文库发布:1agent技术概要_第1页
文库发布:1agent技术概要_第2页
文库发布:1agent技术概要_第3页
文库发布:1agent技术概要_第4页
文库发布:1agent技术概要_第5页
已阅读5页,还剩58页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

软件Agent技术课程介绍钟元生2008.2.251IntroductiontoTechnologyDefinitionsofSoftwareAgents24.1.20062IntroductiontoTechnologyIntroductionLittlebithistory.

Mid50’sfirstideasaboutsoftwareagents.JohnMcCarthy,OliverG.Selfridge.Researchonsoftwareagentscanbesplitintotwomainstrands:线绳firstoneaboutmid70’sandsecondoneatthebeginningof90’s.Todaysoftwareagentsarestillahot,diverse变化多的topic.distributedartificialintelligence,robotics,artificiallife,distributedobjectcomputing,human-computerinteraction,intelligentandadaptiveinterfaces,intelligentsearchandfiltering,informationretrieval检索,etc.24.1.20063IntroductiontoTechnologyIntroductionAgents

Manydefinitions:

"Anagentisanythingthatcanbeviewedasperceivingitsenvironmentthroughsensorsandactinguponthatenvironmentthrougheffectors.”

"Autonomousagentsarecomputationalsystemsthatinhabitsomecomplexdynamicenvironment,senseandactautonomouslyinthisenvironment,andbydoingsorealizeasetofgoalsortasksforwhichtheyaredesigned."

"Intelligentagentsaresoftwareentitiesthatcarryoutsomesetofoperationsonbehalfofauseroranotherprogramwithsomedegreeofindependenceorautonomy,andinsodoing,employsomeknowledgeorrepresentationoftheuser'sgoalsordesires."24.1.20064IntroductiontoTechnologyIntroductionAgents

Manydefinitions:

“Letusdefineanagentasapersistent持久稳固的softwareentitydedicatedtoaspecificpurpose.‘Persistent’distinguishesagentsfromsubroutines子程序;agentshavetheirownideas主意abouthowtoaccomplishtasks,theirownagendas日程.‘Specialpurpose’distinguishesthemfromentiremultifunction多函数applications;agentsaretypicallymuchsmaller."

Anautonomousagentisasystemsituatedwithinandapartofanenvironmentthatsensesthatenvironmentandactsonit,overtime,inpursuitofitsownagendaand如此……以致于soastoeffectwhatitsensesinthefuture.24.1.20065IntroductiontoTechnologyIntroductionAutonomousAgentsBiologicalAgentsComputationalAgentsRoboticAgentsSoftwareAgentsArtificialLifeAgentsTask-specificAgentsEntertainmentAgentsVirusesCategoriesofagents(justoneofthem)机器人Agents生物Agents娱乐Agents特定任务的Agents24.1.20066IntroductiontoTechnologyIntroductionCategoriesofagents(justanotheroneofthem).

MobileagentsInterfaceagentsCollaborativeagentsInformationagentsReactive反应agentsHybrid混合agentsHeterogeneous异质agents24.1.20067IntroductiontoTechnologyIntroductiontoTechnologyWhatisanAgent?Attributes属性AutonomyGoal-orientedMobileCollaborativeTemporalcontinuityAdaptivityCommunicativeProactive主动的自治移动主动的适应性沟通的协作的目标导向24.1.20068IntroductiontoTechnologyIntroductiontoTechnologyObjectsdoit,becauseAgentsdoitformoney...24.1.20069IntroductiontoTechnologyIntroductiontoTechnologyAgents'Architectures24.1.200610IntroductiontoTechnologyIntroductiontoTechnologyAgentArchitecturesMaes更多:Anarchitectureproposesaparticularmethodologyforbuildinganautonomousagent.Itspecifieshowtheoverallproblemcanbedecomposedintosubproblems,i.e.howtheconstructionoftheagentcanbedecomposedintotheconstructionofasetofcomponentmodulesandhowthesemodulesshouldbemadetointeract.Thetotalsetofmodulesandtheirinteractionshastoprovideananswertothequestionofhowthesensordataandthecurrentinternalstateoftheagentdeterminetheactions(effectoroutputs)andfutureinternalstateoftheagent.Anarchitectureencompassestechniquesandalgorithmsthatsupportthismethodology.24.1.200611IntroductiontoTechnologyIntroductiontoTechnologyAgentArchitectures

Reactive反应的agentDeliberative协商的?agentHybrid混合的agent24.1.200612IntroductiontoTechnologyIntroductiontoTechnologyReactiveAgentReactiveagentshaveatmostaverysimpleinternalrepresentationoftheworld,butprovidetightcoupling紧耦合ofperception感知andaction行为Behaviour-basedparadigm范例Intelligenceisaproductofinteractionbetweenanagentanditsenvironment.24.1.200613IntroductiontoTechnologyIntroductiontoTechnologyReactiveAgentEachbehaviourcontinuallymapsperceptualinputtoactionoutputReactivebehaviour: action:S->AwhereSdenotesthestatesoftheenvironment,andAtheprimitiveactionstheagentiscapableofperform.Example:action(s)= Heateroff,iftemperatureOKHeateron,otherwise24.1.200614IntroductiontoTechnologyIntroductiontoTechnologySensorsEffectorsAgentStimulus-responsebehavioursState1State2StatenAction1Action2Actionn......ReactiveAgent24.1.200615IntroductiontoTechnologyIntroductiontoTechnologyDeliberative协商?AgentExplicit清晰的symbolicmodeloftheworldinwhichdecisionsaremadevialogicalreasoning,basedonpatternmatchingandsymbolicmanipulationsense-plan-actproblem-solvingparadigmofclassicalAIplanningsystems24.1.200616IntroductiontoTechnologyIntroductiontoTechnologySensorsEffectorsWorldModelPlannerPlanexecutorAgentDeliberativeAgent24.1.200617IntroductiontoTechnologyIntroductiontoTechnologyAgents(intelligent).Internalarchitecture:Agentsmaintainaninternalrepresentationoftheirworld.Thereisanexplicitmental精神stateCanbemodifiedbysomeformofsymbolicreasoning.BDIagentsBeliefDesireIntentions24.1.200618IntroductiontoTechnologyIntroductiontoTechnologyAgent–aBDImodelEnvironmentperceptionworldmodelmentalmodelsocialmodelroutine/emergencysituationslocalplanningsituationsco-operatingsituationsreactionslocalgoalsco-operativegoalsoptionsintentionsPOBlocalplansjointplansACTIONbeliefssituationsgoalsselectionexecutiongoalactivationBeliefgenerationsituationrecognitionplanningandscheduling24.1.200619IntroductiontoTechnologyIntroductiontoTechnologyHybridAgentCombinationofdeliberativeandreactivebehaviourAnagentconsistsofseveralsubsystemsSubsystemsthatdevelopplansandmakedecisionsusing symbolicreasoning(deliberativecomponent)Reactivesubsystemsthatareabletoreactquicklytoeventswithoutcomplexreasoning(reactivecomponent)Layeredarchitectures24.1.200620IntroductiontoTechnologyIntroductiontoTechnologyAgentSensorsEffectorsReactivecomponentState1State2StatenAction2Actionn......WorldModelPlannerPlanexecutorDeliberativecomponentAction1observationsmodificationsHybridAgent24.1.200621IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentofSoftwareAgents24.1.200622IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentofSoftwareagents

AgentsPlatformsCommunicationsOntologyApplications24.1.200623IntroductiontoTechnologyIntroductiontoTechnologyPlatform24.1.200624IntroductiontoTechnologyIntroductiontoTechnologyPlatform

AplacewhereagentslivenotalwaysneededAgentmanagementcreation–terminationsecurityAgentcommunicationservicesAgentdirectoryservices24.1.200625IntroductiontoTechnologyIntroductiontoTechnologyExampleAgentPlatformAgentPlatformAgentManagementSystemDirectoryFacilitatorAgentCommunicationChannelAgentBaselineProtocolSoftwareAgentCommunicationChannelFIPA24.1.200626IntroductiontoTechnologyIntroductiontoTechnologyPlatforms-FIPAImplementations

Jade

(TiLab-TelecomItaly)LEAP (EU-project)BlueJade

(HP)FIPA-OS (Nortel-Emorphia)

MicroFIPA-OS (HY)24.1.200627IntroductiontoTechnologyIntroductiontoTechnologyQuestions?Comments.24.1.200628IntroductiontoTechnologyIntroductiontoTechnologyCommunications24.1.200629IntroductiontoTechnologyIntroductiontoTechnologyAgentCommunications

Modelledathigherabstractionlevelthan’traditional’datacommunications.

SpeechActtheoryJ.R.Searle(1969)SpeechActsClassicexample:Presidentsays”Ideclarewar”

isactuallyanaction.24.1.200630IntroductiontoTechnologyIntroductiontoTechnologySpeechacts

PragmatictheoryoflanguagedealingwiththesyntaxandsemanticsACLmessagetypesusuallymodeledafterspeechactsLocutionaryactutteringasyntacticallyacceptablesentenceIllocutionaryact(speechact)communicatingofthespeakers’intentionstothehearerPerlocutionaryactbringingaboutoftheeffectontheheareroftheutterance24.1.200631IntroductiontoTechnologyIntroductiontoTechnologyAgentCommunicationsKnowledgesharingAgentCommunicationLanguageInteractionProtocolsContentLanguage24.1.200632IntroductiontoTechnologyIntroductiontoTechnologyAgentCommunications24.1.200633IntroductiontoTechnologyIntroductiontoTechnologyAgentCommunicationLanguage(ACL)

KQML

(KnowledgeQueryandManipulationLanguage)

FIPA-ACL

(AgentCommunicationLanguage)24.1.200634IntroductiontoTechnologyIntroductiontoTechnologyAgentCommunicationLanguageFIPA-ACL24.1.200635IntroductiontoTechnologyIntroductiontoTechnologyCommunicativeactsAccept-proposalAgree Cancel Cfp Confirm Disconfirm Failure Inform

Inform-if(macroact)Inform-ref(macroact)Not-understood

Propagate Propose Proxy Query-if Query-ref Refuse Reject-proposalRequest

Request-when Request-wheneverSubscribe 24.1.200636IntroductiontoTechnologyIntroductiontoTechnologyThedoorisopen?Openthedoor(forme)OK!I’llopenthedoorThedoorisopen IamunabletoopenthedoorIwillnotopenthedoorSaywhenthedoorbecomesopenAnyonewanttoopenthedoor?Icanopenthedoorforyou..atapriceDoor?What’sthat?Don’tunderstand...queryrequestagreeinformfailurerefusesubscribecfpproposenot-understoodCommunicativeActs(cf.Dr.Poslad’sLecture)24.1.200637IntroductiontoTechnologyIntroductiontoTechnologyParameters

:sender:receiver:content:reply-with:reply-by:in-reply-to:envelope:language:ontology:protocol:conversation-id24.1.200638IntroductiontoTechnologyIntroductiontoTechnologyExampleofrequest

Thesenderrequeststhereceivertoperformsomeaction.(request

:senderi

:receiverj

:content"open\"db.txt\"forinput"

:languagevb)24.1.200639IntroductiontoTechnologyIntroductiontoTechnologySemanticofRequest

<i,request(j,a)>

FP:FP(a)[i\j]

BiAgent(j,a)

BiPGjDone(a)

RE:Done(a)i=sender,j=receiver,a=actiontoperformDone(a)=actionperformedBiAgent(j,a)=agentibelievesthatjcanperforma

BiPGjDone(a)=idoesnotbelievethatDone(a)isj

’spersistentgoal24.1.200640IntroductiontoTechnologyIntroductiontoTechnologyContentlanguages

FIPA-SLsubsetsSL0,SL1,andSL2FIPA-RDF0RDFbasedcontentlanguageFIPA-CCLConstraintChoiceLanguageKIF&FIPA-KIF24.1.200641IntroductiontoTechnologyIntroductiontoTechnologyContentlanguages

KIF(KnowledgeInterchangeFormat)

(salary015-46-3946widgets72000)(salary026-40-9152grommets36000)(salary415-32-4707fidgets42000)

(>(*(widthchip1)(lengthchip1))

(*(widthchip2)(lengthchip2)))24.1.200642IntroductiontoTechnologyIntroductiontoTechnologyContentlanguagesFIPA-SL

FormallanguageusedtodefinethesemanticsoftheFIPAACL

Logicalpropositionsareexpressedinalogicofmentalattitudesandactions,formalisedinafirstordermodallanguagewithidentity24.1.200643IntroductiontoTechnologyIntroductiontoTechnologyContentlanguageFIPA-SLexample(query-ref:sender(agent-identifier:nameB):receiver(set(agent-identifier:nameA)):content

(iota?x(UKPrimeMinister?x)):languageFIPA-SL:reply-withquery2)

(inform:sender(agent-identifier:nameA):receiver(set(agent-identifier:nameB)):content

(=(iota?x(UKPrimeMinister?x))"TonyBlair"):languageFIPA-SL:in-reply-toquery2)

24.1.200644IntroductiontoTechnologyIntroductiontoTechnologyInteractionprotocolsSpecifyagentcommunicationpatterns.Example:24.1.200645IntroductiontoTechnologyIntroductiontoTechnologyFIPAinteractionprotocolsRequestQueryRequestWhenContractNetIteratedContractNetEnglishAuctionDutchAuctionBrokeringSubscribeBrokeringPropose24.1.200646IntroductiontoTechnologyIntroductiontoTechnologyQuestions?Comments.24.1.200647IntroductiontoTechnologyIntroductiontoTechnologyOntology24.1.200648IntroductiontoTechnologyIntroductiontoTechnologyOntology

Howdocommunicatingagentsunderstandeachother?

Systemswhichcommunicateandworktogethermustshareanontology.

Ontology

Acommonvocabularyandagreeduponmeaningstodescribeasubjectdomain.Anontologyisaspecificationofaconceptualization.Anontologyisadescriptionoftheconceptsandrelationshipsthatcanexistforanagentoracommunityofagents.24.1.200649IntroductiontoTechnologyIntroductiontoTechnologyOntology

Thesharedontologycanbeimplicitorexplicit.Implicitontologyaretypicallyrepresentedonlybyprocedures.Explicitontologyare(ideally)givenadeclarativerepresentationinawelldefinedknowledgerepresentationlanguage.24.1.200650IntroductiontoTechnologyIntroductiontoTechnologyOntologyExample24.1.200651IntroductiontoTechnologyIntroductiontoTechnologyOWL(WebOntologyLanguage)byW3COWLisdesignedforusebyapplicationsthatneedtoprocessthecontentofinformationinsteadofjustpresentinginformationtohumans.OWLisanextensiontoRDFandRDFSchemaThreesublanguages:OWLLiteclassificationhierarchyandsimpleconstraintsOWLDLmaximumexpressiveness,computationalcompleteness,anddecidabilityOWLFullmaximumexpressivenessandthesyntacticfreedomofRDFwithnocomputationalguarantees.24.1.200652IntroductiontoTechnologyIntroductiontoTechnologyOWLOWLextendsvocabularydescriptionallowingtoexpressclaimssuchas...Nothingcanbebothwineandbeer.grandParentandgrandChildareinverses.homepageisuniquelyidentifyingproperty.AW3CTeamPersonisaPersonwhoseworkplaceHomepageis/24.1.200653IntroductiontoTechnologyIntroductiontoTechnologyOWLOWLaddsvaluetotheSemanticWeballowingmachine-checkableontologydesignallowinginferringnewknowledgeallowingautomaticdetectionofcontradictionsprovidingprinciplesforcomplexdatamerging24.1.200654IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentsofAgent24.1.200655IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentsofsoftwareagents-anotherviewpoint

Accessiblevsinaccessible

Anaccessibleenvironmentisoneinwhichtheagentcanobtaincomplete,accurate,up-to-dateinformationabouttheenvironment’sstate.Mostmoderatelycomplexenvironments(including,forexample,theeverydayphysicalworldandtheInternet)areinaccessible.Themoreaccessibleanenvironmentis,thesimpleritistobuildagentstooperateinit.24.1.200656IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentsofsoftwareagents

Deterministicvsnon-deterministic

Adeterministicenvironmentisoneinwhichanyactionhasasingleguaranteedeffect—thereisnouncertaintyaboutthestatethatwillresultfromperforminganaction.Thephysicalworldcantoallintentsandpurposesberegardedasnon-deterministic.Non-deterministicenvironmentspresentgreaterproblemsfortheagentdesigner.24.1.200657IntroductiontoTechnologyIntroductiontoTechnologyEnvironmentsofsoftwareagents

Episodicvsnon-episodic

Inanepisodicenvironment,theperformanceofanagentisdependentonanumberofdiscreteepisodes,withnol

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论