semantic web semantic web-based agent applications based agent applications – from design to...

37
Semantic Web Semantic Web - - based Agent Applications based Agent Applications from Design to Collaborative Deployment from Design to Collaborative Deployment Sabin C. Buraga Faculty of Computer Science “A.I.Cuza” University of Iasi, Romania http://www.infoiasi.ro/~busaco/ Teodora Rusu “P.Poni” Institute of Macromolecular Chemistry, Iasi Marius Cioca Faculty of Engineering, “L. Blaga” University of Sibiu

Upload: sabin-buraga

Post on 15-Jan-2015

1.452 views

Category:

Technology


5 download

DESCRIPTION

 

TRANSCRIPT

Page 1: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Semantic WebSemantic Web--based Agent Applications based Agent Applications –– from Design to Collaborative Deploymentfrom Design to Collaborative Deployment

Sabin C. Buraga Faculty of Computer Science

“A.I.Cuza” University of Iasi, Romaniahttp://www.infoiasi.ro/~busaco/

Teodora Rusu“P.Poni” Institute of Macromolecular Chemistry, Iasi

Marius Cioca Faculty of Engineering, “L. Blaga” University of Sibiu

Page 2: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 2

I*P

RO

MS

200

7I*

PR

OM

S 20

07

AgendaAgenda

• Introduction

• Semantic Web – short description

• Multi-agent systems– Omega – multi-agent system infrastructure

– ADF – framework for agent developing

• Case studies

• Conclusions & Further Work

Page 3: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 3

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Web became a distributed environment for sharing information, without concerns for its type

• Information can hardly be found and processed–in a uniform and intelligent manner–by heterogeneous applications

IntroIntro

Page 4: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 4

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• A solution is provided by the agent-oriented technologies– A problem can be divided into smaller parts

(objectives), which can be solved independently problem modularization and robust

distributed calculation – We’ll describe 2 original agent-oriented systems,

designed and developed based on semantic Web technologies, to capture, represent and manage the knowledge available on Web

IntroIntro

Page 5: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 5

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Web ≡

data storage & presentation (Berners-Lee, 1989)

• Semantic Web (Berners-Lee et al., 2001)–Relations between objects that can be

identified ⇒ unified management of distributed information

–Information needs no longer to be intended for human readers only, but also for machine processing

Semantic WebSemantic Web

Page 6: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 6

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Semantic Web is…–A vision of Sir Tim Berners-Lee

(the creator of World Wide Web)–A transition from “classical” Web

to a more structured and flexible one–A way of describing data, in terms of

metadata (data about data) and explain certain relations between resources

Semantic WebSemantic Web

Page 7: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 7

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• From “opaque” documents to machine understandable data (resources + links between them)

• A plenty of tools, applications, services, expertise available

Semantic WebSemantic Web

Page 8: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 8

I*P

RO

MS

200

7I*

PR

OM

S 20

07

RDF (Resource Description Framework)– Defining & describing data and

relations among data

– Suitable to attach metadata to certain resources

– Data modeling must make the implicit explicit: “Teodora is a person.” “I*PROMS is an event attended by Teodora.” “Teodora knows Sabin.”

Some people can understand these assertions. How about the machines?

Semantic Web: technologiesSemantic Web: technologies

Page 9: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 9

I*P

RO

MS

200

7I*

PR

OM

S 20

07

RDF (Resource Description Framework)– A triple {subject, property, object} model

hasWebsite ("#iproms", "http://www.iproms.org/") isA ("#teodora", "#person") knowns ("#teodora", "#sabin")

– Triples of URIs (Uniform Resource Identifiers)

– Many syntactic representations, including XML (Extensible Markup Language)

Semantic Web: technologiesSemantic Web: technologies

Page 10: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 10

I*P

RO

MS

200

7I*

PR

OM

S 20

07

RDF (Resource Description Framework)<rdf:RDF>

<rdf:Description rdf:about="http://www.iproms.org/"><s:Title>A great online conference.</s:Title><s:Participants s:year="2007" s:total="300" />

</rdf:Description></rdf:RDF>

Semantic Web: technologiesSemantic Web: technologies

Our defined metadata vocabulary

Page 11: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 11

I*P

RO

MS

200

7I*

PR

OM

S 20

07

OWL (Web Ontology Language)– Structuring and characterizing

resources/relations (e.g., “Teodora” is an instance of “Person” class)

– Defining logical relationships among resources and the relationships (1-to-1, 1-to-many,…)

– Based on RDF

– Used to denote taxonomies, thesauri, ontologies

Semantic Web: technologiesSemantic Web: technologies

Page 12: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 12

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Agent paradigm is one of the promising technologies for information retrieval in general and for certain types of resource discovery in particular

• Mobile agents provide a way to think about solving software problems in a distributed environment (like the WWW)

AgentsAgents

Page 13: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 13

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Important attributes (Bradshow, 1997):– reactive – ability to respond to changes

within agent environment – autonomous – able to exercise control

over its own actions– goal-oriented – having a planned aim– communicative – ability to communicate

with other agents, by exchanging information and/or knowledge

– mobile – agents can transfer themselves from one host to another

AgentsAgents

Page 14: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 14

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Agents can be integrated into a coherent and consistent software system – e.g., a multi-agent system – in which they work together to better meet the needs of the entire application–utilizing autonomy, responsiveness,

pro-activeness and social ability

AgentsAgents

Page 15: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 15

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• A multi-agent system infrastructure• Implemented in C++• Started as a project in 2002• Offers an addressing space (viewed as a tree)

for the Web objects and different techniques to remotely access the Web distributed resources

• Details in (Alboaie & Buraga, 2004) and (Buraga & Alboaie, 2005)

OmegaOmega

Page 16: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 16

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Each object can be viewed as a collection of objects included in that one– Links (edges) between the vertices of the tree are

given by the aggregation relationship exposed by the object-oriented methodologies

• A basic scripting support is also available via OmegaKernel component

OmegaOmega

Page 17: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 17

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• For each object, different metadata constructs (expressed in RDF) are attached to specify semantic properties

• We are using different XML vocabularies• Relations between agents are expressed

by RDF statements• Omega was used within ITW system

(Buraga & Rusu, I*PROMS 2006)

OmegaOmega

Page 18: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 18

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• For each Web resource, a RDF document is generated in order to: – attach metadata: location, type (e.g., XHTML

page, JPEG image, etc.), owner, access method, timestamp of last modification – XFiles (Buraga, 2002) language based on XML

– express temporal information regarding the relationship between two resources – TRSL – Temporal Relation Specification Language (Buraga & Ciobanu, 2002; Buraga & Gabureanu, 2003) – based on ITL (Interval Temporal Logic)

OmegaOmega

Page 19: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 19

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Agent Development Framework• Second, more recent (since 2004),

open-source Java-based approach• Built as a Service-Oriented Architecture

(SOA) to assure loose coupling between the interacting software agents

• http://adf.sourceforge.net/

ADFADF

Page 20: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 20

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Main goals (Nichifor & Buraga, 2004; Hritcu & Buraga, 2005):–Interoperability–Flexibility and Extensibility–Scalability–Platform independence–Easy to use–Pragmatism

ADFADF

Page 21: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 21

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Agents invoke and use the services provided by other agents– to be able to offer their own specialized services

• Every agent can specialize only in several tasks, while delegating the remaining tasks to other expert agents separation of concerns

• A provider can register its services and a potential consumer to discover the providers that offer the services it needs, in a dynamic manner

ADFADF

Page 22: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 22

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Overview of the ADF platform:

ADFADF

Page 23: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 23

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Interaction between agents is done by exchanging descriptive messages through the standard interfaces

• Agent communication is message-oriented and asynchronous – Compatible with FIPA (Foundation for Intelligent

Physical Agents) model– Exchanging messages via a message transport

service: HTTP (HyperText Transport Protocol) or JMS (Java Messaging Service)

ADFADF

Page 24: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 24

I*P

RO

MS

200

7I*

PR

OM

S 20

07

1. Describing Omega agents via RDF– the use of RDF to capture the state of

the Omega internal objects2. Meeting Scheduling in ADF

– involves ADF agents collaborating in order to schedule a meeting on behalf of the users

Case StudiesCase Studies

Page 25: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 25

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• For each object stored within the Omega multi-agent system, we can attach different metadata – Assuring control versioning and security

(via permissions)• The system keeps these descriptions as RDF

assertions that can be transported to other objects during the information updating activities (i.e., object replication)

Case Study #1Case Study #1

Page 26: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 26

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• To catch the dynamics of the involved agents built within Omega and the links between them, a high-level RDF-based description of temporal relations is adopted – We use our TRSL (Temporal Relation

Specification Language), based on ITL –Interval Temporal Logic (Allen, 1991)

Case Study #1Case Study #1

Page 27: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 27

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Scenario: an object-maintainer Web agent can discover different temporal relations between the Omega objects that are distributed in the intranet of an organization and can automatically generate a TRSL document

Case Study #1Case Study #1

Page 28: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 28

I*P

RO

MS

200

7I*

PR

OM

S 20

07

<rdf:Bag rdf:id="RecentlyChanged"><rdf:li resource="object1" /><rdf:li resource="object2" /></rdf:Bag><rdf:Description rdf:about="#RecentlyChanged">

<f:Location f:dns="www.site.org">193.231.30.1 <!-- where objects are stored --></f:Location>

<f:Owner> <!-- metadata information --><rdf:Description rdf:about="http://www.omega.site/">

<f:Login f:uid="714">busaco</f:Login></rdf:Description></f:Owner>

...<!-- temporal information --><t:link t:type="temporal" t:action="Serialize" t:end="20070102T102033">

<t:Finishes t:dur="T000002" /></t:link>

</rdf:Description>

Case Study #1Case Study #1

Collection of objects to be serialized

The serialization action is planned to be performed on this date

Page 29: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 29

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• ADF agents are collaborating in order to schedule a meeting on behalf of their owners (for example, team workers)

• Each agent has access to relevant information about its owner’s identity, schedule and the relationships between the owner and other people – Schedule is represented in RDF – Relationships between persons is expressed

by using an extension of the FOAF (Friend of a Friend) vocabulary

Case Study #2Case Study #2

Page 30: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 30

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• The exchange of agent messages follows the Contract Net interaction protocol (Smith, 1980)

Case Study #2Case Study #2

Page 31: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 31

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• We presented two multi-agent systems – Omega and ADF – that are designed, implemented and deployed by using actual (semantic) Web technologies

• Our general goal was to present the advantages of metadata and ontologies used to rigorously capture, represent and manage the knowledge available on Web, in a pragmatic manner

ConclusionsConclusions

Page 32: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 32

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Our studies denote that various properties and relations established between the components (agents, objects, processes) of a multi-agent system can be expressed in a standardized and machine-understandable manner

• This approach could be an interesting solution for exchanging knowledge within an e-enterprise via the actual service oriented architecture technologies

ConclusionsConclusions

Page 33: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 33

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• We intend to study the integration of agent software into the social Web systems, such as blogs/wikis

• Other interest is to build a more complex multi-agent platform and to integrate it into a wide-area distributed system, such a Grid

Further workFurther work

Page 34: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 34

I*P

RO

MS

200

7I*

PR

OM

S 20

07

• Sabin Buraga thanks to L. Alboaie, S. Alboaie, C. Hritcu, and O. Nichifor for their collaboration regarding several aspects of the paper

• “P. Poni” Institute of Macromolecular Chemistry, Iasi, Romania is associated partner of the EU-funded FP6 Innovative Production Machines and Systems (I*PROMS) Network of Excellence

AcknowledgmentAcknowledgment

Page 35: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 35

I*P

RO

MS

200

7I*

PR

OM

S 20

07

Online ResourcesOnline Resources

• AgentLink: http://www.agentlink.org/• Agent Development Framework (ADF):

http://adf.sourceforge.net/• Foundation for Intelligent Physical Agents (FIPA):

http://www.fipa.org/• Semantic Web: http://www.semanticweb.org/• Web Services: http://www.webservices.org/• World-Wide Web Consortium:

http://www.w3.org/

• Our publications: http://www.infoiasi.ro/~busaco/pub.html

Page 36: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 36

I*P

RO

MS

200

7I*

PR

OM

S 20

07

SummarySummary

• Introduction

• Semantic Web – short description

• Multi-agent systems– Omega – multi-agent system infrastructure

– ADF – framework for agent developing

• Case studies

• Conclusions & Further Work

Page 37: Semantic Web Semantic Web-based Agent Applications based Agent Applications – from Design to Collaborative Deployment

Sabin Buraga, Teodora Rusu, Marius Cioca 37

I*P

RO

MS

200

7I*

PR

OM

S 20

07

Thank you for your attention!