Rule Responder Agents inRule Responder Agents inVirtual OrganizationsVirtual Organizations

Harold BoleyHarold Boley

Benjamin Craig Benjamin Craig

Institute for Information TechnologyInstitute for Information TechnologyNational Research Council, Canada National Research Council, Canada

Fredericton, NB, CanadaFredericton, NB, Canada

Talk atTalk atInstitute of Computer TechnologyInstitute of Computer Technology

Faculty of Electrical Engineering and Information Faculty of Electrical Engineering and Information Technology Vienna University of Technology, AustriaTechnology Vienna University of Technology, Austria

December 3, 2008December 3, 2008

22

OutlineOutline Rule Responder OverviewRule Responder Overview AgentsAgents

Personal / Organizational / ExternalPersonal / Organizational / External InfrastructureInfrastructure

Reaction RuleML MessagesReaction RuleML Messages Message PerformativesMessage Performatives Agent Communication ProtocolsAgent Communication Protocols Mule ESB (Communication Middleware)Mule ESB (Communication Middleware)

Rule Engines (for Realizing Agents)Rule Engines (for Realizing Agents) ProvaProva OO jDREWOO jDREW

Symposium Planner Use CaseSymposium Planner Use Case Query Delegation/AnsweringQuery Delegation/Answering Shared Knowledge between PasShared Knowledge between Pas Ontology DescriptionOntology Description

Future Work and ConclusionFuture Work and Conclusion

33

Overview of Rule Responder Overview of Rule Responder (I)(I)

Rule Responder is an experimentalRule Responder is an experimentalmulti-agent system for collaborative multi-agent system for collaborative teams and virtual communities on the teams and virtual communities on the WebWeb

Supports rule-based collaboration Supports rule-based collaboration between the distributed members of between the distributed members of such a virtual organizationsuch a virtual organization

Members of each virtual organization Members of each virtual organization are assisted by semi-automated rule-are assisted by semi-automated rule-based agents, which use rules to based agents, which use rules to describe the decision and behavioral describe the decision and behavioral logiclogic

44

Overview of Rule Responder Overview of Rule Responder (II)(II)

Uses languages and engines of the Uses languages and engines of the RuleML family for rule serialization, RuleML family for rule serialization, based on logic and XML: based on logic and XML: Hornlog RuleML: ReasoningHornlog RuleML: Reasoning Reaction RuleML: Interaction Reaction RuleML: Interaction

Implemented on top of a Mule-basedImplemented on top of a Mule-basedService Oriented Architecture (SOA)Service Oriented Architecture (SOA)

55

Personal AgentsPersonal Agents

A personal agent assists a single personA personal agent assists a single person

of an organization, (semi-autonomously) of an organization, (semi-autonomously)

acting on his/her behalfacting on his/her behalf It contains a FOAF*-like It contains a FOAF*-like factfact profile plus profile plus

FOAF-extending FOAF-extending rulesrules to encode selected to encode selected

knowledge of its human ownerknowledge of its human owner

* The Friend of a Friend (FOAF) project: http://www.foaf-project.org

66

Organizational AgentsOrganizational Agents

An organizational agent represents An organizational agent represents goals and strategies shared by each goals and strategies shared by each member of the organizationmember of the organization

It contains rule sets that describe the It contains rule sets that describe the policies, regulations, opportunities, policies, regulations, opportunities, etc. of its organization etc. of its organization

77

External AgentsExternal Agents External agents exchange messages External agents exchange messages

with (the public interface of) with (the public interface of) organizational agents, asking queries, organizational agents, asking queries, receiving answers, or interchanging receiving answers, or interchanging complete rule sets complete rule sets

End users, as external agents, employ a End users, as external agents, employ a Web (HTTP) interface of Rule Responder Web (HTTP) interface of Rule Responder (currently an API-like browser interface)(currently an API-like browser interface)

Support for simultaneous external Support for simultaneous external agents: Currently, end users (B2C)agents: Currently, end users (B2C)Ultimately, other organizations (B2B)Ultimately, other organizations (B2B)

88

Rule Responder as a Rule Responder as a Multi-Agent InfrastructureMulti-Agent Infrastructure

Realizes a System of OAs, PAs, and EAsRealizes a System of OAs, PAs, and EAs Built on the Mule ESBBuilt on the Mule ESB The OAs and PAs are realized each with The OAs and PAs are realized each with

an instance of a Rule Enginean instance of a Rule Engine Combines the ideas of multi-agent Combines the ideas of multi-agent

systems, distributed rule management systems, distributed rule management systems, as well as service-oriented and systems, as well as service-oriented and event-driven architecturesevent-driven architectures

99

Infrastructure - OverviewInfrastructure - Overview

1010

Translation Between PAs' Native Languages and OA's Interchange

Language Each rule engine can use its own rule Each rule engine can use its own rule

languagelanguage Agents require an interchange language Agents require an interchange language

so that they can understand each otherso that they can understand each other Rule Responder uses Reaction RuleML Rule Responder uses Reaction RuleML

as its interchange languageas its interchange language Translation is done with an XSLT Translation is done with an XSLT

stylesheetstylesheet

1111

Reaction RuleMLReaction RuleML

Reaction RuleML is a branch of the Reaction RuleML is a branch of the RuleML family that supports actions and RuleML family that supports actions and events events

When two agents need to communicate, When two agents need to communicate, each others’ Reaction RuleML messages each others’ Reaction RuleML messages are sent through the ESBare sent through the ESB

Carries RuleML queries, answers, and Carries RuleML queries, answers, and rule bases to/from agentsrule bases to/from agents

1212

Example Reaction RuleML MessageExample Reaction RuleML Message <RuleML xmlns="http://www.ruleml.org/0.91/xsd"<RuleML xmlns="http://www.ruleml.org/0.91/xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.ruleml.org/0.91/xsdxsi:schemaLocation="http://www.ruleml.org/0.91/xsd http://ibis.in.tum.de/research/ReactionRuleML/0.2/rr.xsd"http://ibis.in.tum.de/research/ReactionRuleML/0.2/rr.xsd" xmlns:ruleml2007="http://ibis.in.tum.de/projects/paw#">xmlns:ruleml2007="http://ibis.in.tum.de/projects/paw#"> <Message mode<Message mode="outbound"="outbound" directive directive="query-="query-

sync">sync"> <oid> <oid> <Ind>RuleML-2008</Ind> <Ind>RuleML-2008</Ind> </oid></oid> <protocol><protocol> <Ind>esb</Ind> <Ind>esb</Ind> </protocol></protocol> <sender><sender> <Ind>User</Ind> <Ind>User</Ind> </sender></sender> <content><content> <Atom><Atom> <Rel>getContact</Rel><Rel>getContact</Rel>

<Ind>ruleml2008_PanelChair</Ind><Ind>ruleml2008_PanelChair</Ind> <Ind>update</Ind><Ind>update</Ind> <Var>Contact</Var><Var>Contact</Var> </Atom></Atom> </content></content> </Message></Message> </RuleML></RuleML>

1313

Message PerformativesMessage Performatives

The attribute The attribute directive="..."directive="..." corresponds to the pragmatic performativecorresponds to the pragmatic performative

Specify message exchange/interaction Specify message exchange/interaction protocolsprotocols

Rule Responder PerformativesRule Responder Performatives In tradition of KQML and FIPA-ACLIn tradition of KQML and FIPA-ACL Currently implemented: Query and Currently implemented: Query and

AnswerAnswer Retract and Update in collaboration with Retract and Update in collaboration with

RIF-PRDRIF-PRD

1414

Agent Communication Agent Communication ProtocolsProtocols

WSDL-like protocolsWSDL-like protocols In-OnlyIn-Only

Message is sent to agentMessage is sent to agent1 1 from agentfrom agent22; then agent; then agent11 executes performativeexecutes performative

Request-ResponseRequest-Response Performs above in-only; then agentPerforms above in-only; then agent11 sends sends

response back to agentresponse back to agent22

Request-Response-AcknowledgeRequest-Response-Acknowledge Does Request-Response; then agentDoes Request-Response; then agent22 sends a sends a

response back to agentresponse back to agent11

WorkflowsWorkflows Generalizes the above protocols to allow arbitrary Generalizes the above protocols to allow arbitrary

compositions of agent messagescompositions of agent messages

1515

Communication MiddlewareCommunication Middleware MuleMule Enterprise Service Bus (ESB) Enterprise Service Bus (ESB)

Mule* is used to create communication Mule* is used to create communication end points at each personal and end points at each personal and organizational agent of Rule Responderorganizational agent of Rule Responder

Mule supports various transport Mule supports various transport protocolsprotocols(e.g. HTTP, JMS, SOAP)(e.g. HTTP, JMS, SOAP)

Rule Responder currently uses HTTP and Rule Responder currently uses HTTP and JMS as transport protocolsJMS as transport protocols

* Mule – The open source SOA infrastructure: http://mulesource.com

1616

Rule EnginesRule Engines

Prova: Prolog + JavaProva: Prolog + Java

OO jDREW: Object OrientedOO jDREW: Object Orientedjava Deductive Reasoning Engine for java Deductive Reasoning Engine for the Webthe Web

1717

ProvaProva

Prova is mainly used to realize the Prova is mainly used to realize the organizational agents of Rule organizational agents of Rule ResponderResponder

It implements Reaction RuleML for It implements Reaction RuleML for agent interaction (event-condition-agent interaction (event-condition-action rules)action rules)

1818

OO jDREW OO jDREW

OO jDREW is used to realize theOO jDREW is used to realize thepersonal agents of Rule Responderpersonal agents of Rule Responder

It implements Hornlog RuleML for It implements Hornlog RuleML for agent reasoning (Horn logic rules)agent reasoning (Horn logic rules)

Supports rules in two formats:Supports rules in two formats: POSL: POSL: Positional Slotted presentation Positional Slotted presentation

syntaxsyntax RuleML: RuleML: XML interchange syntaxXML interchange syntax

(can be generated from POSL) (can be generated from POSL)

1919

Use Case: Symposium Use Case: Symposium Planner Planner

RuleML-20xy SymposiaRuleML-20xy Symposia An organizational agent acts as the single point An organizational agent acts as the single point

of entry to of entry to assistassist with the symposium: with the symposium: Currently, query answering about the symposium Currently, query answering about the symposium Ultimately, preparing and running the symposium Ultimately, preparing and running the symposium

Personal agents have supported symposium Personal agents have supported symposium chairs since 2007 (deployed as Q&A in 2008)chairs since 2007 (deployed as Q&A in 2008)

General Chair, Program Chair, Panel Chair, Publicity General Chair, Program Chair, Panel Chair, Publicity Chair, etc.Chair, etc.

2020

Online Use Case DemoOnline Use Case Demo Rule Responder:Rule Responder:

http://responder.ruleml.orghttp://responder.ruleml.org

RuleML-2007/RuleML-2008 Symposia:RuleML-2007/RuleML-2008 Symposia:http://ibis.in.tum.de/projects/paw/ruleml-2007http://ibis.in.tum.de/projects/paw/ruleml-2007http://ibis.in.tum.de/projects/paw/ruleml-2008http://ibis.in.tum.de/projects/paw/ruleml-2008

Personal agents:Personal agents:Supporting all ChairsSupporting all Chairs

Organizational agent:Organizational agent:Supporting Symposium as a wholeSupporting Symposium as a whole

Online

2121

Query DelegationQuery Delegation

Query delegation to personal agents is Query delegation to personal agents is done by the organizational agentdone by the organizational agent

Tasks for the symposium organization are Tasks for the symposium organization are managed via a role assignment matrixmanaged via a role assignment matrix

Is defined here by an OWL Lite OntologyIs defined here by an OWL Lite Ontology(alternatives: RDFS, RuleML, ...)(alternatives: RDFS, RuleML, ...)

Assigns (meta)topics to agents within the Assigns (meta)topics to agents within the virtual organization: virtual organization: ... see next slide ...... see next slide ...

2222

Role Assignment OntologyRole Assignment Ontology

Publicity Chair

Topics

Agents Sponsoring

Publicity Chair

.

..

Panel Chair

General Chair

..

.

.

.

.

Challenge Chair

Challenge

Demos

Panel

Participants

Responsible

Accountable

Press Release

Challenge Chair

Panel Chair

General Chair

Metatopics...

.

..

.

..Registration

Visa Letter

.

..

Program Chair

.

..

.

..

Program Chair

Submissions

Presentations

.

..

2323

Multiple Query Answers by Multiple Query Answers by PAsPAs

Some queries have more than one answerSome queries have more than one answer

The PA will send the answers one at a The PA will send the answers one at a time to the OA (interleaved backtracking time to the OA (interleaved backtracking and transmission)and transmission)

When no more answers are computed, an When no more answers are computed, an end-of-transmission message is sent backend-of-transmission message is sent back

2424

Shared Knowledge Between Personal Agents

Rules can be shared among personal Rules can be shared among personal agentsagents

Rules that apply to all PAs can be Rules that apply to all PAs can be moved up to the OA levelmoved up to the OA level

... see next slide ...... see next slide ...

2525

Organizational SymposiumOrganizational Symposium Agent Knowledge Base Agent Knowledge Base

% Sample Prova-like rule (in POSL syntax)% Sample Prova-like rule (in POSL syntax)stored in the OA:stored in the OA:

getContact(?topic, ?request, ?contact) :-getContact(?topic, ?request, ?contact) :-

% Uses the topic and request to delegate % Uses the topic and request to delegate the following query to appropriate PAthe following query to appropriate PA

person(person(

?contact, ?role, ?title, ?email, ??contact, ?role, ?title, ?email, ?telephone).telephone).

2626

Personal Panel Chair Agent Personal Panel Chair Agent Knowledge BaseKnowledge Base

% Sample FOAF-like facts used by the OA rule:% Sample FOAF-like facts used by the OA rule:% Example fact stored in the Panel Chair’s PA% Example fact stored in the Panel Chair’s PAperson(John, PanelChair, PHD, person(John, PanelChair, PHD,

john@email.com, 1-555-555-5555)john@email.com, 1-555-555-5555)% Example fact stored in the Publicity Chair’s % Example fact stored in the Publicity Chair’s

PAPAperson(Tracy, PublicityChair, PHD, person(Tracy, PublicityChair, PHD,

tracy@email.com, 1-444-444-4444)tracy@email.com, 1-444-444-4444)% Sample query in RuleML syntax:% Sample query in RuleML syntax:... see next slide ...... see next slide ...

2727

Sample Message to Organizational Sample Message to Organizational AgentAgent

<RuleML xmlns="http://www.ruleml.org/0.91/xsd"<RuleML xmlns="http://www.ruleml.org/0.91/xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.ruleml.org/0.91/xsdxsi:schemaLocation="http://www.ruleml.org/0.91/xsd http://ibis.in.tum.de/research/ReactionRuleML/0.2/rr.xsd"http://ibis.in.tum.de/research/ReactionRuleML/0.2/rr.xsd" xmlns:ruleml2007="http://ibis.in.tum.de/projects/paw#">xmlns:ruleml2007="http://ibis.in.tum.de/projects/paw#"> <Message mode="outbound" directive="query-sync"><Message mode="outbound" directive="query-sync"> <oid><oid> <Ind>RuleML-2008</Ind><Ind>RuleML-2008</Ind> </oid></oid> <protocol><protocol> <Ind>esb</Ind><Ind>esb</Ind> </protocol></protocol> <sender><sender> <Ind>User</Ind><Ind>User</Ind> </sender></sender> <content><content> <Atom><Atom> <Rel>getContact</Rel><Rel>getContact</Rel> <Ind>ruleml2008_PanelChair</Ind><Ind>ruleml2008_PanelChair</Ind> <Ind>update</Ind><Ind>update</Ind> <Var>Contact</Var><Var>Contact</Var> </Atom></Atom> </content></content> </Message></Message> </RuleML></RuleML>

Online

http://www.ruleml.org/RuleML-2008/RuleResponder

Query Selection: Panel Chair Contact

2828

Architecture - ExecutionArchitecture - Execution

2929

Architecture - ExecutionArchitecture - Execution

3030

Architecture - ExecutionArchitecture - Execution

3131

Architecture - ExecutionArchitecture - Execution

3232

Architecture - ExecutionArchitecture - Execution

3434

3535

Sample Message to Publicity Chair Agent (I)Sample Message to Publicity Chair Agent (I) <content><content> <Atom><Atom> <Rel>sponsor</Rel><Rel>sponsor</Rel> <Expr><Expr> <Fun>contact</Fun><Fun>contact</Fun> <Ind>Mark</Ind><Ind>Mark</Ind> <Ind>JBoss</Ind><Ind>JBoss</Ind> </Expr></Expr> <Ind type="integer"><Ind type="integer">500500</Ind></Ind> <Expr><Expr> <Fun>results</Fun><Fun>results</Fun> <Var>Level</Var><Var>Level</Var> <Var>Benefits</Var><Var>Benefits</Var> <Var>DeadlineResults</Var><Var>DeadlineResults</Var> </Expr></Expr> <Expr><Expr> <Fun>performative</Fun><Fun>performative</Fun> <Var>Action</Var><Var>Action</Var> </Expr></Expr> </Atom></Atom> </content></content>

Online

http://www.ruleml.org/RuleML-2008/RuleResponder

Query Selection: Publicity Chair Sponsoring

3636

3737

Sample Message to Publicity Chair Agent (II)Sample Message to Publicity Chair Agent (II) <content><content> <Atom><Atom> <Rel>sponsor</Rel><Rel>sponsor</Rel> <Expr><Expr> <Fun>contact</Fun><Fun>contact</Fun> <Ind>Mary</Ind><Ind>Mary</Ind> <Ind>Super</Ind><Ind>Super</Ind> </Expr></Expr> <Ind type="integer"><Ind type="integer">50005000</Ind></Ind> <Expr><Expr> <Fun>results</Fun><Fun>results</Fun> <Var>Level</Var><Var>Level</Var> <Var>Benefits</Var><Var>Benefits</Var> <Var>DeadlineResults</Var><Var>DeadlineResults</Var> </Expr></Expr> <Expr><Expr> <Fun>performative</Fun><Fun>performative</Fun> <Var>Action</Var><Var>Action</Var> </Expr></Expr> </Atom></Atom> </content></content>

Online

http://www.ruleml.org/RuleML-2008/RuleResponder

Query Selection: Publicity Chair Sponsoring (edit)

3838

3939

Conclusion (I)Conclusion (I) Rule Responder was implemented & Rule Responder was implemented &

tested for various use cases tested for various use cases ((http://responder.ruleml.orghttp://responder.ruleml.org) and deployed for ) and deployed for RuleML-2008 Q&ARuleML-2008 Q&A

Its organizational agents delegate Its organizational agents delegate external queries to topic-assigned external queries to topic-assigned personal agentspersonal agents

It couples rule engines OO jDREW & It couples rule engines OO jDREW & Prova via Mule middleware and RuleML Prova via Mule middleware and RuleML 0.91 XML interchange format0.91 XML interchange format

4040

Conclusion (II)Conclusion (II) Without a Reaction Rule Dialect, RIF could not Without a Reaction Rule Dialect, RIF could not

be used for behavioral Responder logicbe used for behavioral Responder logic Current system is reusable on all levels:Current system is reusable on all levels:

Symposium Planner, Rule Responder, POSL, Symposium Planner, Rule Responder, POSL, RuleML, OO jDREW, Prova, MuleRuleML, OO jDREW, Prova, Mule

RuleML Techn. Group with Adrian Paschke, RuleML Techn. Group with Adrian Paschke, Alexander Kozlenkov and Nick Bassiliades:Alexander Kozlenkov and Nick Bassiliades:Looking for more ‘partner engines’ (mainly Looking for more ‘partner engines’ (mainly Flora-2) for use case, e.g. on RuleML FOAFFlora-2) for use case, e.g. on RuleML FOAF

4141

Future Work (I)Future Work (I)

Communication Between Personal Agent and Communication Between Personal Agent and Expert OwnerExpert Owner The PA at some point may need to interact The PA at some point may need to interact

with its expert ownerwith its expert owner The formal interaction between PAs and their The formal interaction between PAs and their

owners is email (SMTP)owners is email (SMTP) The interaction language of these emails is The interaction language of these emails is

Reaction RuleMLReaction RuleML Query DecompositionQuery Decomposition

Each premise of a rule can be delegated to Each premise of a rule can be delegated to different PAs, followed by Integrationdifferent PAs, followed by Integration

4242

Future Work (II)Future Work (II)

Centralized, Hierarchical (Distributed), and Centralized, Hierarchical (Distributed), and Networked (Distributed) Query AnsweringNetworked (Distributed) Query Answering

Centralized and Distributed Knowledge Centralized and Distributed Knowledge MaintenanceMaintenance How to keep your rules updatedHow to keep your rules updated

Distributed: Fault ToleranceDistributed: Fault Tolerance Alternative agents when an agent stops Alternative agents when an agent stops

workingworking Communication Overhead vs. Centralized Communication Overhead vs. Centralized

ProcessingProcessing