topic 8 intelligent agents

Upload: optimisticharish

Post on 07-Jul-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/18/2019 Topic 8 Intelligent Agents

    1/38

    ICT619 IntelligentICT619 Intelligent

    SystemsSystems

    Topic 8: Intelligent AgentsTopic 8: Intelligent Agents

  • 8/18/2019 Topic 8 Intelligent Agents

    2/38

    ICT619 2

     ntelligent Agents

    ntelligent Agents

    What is an intelligent agent?What is an intelligent agent?

    Why intelligent agents?Why intelligent agents?

    What intelligent agents can do for usWhat intelligent agents can do for us Characteristics of a good agentCharacteristics of a good agent

    Types of agentsTypes of agents

    Building intelligent agentsBuilding intelligent agents Intelligent agents in E-CommerceIntelligent agents in E-Commerce

    Intelligent agent design - state-of-the-artIntelligent agent design - state-of-the-art

    and futureand future

  • 8/18/2019 Topic 8 Intelligent Agents

    3/38

    ICT619 3

    What is an intelligent agent?What is an intelligent agent?

    Underlying concept -Underlying concept -

     An autonomous computational entity designed to perform a An autonomous computational entity designed to perform aspecific task, it!out direct initiation and continuous monitoring onspecific task, it!out direct initiation and continuous monitoring onpart of t!e user part of t!e user 

    "merged in t!e last 1# years or so"merged in t!e last 1# years or so $istinct from con%entional programs, in t!at it is$istinct from con%entional programs, in t!at it is automatic automatic 

     Additional properties& Additional properties& 'ome le%el of intelligence ()ased on any AI tec!nology from fi*ed'ome le%el of intelligence ()ased on any AI tec!nology from fi*ed

    rules to learning engines+ for decisions andor adaptation torules to learning engines+ for decisions andor adaptation toen%ironmental c!angeen%ironmental c!ange

     Acts reacti%ely, )ut also proacti%ely Acts reacti%ely, )ut also proacti%ely

    'ocial a)ility - communicates it! user, system, ot!er agents as'ocial a)ility - communicates it! user, system, ot!er agents asreuiredreuired

    .ig!t cooperate it! ot!er agents to carry out comple* tasks.ig!t cooperate it! ot!er agents to carry out comple* tasks  Agents mig!t mo%e from one system to anot!er to access remote Agents mig!t mo%e from one system to anot!er to access remote

    resources andor meet ot!er agentsresources andor meet ot!er agents

  • 8/18/2019 Topic 8 Intelligent Agents

    4/38

    ICT619 /

    What is an intelligent agent?What is an intelligent agent?

    (cont’d)(cont’d)

    Intelligent agents (also called 0softare agents+ do not necessarilyIntelligent agents (also called 0softare agents+ do not necessarilypossesspossess allall t!ese possi)le featurest!ese possi)le features

    ide range of %ariation in capa)ilities&ide range of %ariation in capa)ilities&

    'ome perform tasks indi%idually !ile ot!ers are cooperati%e'ome perform tasks indi%idually !ile ot!ers are cooperati%e

    'ome are mo)ile- a)le to mo%e across a netork, ot!ers are not'ome are mo)ile- a)le to mo%e across a netork, ot!ers are not 

    .ost communicate %ia coded messages or e%en natural language,.ost communicate %ia coded messages or e%en natural language,some dont communicate at allsome dont communicate at all 

    Multiple agentsMultiple agents ork in groups or sarms to sol%e pro)lemsork in groups or sarms to sol%e pro)lemscollecti%ely, some ork as indi%idual unitscollecti%ely, some ork as indi%idual units

    4ot all agents learn and adapt t!emsel%es4ot all agents learn and adapt t!emsel%es

    5o)ots are p!ysically em)odied agents5o)ots are p!ysically em)odied agents

  • 8/18/2019 Topic 8 Intelligent Agents

    5/38

    ICT619 #

    Why intelligent agents?Why intelligent agents?

    .ore and more e%eryday tasks )ecoming computer-)ased.ore and more e%eryday tasks )ecoming computer-)ased

     An increasing num)er of untrained users using computers An increasing num)er of untrained users using computers Current !uman-computer interfaces reuire users to initiate allCurrent !uman-computer interfaces reuire users to initiate all

    tasks and monitor t!em - manuallytasks and monitor t!em - manually

    Intelligent agents engage in a cooperati%e process it! t!e user toIntelligent agents engage in a cooperati%e process it! t!e user tole%erage t!e effecti%eness and efficiency of !uman-computerle%erage t!e effecti%eness and efficiency of !uman-computerinteractioninteraction

    'taggering grot! in information a%aila)ility'taggering grot! in information a%aila)ility Intelligent agents can )e a tool for relie%ing t!e user of t!isIntelligent agents can )e a tool for relie%ing t!e user of t!is

    information o%erloadinformation o%erload

    Intelligent agents can act as personal assistants to t!e user toIntelligent agents can act as personal assistants to t!e user tomanage informationmanage information

    .ig!t one day take o%er routine tasks in personal management.ig!t one day take o%er routine tasks in personal managementsuc! as appointments, meetings and tra%el arrangementssuc! as appointments, meetings and tra%el arrangements

  • 8/18/2019 Topic 8 Intelligent Agents

    6/38

    ICT619 6

    What intelligent agents can do forWhat intelligent agents can do for

    usus Carry out tasks on t!e users )e!alf Carry out tasks on t!e users )e!alf  Train or teac! t!e user Train or teac! t!e user 

    7elp different users colla)orate7elp different users colla)orate

    .onitor e%ents and procedures.onitor e%ents and procedures

    'pecifically, intelligent agents can !elp us it!'pecifically, intelligent agents can !elp us it!

    Information retrie%alInformation retrie%al

    Information filteringInformation filtering

    .ail management.ail management

    5ecreational acti%ities 8 selection of5ecreational acti%ities 8 selection of)ooks, music, !olidays)ooks, music, !olidays

    ooking of meetings, !otels, ticketsooking of meetings, !otels, tickets

  • 8/18/2019 Topic 8 Intelligent Agents

    7/38

    ICT619 :

    What intelligent agents can do forWhat intelligent agents can do for

    us (cont’d)us (cont’d)

    Information filtering agentInformation filtering agent

    ;ne type is t!e selection of articles from a continuous;ne type is t!e selection of articles from a continuousstream to suit particular user needsstream to suit particular user needs

    User can create 0nes agents and train t!em )y gi%ingUser can create 0nes agents and train t!em )y gi%ingpositi%e or negati%e feed)ack for articles recommendedpositi%e or negati%e feed)ack for articles recommended

    T!e use of key ords alone can )e restricti%eT!e use of key ords alone can )e restricti%e

    Underlying semantics must )e e*tracted for moreUnderlying semantics must )e e*tracted for moreeffecti%enesseffecti%eness

    "g

  • 8/18/2019 Topic 8 Intelligent Agents

    8/38

    ICT619 ?

    What intelligent agents can do forWhat intelligent agents can do for

    us (cont’d)us (cont’d)

    "lectronic mail agent"lectronic mail agent

     Assist users it! electronic mail Assist users it! electronic mail

    @earn to prioritie, delete, forard, sort and arc!i%e@earn to prioritie, delete, forard, sort and arc!i%email messages on )e!alf of t!e user mail messages on )e!alf of t!e user 

    .ay use intelligent system tec!niues like case-)ased.ay use intelligent system tec!niues like case-)asedreasoningreasoning

    Can associate a le%el of confidence it! its action orCan associate a le%el of confidence it! its action orsuggestionsuggestion

    Use of 0do-it and 0tell-me t!res!olds set )y user Use of 0do-it and 0tell-me t!res!olds set )y user  .ay in%ol%e multi-agent colla)oration.ay in%ol%e multi-agent colla)oration

  • 8/18/2019 Topic 8 Intelligent Agents

    9/38

    ICT619 9

    What intelligent agents can do forWhat intelligent agents can do for

    us (cont’d)us (cont’d)

    'election agents for entertainment'election agents for entertainment

    Con%ersational agents s!o potential forCon%ersational agents s!o potential for)ecoming popular and commercially)ecoming popular and commerciallysuccessful eg Cy)elle, A@IC"successful eg Cy)elle, A@IC"

    Use 0social filtering 8 correlation )eteen different usersUse 0social filtering 8 correlation )eteen different usersto make recommendations on )ooks, C$s, films etcBto make recommendations on )ooks, C$s, films etcB

    'o, if user'o, if user A A liked itemsliked items X  X  andand Y Y , and user, and user BB liked itemliked item X  X  

    andand Z Z , t!en item, t!en item Z Z  may )e recommended for usermay )e recommended for user A A

     AmaonBcom !as )een using t!is system for years - AmaonBcom !as )een using t!is system for years -

     

    7i, I am Cy)elleB

    !at is your nameD

  • 8/18/2019 Topic 8 Intelligent Agents

    10/38

    ICT619 1E

  • 8/18/2019 Topic 8 Intelligent Agents

    11/38

    ICT619 11

    What intelligent agents can do forWhat intelligent agents can do for

    us (cont’d)us (cont’d)

    'ome ot!er current and emerging applications of intelligent'ome ot!er current and emerging applications of intelligentagents&agents&

    air traffic controlair traffic control

    air craft mission analysisair craft mission analysis control of telecommunications and netork systemscontrol of telecommunications and netork systems pro%ision and monitoring of medical carepro%ision and monitoring of medical care monitoring and control of industrial processesmonitoring and control of industrial processes on-line fault diagnosis and malfunction !andlingon-line fault diagnosis and malfunction !andling

    super%ision and control of manufacturing en%ironmentssuper%ision and control of manufacturing en%ironments transactions management in )anks and insurancetransactions management in )anks and insurance

    companiescompanies "-commerce, tourism"-commerce, tourism

  • 8/18/2019 Topic 8 Intelligent Agents

    12/38

    ICT619 12

    Characteristics of a good agentCharacteristics of a good agent

     Action Action  Agent must )e a)le to take some action and not >ust Agent must )e a)le to take some action and not >ust

    pro%ide ad%icepro%ide ad%ice =resent state of e) tec!nology limits capa)ility of=resent state of e) tec!nology limits capa)ility of

    Internet agentsInternet agents

    - still no standard interface for agents, )ut agent- still no standard interface for agents, )ut agentcommunication languages suc! as AC@ and FG.@communication languages suc! as AC@ and FG.@mig!t in outmig!t in out

     As t!e Internet )ecomes more agent-friendly, more As t!e Internet )ecomes more agent-friendly, morecapa)le agents ill emergecapa)le agents ill emerge

     Autonomy Autonomy  An agent can )e muc! more useful if it can act An agent can )e muc! more useful if it can act

    autonomouslyautonomously

    T!e rig!t le%el of autonomy for a task must )e foundT!e rig!t le%el of autonomy for a task must )e found

  • 8/18/2019 Topic 8 Intelligent Agents

    13/38

    ICT619 13

    Characteristics of a good agentCharacteristics of a good agent

    (cont)(cont)

    CommunicationCommunication .ust communicate ell it! t!e user .ust communicate ell it! t!e user  '!ould understand users goals, preferences and constraints'!ould understand users goals, preferences and constraints Useful communication reuires s!ared knoledge onUseful communication reuires s!ared knoledge on

    language of communicationlanguage of communication

    pro)lem domainpro)lem domain

    "*ample =ro)lem& e) searc! engines"*ample =ro)lem& e) searc! engines accept key ords and p!rases (some knoledge of t!eaccept key ords and p!rases (some knoledge of t!e

    language+language+

    )ut)ut understand not!ing a)out t!e documents t!ey retrie%e (nounderstand not!ing a)out t!e documents t!ey retrie%e (no

    domain knoledge+domain knoledge+ 'olution& pro%ision of a mac!ine-reada)le'olution& pro%ision of a mac!ine-reada)le ontology ontology  

    - a definition of a )ody of knoledge including its- a definition of a )ody of knoledge including itscomponents and t!eir relations!ips components and t!eir relations!ips

  • 8/18/2019 Topic 8 Intelligent Agents

    14/38

    ICT619 1/

    Characteristics of a good agentCharacteristics of a good agent

    (cont)(cont)

     Adaptation Adaptation

    Can gain user confidence )y learning user preferencesCan gain user confidence )y learning user preferences

    .@ tec!niues suc! as A44', HAs or C5 can )e.@ tec!niues suc! as A44', HAs or C5 can )eusedused

     Adapting to user preferences can )e also ac!ie%ed )y Adapting to user preferences can )e also ac!ie%ed )yusing data mining tec!niues suc! as clusteringusing data mining tec!niues suc! as clustering

     Agent forms clusters of users it! similar features Agent forms clusters of users it! similar features

    Users needs can t!en )e anticipated )y placing t!eUsers needs can t!en )e anticipated )y placing t!e

    user in one of t!ese clusters and analysing t!e cluster user in one of t!ese clusters and analysing t!e cluster  'ocial pro)lem sol%ing met!od, similar to Amaon'ocial pro)lem sol%ing met!od, similar to Amaon

    recommendationsrecommendations

  • 8/18/2019 Topic 8 Intelligent Agents

    15/38

    ICT619 1#

    Ty!es of agentsTy!es of agents

    ased on operational c!aracteristics andased on operational c!aracteristics andfunctional o)>ecti%es&functional o)>ecti%es&

    Colla)orati%e agentsColla)orati%e agents ork toget!er toork toget!er to

    - integrate information and- integrate information and

    - negotiate it! ot!er agents to resol%e conflict- negotiate it! ot!er agents to resol%e conflict- =ro%ide solutions to in!erently distri)uted pro)lems,- =ro%ide solutions to in!erently distri)uted pro)lems,

    eBgB, air traffic controleBgB, air traffic control

    5eacti%e agents5eacti%e agents  Act )y stimulus-response to t!e current state of Act )y stimulus-response to t!e current state of

    t!e en%ironmentt!e en%ironment "ac! reacti%e agent is simple and interacts it!"ac! reacti%e agent is simple and interacts it!

    ot!ers in a )asic ayot!ers in a )asic ay

  • 8/18/2019 Topic 8 Intelligent Agents

    16/38

    ICT619 16

    Ty!es of agents (cont’d)Ty!es of agents (cont’d)

    Interface agentsInterface agents =ro%ide user support and assistance=ro%ide user support and assistance Cooperate it! user in accomplis!ing some task in anCooperate it! user in accomplis!ing some task in an

    applicationBapplicationB

    Interface agents learn&Interface agents learn& )y o)ser%ing and imitating t!e user)y o)ser%ing and imitating t!e user t!roug! recei%ing feed)ack from t!e usert!roug! recei%ing feed)ack from t!e user )y recei%ing e*plicit instructions)y recei%ing e*plicit instructions )y asking ot!er agents for ad%ice (from peers+)y asking ot!er agents for ad%ice (from peers+

    "*amples&"*amples& =ersonal assistants performing information filtering,=ersonal assistants performing information filtering,

    email managementBemail managementB

     

  • 8/18/2019 Topic 8 Intelligent Agents

    17/38

    ICT619 1:

    Ty!es of agents (cont)Ty!es of agents (cont)

    .o)ile agents.o)ile agents =rograms t!at migrate from one mac!ine to anot!erB=rograms t!at migrate from one mac!ine to anot!erB "*ecute in a platform-independent e*ecution en%ironment, like"*ecute in a platform-independent e*ecution en%ironment, like

    a%a applets running on a a%a %irtual mac!inea%a applets running on a a%a %irtual mac!ine =ractical )ut non-functional ad%antages&=ractical )ut non-functional ad%antages&

    5educed communication cost5educed communication cost

     Async!ronous computing (!en you are not connected+ Async!ronous computing (!en you are not connected+

  • 8/18/2019 Topic 8 Intelligent Agents

    18/38

    ICT619 1?

    Ty!es of agents (cont)Ty!es of agents (cont)

    To types of mo)ile agents&To types of mo)ile agents&

    ;ne-!op mo)ile agents (migrates to one ot!er;ne-!op mo)ile agents (migrates to one ot!er

    place+place+

    .ulti-!op mo)ile agents (roam t!e netork.ulti-!op mo)ile agents (roam t!e netorkfrom place to place+from place to place+

    "*ample applications&"*ample applications& $istri)uted information retrie%al$istri)uted information retrie%al

    Telecommunication netork routingTelecommunication netork routing

  • 8/18/2019 Topic 8 Intelligent Agents

    19/38

    ICT619 19

    Ty!es of agents (cont)Ty!es of agents (cont)

    Information agentsInformation agents .anage information.anage information .anipulate or collate information from many distri)uted.anipulate or collate information from many distri)uted

    sourcesBsourcesB Can )e mo)ile or staticBCan )e mo)ile or staticB "*amples&"*amples& argainJinder compares prices among Internet stores forargainJinder compares prices among Internet stores for

    C$sC$s asper orks on )e!alf of a user or community of users andasper orks on )e!alf of a user or community of users and

    stores, retrie%es and informs ot!er agents of usefulstores, retrie%es and informs ot!er agents of useful

    information on t!e information on t!e

  • 8/18/2019 Topic 8 Intelligent Agents

    20/38

    ICT619 2E

    Ty!es of agents (cont)Ty!es of agents (cont)

    .ultiple agent systems.ultiple agent systems Consist of collections, or sarms, of simple agents t!atConsist of collections, or sarms, of simple agents t!at

    interact it! eac! ot!er and t!e pro)lem en%ironmentinteract it! eac! ot!er and t!e pro)lem en%ironment Can )e mo)ile or static, same or different agentsCan )e mo)ile or static, same or different agents

    Comple* patterns of )e!a%iour emerge from collecti%eComple* patterns of )e!a%iour emerge from collecti%einteractioninteraction

    "*amples&"*amples& 'arm of )ees finds an optimal location for t!e !i%e'arm of )ees finds an optimal location for t!e !i%e ********

  • 8/18/2019 Topic 8 Intelligent Agents

    21/38

    ICT619 21

  • 8/18/2019 Topic 8 Intelligent Agents

    22/38

    ICT619 22

    "uilding intelligent agents"uilding intelligent agents

    To main pro)lems to o%ercome&To main pro)lems to o%ercome&

    CompetenceCompetence  7o do e )uild agents it! t!e knoledge needed to decide7o do e )uild agents it! t!e knoledge needed to decide

    !en to !elp t!e user !en to !elp t!e user 

    !at to !elp t!e user it!, and!at to !elp t!e user it!, and !o to !elp t!e userD!o to !elp t!e userD

    Trust Trust   7o to guarantee user comfort (and protectionK+ in7o to guarantee user comfort (and protectionK+ in

    delegating tasks to t!e agentdelegating tasks to t!e agent

     Approac!es to )uilding agents Approac!es to )uilding agents1.1. User-programmed agents - write specialised scriptsUser-programmed agents - write specialised scripts

    2.2. nowledge-!ased agentsnowledge-!ased agents

    ".". Mac#ine-learning approac#Mac#ine-learning approac#

  • 8/18/2019 Topic 8 Intelligent Agents

    23/38

    ICT619 23

    "uilding intelligent agents (cont’d)"uilding intelligent agents (cont’d)

    T!e main pro)lem it!T!e main pro)lem it! user-programmeduser-programmed approac!approac!

    - reuires !ig! le%el of user competency- reuires !ig! le%el of user competency

    - user must )e a)le to- user must )e a)le to

    5ecognise opportunity for employing an agent5ecognise opportunity for employing an agent Take initiati%e to create an agentTake initiati%e to create an agent Impart specific knoledge to agent )y codifying it in aImpart specific knoledge to agent )y codifying it in a

    special languagespecial language .aintain agents knoledge )y updating rule )ase it!.aintain agents knoledge )y updating rule )ase it!

    timetime

    T!e issue of trust is t!en reduced to users trust in t!eirT!e issue of trust is t!en reduced to users trust in t!eiron programming skillson programming skills

  • 8/18/2019 Topic 8 Intelligent Agents

    24/38

    ICT619 2/

    "uilding intelligent agents (cont)"uilding intelligent agents (cont)

    In t!eIn t!e knoledge-)ased approac!,knoledge-)ased approac!,  T!e agent is supplied it! knoledge a)out t!eT!e agent is supplied it! knoledge a)out t!e

    application and user application and user 

     At run-time, agent uses t!e knoledge to At run-time, agent uses t!e knoledge torecognise users plans and find opportunitiesrecognise users plans and find opportunitiesto contri)ute to t!emto contri)ute to t!em

    "*ample of knoledge-)ased agent& t!e UC"go -"*ample of knoledge-)ased agent& t!e UC"go -

    designed to !elp users sol%e pro)lems in usingdesigned to !elp users sol%e pro)lems in usingt!e U4IL operating systemBt!e U4IL operating systemB

  • 8/18/2019 Topic 8 Intelligent Agents

    25/38

    ICT619 2#

    "uilding intelligent agents (cont)"uilding intelligent agents (cont)

    =ro)lems it! knoledge-)ased approac! -=ro)lems it! knoledge-)ased approac! - ot! competence and trust are issues of concernot! competence and trust are issues of concern

    T!e pro)lem of competence relates to t!e competenceT!e pro)lem of competence relates to t!e competence

    of t!e knoledge engineer of t!e knoledge engineer 

    Fnoledge-)ase is fi*ed and cannot )e customised toFnoledge-)ase is fi*ed and cannot )e customised tospecific user needsspecific user needs

    Users trust is affected as agent is programmed )yUsers trust is affected as agent is programmed )ysomeone elsesomeone else

  • 8/18/2019 Topic 8 Intelligent Agents

    26/38

    ICT619 26

    "uilding agents # the machine

    uilding agents # the machine

    learning a!!roach

    earning a!!roach

    .etap!or of a personal office assistant.etap!or of a personal office assistant

     Agents start it! minimum knoledge and learn Agents start it! minimum knoledge and learnfrom&from&

    1B1B ;)ser%ation and imitation of user ;)ser%ation and imitation of user 2B2B User feed)ack 8 direct, indirectUser feed)ack 8 direct, indirect

    3B3B Training )y user Training )y user 

    /B/B ;t!er agents;t!er agents

    User can )uild up model of agent decision making 8User can )uild up model of agent decision making 8more trustmore trust

     Agent capa)le of e*planation Agent capa)le of e*planation

  • 8/18/2019 Topic 8 Intelligent Agents

    27/38

    ICT619 2:

    $e%elo!ment of an agent through

    e%elo!ment of an agent through

    learning

    earning

  • 8/18/2019 Topic 8 Intelligent Agents

    28/38

    ICT619 2?

    "uilding agents # the machine

    uilding agents # the machine

    learning a!!roach

    earning a!!roach

     Ad%antages& Ad%antages& @ess ork from end-user and de%eloper @ess ork from end-user and de%eloper   Agent customises to userorganisation Agent customises to userorganisation

    !a)itspreferences!a)itspreferences 7elps distri)ute kno-!o and competence7elps distri)ute kno-!o and competence

    among different usersamong different users

    'ome e*amples&'ome e*amples&

     Agent for e-mail !andling Agent for e-mail !andling  Agent for meeting sc!eduling Agent for meeting sc!eduling  Agent for electronic nes filtering Agent for electronic nes filtering  Agent for recommending )ooks, music Agent for recommending )ooks, music

  • 8/18/2019 Topic 8 Intelligent Agents

    29/38

    ICT619 29

    Intelligent agents in &'commerceIntelligent agents in &'commerce

    5apid grot! continues in e-commerce5apid grot! continues in e-commerce Information a)out products and %endors is easilyInformation a)out products and %endors is easily

    accessi)leaccessi)le ut transactions are still mostly not automatedut transactions are still mostly not automated

    'i* fundamental stages of t!e )uying process&'i* fundamental stages of t!e )uying process& 4eed identification4eed identification =roduct )rokering=roduct )rokering .erc!ant )rokering.erc!ant )rokering

    4egotiation4egotiation =urc!ase and deli%ery=urc!ase and deli%ery =roduct ser%ice and e%aluation=roduct ser%ice and e%aluation

  • 8/18/2019 Topic 8 Intelligent Agents

    30/38

    ICT619 3E

    Intelligent agents in &'CommerceIntelligent agents in &'Commerce

    (cont’d)(cont’d)

    In t!e need-identification stage, agents can !elp inIn t!e need-identification stage, agents can !elp inpurc!ases t!at are repetiti%e or predicta)lepurc!ases t!at are repetiti%e or predicta)le

    Continuously running agents can monitor a set ofContinuously running agents can monitor a set of

    sensors or data streams and take actions !en certainsensors or data streams and take actions !en certainpre-specified conditions applypre-specified conditions apply

     Agents can use rule-)ased systems or data mining Agents can use rule-)ased systems or data miningtec!niues to disco%er patterns in customer )e!a%iourtec!niues to disco%er patterns in customer )e!a%iourto !elp customers find productsto !elp customers find products

  • 8/18/2019 Topic 8 Intelligent Agents

    31/38

    ICT619 31

    Intelligent agents in &'commerceIntelligent agents in &'commerce

    (cont)(cont)

    In t!e merc!ant )rokering stage, on-lineIn t!e merc!ant )rokering stage, on-lines!opping agents can look up prices for a c!osens!opping agents can look up prices for a c!osenproduct for a num)er of merc!antsproduct for a num)er of merc!ants

    .any )usiness-to-)usiness transactions are.any )usiness-to-)usiness transactions arecan%assedcan%assed

    In a e) auction, customers are reuired toIn a e) auction, customers are reuired tomanage t!eir on negotiation strategiesmanage t!eir on negotiation strategies

    Intelligent agents can !elp it! t!isIntelligent agents can !elp it! t!is

  • 8/18/2019 Topic 8 Intelligent Agents

    32/38

    ICT619 32

    &am!les of on'line sho!!ing

    am!les of on'line sho!!ing

    frameor* ith agent mediation

    rameor* ith agent mediation

     

    ="5';4A="5';4A@ogic@ogic

    JireflyJirefly argainargainJinder Jinder 

     Auction Auction

    otot

    angoango  Auction Auctionotot

    TMTTMT

    4eed4eedidentificationidentification

    =roduct=roduct)rokering)rokering NN NN NN NN

    .erc!ant.erc!ant)rokering)rokering

    NN NN NN

    4egotiation4egotiation NN NN NN

    =ayment O=ayment Odeli%erydeli%ery

    'er%ice O'er%ice O

    "%aluation"%aluation

  • 8/18/2019 Topic 8 Intelligent Agents

    33/38

    ICT619 33

    &am!les of on'line sho!!ing

    am!les of on'line sho!!ing

    frameor* ith agent mediation

    rameor* ith agent mediation

     

  • 8/18/2019 Topic 8 Intelligent Agents

    34/38

    ICT619 3/

    &am!les of on'line sho!!ing

    am!les of on'line sho!!ing

    frameor* ith agent mediation

    rameor* ith agent mediation

     

  • 8/18/2019 Topic 8 Intelligent Agents

    35/38

    ICT619 3#

    &am!les of on'line sho!!ing

    am!les of on'line sho!!ing

    frameor* ith agent mediation

    rameor* ith agent mediation

    (cont’d)

    cont’d)

    'oftare agents are !elping )uyers and sellers cope'oftare agents are !elping )uyers and sellers copeit! information o%erload and e*pedite t!e onlineit! information o%erload and e*pedite t!e online)uying process)uying process

     Agents are creating ne markets (eg, lo-cost Agents are creating ne markets (eg, lo-costconsumer goods+ and reducing transaction costsconsumer goods+ and reducing transaction costs

    Use of agents in e-commerce still at an early stageUse of agents in e-commerce still at an early stage

  • 8/18/2019 Topic 8 Intelligent Agents

    36/38

    ICT619 36

    Intelligent agent design ' state'of'Intelligent agent design ' state'of'

    the'art and futurethe'art and future

    Je agents are a%aila)le it! all t!e desiredJe agents are a%aila)le it! all t!e desiredc!aracteristicsc!aracteristics

     Agent tec!nology still in e*perimental stage Agent tec!nology still in e*perimental stage

     Autonomy and mo)ility already ac!ie%a)le Autonomy and mo)ility already ac!ie%a)le 

    "*ample& a%a applets !ic! e*ecute independently"*ample& a%a applets !ic! e*ecute independently

    across netorksacross netorks

    ut autonomy limited so far in practical use due to t!eut autonomy limited so far in practical use due to t!eagent-unfriendliness of t!e current e) tec!nologyagent-unfriendliness of t!e current e) tec!nology

  • 8/18/2019 Topic 8 Intelligent Agents

    37/38

    ICT619 3:

    Intelligent agent design ' state'of'Intelligent agent design ' state'of'

    the'art and future (cont’d)the'art and future (cont’d)

     A ma>or limiting factor is lack of ontologies A ma>or limiting factor is lack of ontologiesessential for effecti%e communicationessential for effecti%e communication

    uilding and maintaining ontologies remains auilding and maintaining ontologies remains ama>or c!allengema>or c!allenge

    'ome of t!e proposed capa)ilities to )e'ome of t!e proposed capa)ilities to )ede%eloped in future intelligent agents include&de%eloped in future intelligent agents include& @earning as ell as reasoning, !ic! are@earning as ell as reasoning, !ic! are

    c!aracteristics of mac!ine intelligencec!aracteristics of mac!ine intelligence Interacting it! t!e e*ternal en%ironment t!roug!Interacting it! t!e e*ternal en%ironment t!roug!

    sensorssensors

  • 8/18/2019 Topic 8 Intelligent Agents

    38/38

    ICT619 3?

    +&,&+&-C&S

    &,&+&-C&S

    Chin, !,Chin, !, Intelligent Interfaces as Agents! InIntelligent Interfaces as Agents! In Intelligent UserIntelligent User

    InterfacesInterfaces, "! #ulli$an and #! Tyler%eds&, AC' (ress, )e* +or,, "! #ulli$an and #! Tyler%eds&, AC' (ress, )e* +or,..!..! /endler, "!,/endler, "!, Making Sense out of AgentsMaking Sense out of Agents, IEEE Intelligent, IEEE Intelligent

    #ystems, 'arch0April ..., pp!12-13!#ystems, 'arch0April ..., pp!12-13! /endler, "!,/endler, "!, Is There an intelligent Agent in Your Future?Is There an intelligent Agent in Your Future?

    http//www.nature.com/nature/webmatters/agents/agents.html http//www.nature.com/nature/webmatters/agents/agents.html 

    'aes, (!,'aes, (!, Agents that e!uce "ork an! Information #$erloa!  Agents that e!uce "ork an! Information #$erloa! ,,Communications of the AC', 4olume 13 , Issue 3 %"uly ..5&,Communications of the AC', 4olume 13 , Issue 3 %"uly ..5&,pp!pp! 16-56!16-56!

    'aes, (!,'aes, (!, Agents that %u& an! Sell  Agents that %u& an! Sell , Communications of the, Communications of theAC', 4olume 52 , Issue 1 %'arch ...&,AC', 4olume 52 , Issue 1 %'arch ...&, pp!pp! 8-.!8-.!

    #heth, B! and 'aes, (! E$ol$ing Agents for (ersonali7ed#heth, B! and 'aes, (! E$ol$ing Agents for (ersonali7ed

    Information iltering! In (roceedings of the )inth Conf! onInformation iltering! In (roceedings of the )inth Conf! onArtificial Intelligence for Applications! IEEE Computer #ocietyArtificial Intelligence for Applications! IEEE Computer #ociety(ress, ..1(ress, ..1

    9'BC Agent )e*s -9'BC Agent )e*s - http:00http:00agents!umc!edu0agentne*s0currentagents!umc!edu0agentne*s0current 0 0

    http:00***!agentland!com0http:00***!agentland!com0

    http://pattie.www.media.mit.edu/people/pattie/CACM-94http://pattie.www.media.mit.edu/people/pattie/CACM-94http://pattie.www.media.mit.edu/people/pattie/CACM-94http://pattie.www.media.mit.edu/people/pattie/CACM-94http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://agents.umbc.edu/agentnews/current/http://pattie.www.media.mit.edu/people/pattie/CACM-94http://pattie.www.media.mit.edu/people/pattie/CACM-94