E-Gov-Project 2006/05/18

ControllingExecution


GUIOntology


Process analysis

Status ReportDate

Responsible Start End Budget

April 06 Johnny B. Good1 March

20068 kEuro

Workpackages:

Web Services

Registry

Orchestration

SOA

Presenting subprojects and workpackages differently

versus

Project: E-Gov-Project










GUIOntology


Process analysis

Status ReportDate



20068 kEuro

Workpackages:

Web Services

Registry

Orchestration

SOA


versus



Portfolio

Project

Sub-project

Workpackage


status

Responsible

Person

is-a

contains

has-part

has-part

Navigation tree

proj-root

proj-node-l1

proj-node-l2

proj-tree “Portfolio”

is-ahas-root

has-successor

has-successor

label

Information objects

GUI objectspositionsizehas-size has-position

Mapping GUI objects to information objects


Portfolio

Project

Sub-project

Workpackage


status

Responsible

Person

is-a

contains

has-part

has-part

Navigation tree

proj-root

proj-node-l1

proj-node-l2

proj-tree “Portfolio”

is-ahas-root

has-successor

has-successor

label

Information objects


Mapping GUI objects to information objects: Rule


Portfolio

Project

Sub-project

Workpackage


status

Responsible

Person

is-a

contains

has-part

has-part

proj-tree

root-1

node-1

node-12

tree-1 “Portfolio”

“E-Gov-Project”

“Process analysis”

instance-ofhas-root

has-successorlabel

has-successor

label

label

e-gov-project

sp-1

has-part

instance-of

instance-of

Information objects



top-left

has-position










GUIOntology


Process analysis

Status ReportDate



20068 kEuro

Workpackages:

Web Services

Registry

Orchestration

SOA


versus



Portfolio

Project

Sub-project

Workpackage


status

Responsible

Person

is-a

contains

has-part

has-part

Tab sequence

tab-11tab-sequence-1 “Sub-Projects”

“SOA”

instance-offirst label

e-gov-project

sp-1

has-part

instance-of

instance-of

Information objects


tab-sequence-2tab-21firstlabel

“Process analysis”

tab-22

has-successorlabel



Project

Milestone

IssueClient

Document

Milestone

Issue

Client

Document

Project 1

Relationships in ontology: Navigation hierarchy:

Navigation hierarchy from relations between concepts

Milestone 1

Milestone 2

+-

++++


Project

Milestone

IssueClient

Document

Milestone

Issue

Client

Document

Project 1


Navigation hierarchy from relations between concepts

Milestone 1

Milestone 2

+-

++++

instance


Project

Sub-project

Workpackage

has-part

has-part

Navigation hierarchy from instances of related concepts

Sub-project 1

Sub-project 3

Sub-project 4

Sub-project 2

Project 1

Workpackage 11

Workpackage 12

+-

++++



Project

Sub-project

Workpackage

has-part

has-part

Navigation hierarchy from instances of related concepts

Sub-project 1

Sub-project 3

Sub-project 4

Sub-project 2

Project 1

Workpackage 11

Workpackage 12

+-

++++


instance


Project

Project

Navigation hierarchy from attribute values: classification

Attribute in ontology: Navigation hierarchy:

Projects: in time

Projects: delayed++

delay: { yes, no }


Project

Project

Navigation hierarchy from attribute values: classifiction

Attribute in ontology: Navigation hierarchy:

Projects: in time

Projects: delayed++

delay: { yes, no }

concept


Project Project

Navigation hierarchy from type of related instance


Client: Insurance

Client: Bank++

Client

Bank Insurance

is-ais-a


Project Project

Navigation hierarchy from attribute values


Client: Insurance

Client: Bank++

Client

concept

Bank Insurance

is-ais-a


app

licat

ion

alG

UI

Many aspects to model

domain ontology

rules

workflows

functions

roles & rights

GUI ontology

Some aspects are declarative

(i.e. model elements),

some are procedural

(i.e. pieces of code).


Reference modelling


for Automotive

Projectontology forAutomotive

Projectontology

Reference Project Model

Generic projectinformation system


for Acme Inc.

Projectontology

for Acme Inc.

Reference Automotive Project Model

Specialization

Specialization


Holistic Project Information System

Project Info System: Basic

Project Info System: Milestones

Project Info System: Budget

etc.

Aggregation of sub-models: Holistic vs. modular

Modular Project Information System

app

licat

ion

alG

UI

domain ontology

rules

workflows

functions

roles & rights

GUI ontology

applicational ontologies

GUI ontology


GUI ontology


GUI ontology


Reference modelling: Specialization


for Automotive


Projectontology




for Acme Inc.

Projectontology

for Acme Inc.


Controllinginformation system

for Automotive

Controllingontology forAutomotive

Controllingontology

Reference Controlling Model

Generic controllinginformation system


for Acme Inc.

Controllingontology

for Acme Inc.

Reference Automotive Controlling Model

Specialization

Specialization


Reference modelling: Specialization and Aggregation


for Automotive


Projectontology




for Acme Inc.

Projectontology

for Acme Inc.



for Automotive


Controllingontology




for Acme Inc.

Controllingontology

for Acme Inc.


Specialization

Specialization


Reference modelling: Specialization and Aggregation


for Automotive


Projectontology





for Automotive


Controllingontology




Specialization

Specialization


for Acme Inc.

Projectontology

for Acme Inc.


for Acme Inc.

Controllingontology

for Acme Inc.


project project manager

project staff

sub-project

workpackage

employee assignment

task

Ontology forResource Management

has-parthas-part

has-part

has

for

is-resp

has-assigned

Basic Project Ontology

Different concepts for same/similar thingshamper integration



project staff

sub-project

workpackage

employee assignment

taskhas-parthas-part

has-part

has

for

is-resp

has-assigned

project taskemployee

is-ais-ais-a

is-a

is-a

Ontology forResource ManagementBasic Project Ontology

Extension of sub-ontologies destroys independence

project task

is-a



project staff

sub-project

workpackage

employee assignment

taskhas-parthas-part

has-part

has

for

is-resp

has-assigned

project task employee

is-a

is-a

is-ais-a is-a is-a

is-a

Ontology forResource ManagementBasic Project Ontology

Upper-LevelOntology

Instead relate sub-ontologies to an upper-level-ontology


Project Info System: Basic Project Info System: Milestones

Upper-level ontology facilitates aggregation of sub-models

Upper-Level Ontology

Project Info System: MilestonesProject Info System: Basic

Upper-Level OntologyAggregatedmodel

+

=


GUI ontology


GUI ontology


GUI ontology


GUI ontology


Co

nfi

gu

rati

on

wo

rkb

ench

Model-driven development: From models to code




Projectontology

for Acme Inc.



for Acme Inc.

Compilation

Specialization

Run-time system


Overview

1. Introduction








A-Bank

Trustee

Company

Revenue tax declaration: Chaotic process withmany media breaks

B-Bank

Tax authority 1

Tax authority 2


tax documents

Process & information integration

A-Bank

Company

B-Bank

Tax authority 1

Tax authority 2

Trustee


Different process for each province

A-Bank

Company

B-Bank

Tax authority 1

Tax authority 2

Trustee

tax documents


tax documents

Reference process model for revenue tax declaration

A-Bank

Tax authority 1

Company

B-Bank

Tax authority 2

Reference Model:Revenue tax declaration

Trustee

generic process model


tax documents

Reference process model plus specializations of it

A-Bank

Trustee

Company

B-Bank

Reference Model:Revenue tax declaration

Specialized model for Province 1


Tax authority 1

Tax authority 2


Different vocabulary for each province

A-Bank

Company

B-Bank

Tax authority 1

Tax authority 2

Trustee

tax documents

Ontology 1

Ontology 2


Reference ontology with mappings to local ontologies

A-Bank

Company

B-Bank

Tax authority 1

Tax authority 2

Trustee

tax documents

Ontology 1

Ontology 2

Reference Ontology


Process participants can get different views

Reference Model: Revenue tax declaration

Law &regulations

Processdescriptions

Ontology


Law &regulations

Processdescriptions

Ontology


Law &regulations

Processdescriptions

Ontology

specialization

specialization

Company

Taxauthority 1

Taxauthority 2


How to get from models to IT systems?

????

OntologyOntologyOntology OntologyOntologyProcessdescriptions

OntologyOntologyLaw &regulations

Reference model with all specializations


Process control

WS

Company

WS

Tax authority

WS

Bank

WS

Trustee

electronictax documents

Orchestration Engine

Web Services

OrchestratedWeb Services

(BPEL)Rule Engine

OntologyOntologyOntology OntologyOntologyProcessdescriptions

OntologyOntologyLaw &regulations

Reference model with all specializationsModel-drivendevelopment:From themodelto SOA

Service-orientedarchitecture


WSWS WSWSWSWSWSWSWeb Services / SOA:• legacy independence• reusability• integration

Model-driven development: The ideal case

Business process view:• application-driven


WSWS WSWSWSWSWSWS


Top-down:From process modelto orchestratedweb service

Web Services / SOA:• legacy independence• reusability• integration



WSWS

WSWS

Top-down:From process modelto orchestratedweb service

WSWS WSWSWSWSWSWS





WSWS WSWSWSWSWSWS

Model-driven development: The reality



semantic gap


WSWS WSWSWSWSWSWS

Model-driven development: The reality

It does not fit!




WSWS WSWSWSWSWSWS

Model-driven development: The solution

Top-down:From the modelto orchestratedweb service

Bottom-up:Build the modelfrom the existingweb services




WSWS WSWSWSWSWSWS

Model-driven development: The solution

Top-down:From the modelto orchestratedweb service

Bottom-up:Build the modelfrom the existingweb services

WSWS

WSWS




Overview

1. Introduction








Travel industry: Emergence of new business models

after: M.Y. Kabbaj: Strategic and Policy Prospects for Semantic Web Services Adoption in US Online Travel Industry.M.Sc. Thesis. June 2003.

. . .

. . .

. . .

Current business models:

• manual bundling

• pre-fabricated travel packages

Advantages of travel packages:

• higher margins

• simple business processes

Disadvantages:

• fixed itineraries

• inflexible dates

• limited options


. . .

. . .

. . .

Dynamic Packaging: New business model

Advantages:

• single point of contact

• satisfaction of individual customer needs

• increased revenues by revenue management


Traditional Distribution Model

Travel Supplier /Inventory Owner

Tour Operator /Wholesaler

Retailer /Sales OrderProcessing

Consumer

New distribution model needed

Current distribution model impedes dynamic packaging


Retailer



Consumer

ElectronicDemand-driven

Marketplace

Sales OrderProcessing

New distribution model:Combine information from distributed sources


Requirements for dynamic packaging:

• reach across multiple markets, countries, currencies, suppliers

• automatic comparison of products

• automatic aggregation of products

• automatic negotiation of discounts

• trustworthy transactions

With current technology not efficient and not profitable!

Electronic marketplace needs Semantic Web Services


car rentalshotelsairlines

conventional registriesrequire programming

railway

Dynamic information gathering and transactions needed


machine understandable service descriptions

car rentalshotelsairlines

Dynamic information gathering and transactions needed

railway


Input: <place>Output: bag-of( url( <car-rentals>) )

Precondition:¬∃ r: ( known( car-rental( r ) ∧ place-of( r ) = <place> )∧ ¬known(¬∃ r: car-rental( r ) ∧ place-of( r ) = <place> )

Postcondition:∃ r: ( known( car-rental( r ) ∧ place-of( r ) = <place> )∨ known(¬∃ r: car-rental( r ) ∧ place-of( r ) = <place> )

Ontology

Semantic Web Services:Ontology-based service description



ElectronicDemand-driven

Marketplace

Sales OrderProcessing

business logicrepresented internally

business logicas rule markup

Business logic needed for negotiation and aggregation


The full potential of the Semantic Web

Dynamic

Static

Web ServicesUDDI, WSDL, SOAP

WWWURI, HTML, HTTP

Semantic Web ServicesOWL-S, WSMF, etc.

Semantic WebRDF Schema, OWL

SemanticTexts, images,

Web Service descriptions


Overview

1. Introduction








• Alleviation of information overload

• Querying facts instead of searching texts

• Integration of heterogeneous information sources

• Process integration

• Dynamic formation of business transactions

• Model-driven development of information systems

• Specialization and aggregation of reference models

Semantic Web Technologies: Opportunities


Semantic Web Technologies: Barriers

• Where do the ontologies come from?

But: Many ontologies already exist

• Where do the semantic content descriptions come from?Ontology development is expensive.

• Not many people have the expertise to build ontologies.

• How to achieve good cost/benefit ratio?


UMLS (Unified Medical Language System):• semantically related medical terms• terms from ca. 100 heterogeneous classification systems and medical

terminologies in 15 languages

UNSPSC (United Nations Standard Products and Services Code):• hierarchical classification system for products and services

Geneontology:• controlled vacabulary to describe genes and proteins • 19’000 concepts• is-a and part-of relationships

and many more

Ontologies do not necessarily need to be big!

Many ontologies or predecessors already exist


Conclusion

• Semantic Web Semantic Web technology

• Modeling will gain in importance over programming.

• Models will be ubiquitous because

– they are easier to create, modify, maintain

– they are easier to exchange and adapt (specialized, aggregated)

– versions can be better maintained (changes propagated top-down)

• Practical applications mostly need lightweight semantics, whereasmost research is about heavy-weight semantics.


Semantic Web Technologies for Real World Applications€¦ · Semantic Web Technologies, Japan 2006...

Documents

Transcript of Semantic Web Technologies for Real World Applications€¦ · Semantic Web Technologies, Japan 2006...