о뷢չ
JOURNAL OF COMPUTER RESEARCH
AND DEVELOPMENT
1999 36 12 Vol.36 No.12 1999



Agent֯ɿIFVO
ߡáֶ
ժҪ֯Ϊ֯ˣЭͬҪʽҪ󼯳ɻϢ֧.һAgent֯ɿIFVOֵΪҪֶδ3棺ϵṹʽAgentЭ廥ּ֧ɻ֯ϢĿЭ.IFVO֯Ϣװ͹ΪɿϺͲжǶAgentӶЧ֧֯Ķ̬Ӫ̵Э̵Э.
ؼʡ֯ɿܣϵṹʽAgentЭ廥
ͼšTP311.5
IFVO:AN AGENT BASED INTEGRATIVE FRAMEWORK FOR
VIRTUAL ORGANIZATIONS
GAO Ji and LIN Dong-Hao
(Department of Computer Science & Technology,
Zhejiang University, Hangzhou 3100 27)
AbstractVirtual organizations are becoming an important form for cooperation among regul ar organizations (or persons), which needs the support of integrative informatio n infrastructure. In this paper, an agent based integrative framework for virtua l organizations, called IFVO is introduced, which, using joint intentions as the main means, supports the development and coordinated running of integrative info rmation infrastructure for virtual organizations in three aspects: architecture style, agent cooperation, and semantic interaction. IFVO allows the infrastructu re of a virtual organization to be packed and constructed into the nested agent federation which can be composed and decomposed quickly, in order to support ef fectively the dynamic creation of virtual organizations, the reengineering of bu siness processes, and the coordination of cooperation processes.
Key wordsvirtual organizationintegrative frameworkarchitec ture styleagent cooperation, semantic interaction
1
ɼֵ֧ķֲ㻷£֯ܽʱϷֲԴĳ֯ˣЧ߽ǿսԻĿӦҲظıԭеĹӶǵĹ㷺[12].ҵҵͿְ⹤ѳΪ21һҵĺĸ.
֯ҪԽϵʹۿٽ֧ԱЭͬϢ輯ɼṩ֧֣֪ʶ[3]Ӫ˹;Ľϸ߳󼶣ͨϳԱϢĸ߲ϵṹٽ֯ϢڽԱϢĲͰװΪȵڹ߲ϵṹṩЧЭֶΣʹЭͬܽһµؽУ֧廥ʵЭʱ.
ĿǰɿܵоҪڻ/߷ʽķֲϣϢĲ㣬ʳ̶ȽϵͣټıͲṩЭЭֶκͲ֧廥⣬Ч֧֯ϢĿٽеЭһ[45].
AgentԡԺӦԼAgentЭ֯ԣʹAgentǱṩ֧[67].ĿǰAgentо֪ۺӦùܵʵ֣AgentЭĹϵάAgent״̬BDIУûAgentЭеĵλãҲûдѵĽǶ͸AgentЭɻӦϵͳϵṹв֧֯ϢĽ.
Agent֯ɿIFVO(integrative framework for virtual or ganization)ֵΪҪֶ͸AgentЭͬAgentΪİݣplug-inϵṹ֧AgentЭһάֺ廥Ӷ֯Ϣɼ.
ĽȥȲɿIFVOȻ֧IFVOԪ.
2ɿIFVO
ǰ֯ϢΪֲ㻷Ӧ趯̬ļɻӦϵͳ³IAS.ΪɿܣIFVOȰAgentΪִʵĴȻAgent֪ȥڼļɺAgentЭе.
IFVOڷֲ֮ϣ3ּ֧ļɺAgentЭɴ֧IASĿٽЭ.
IFVO=ASACSI
(1)ASϵṹʽarchitectural styleָAgentΪİϵṹ
(2)ACAgentЭΪɺЭֲ㻷¼ĻֶΣ
(3)SI廥ԽAgentЭʱֵܳͨͲʧ.
IFVOļɺЭһ£Agent״̬仯ΪAgentͶAgentЭߣԻڻֵΪҪֶ𲽾AgentЭͰϵṹ.
2.1ϵṹʽ
Ϊ֯ϢһIASΪЭAgentɵAgent.԰AgentΪ弴á֪ʶͨŻ֧Agentߣ֪ʶĺμ3ڣ.ÿAgentӵִмͼ.ѻֽΪӻȲ԰ǷִеʵAgentִлٷֽ⣩ļͨɺۺϳṩļ㹦ʵּṩļ.
Ķ幹Agent֯Эͬ.νΪҪAgentеֻȫӻ伯йķֵӶΪ֯άЭͬŦ.IASͨһ鶥ṩЩֲڶAgentɻ¼仯ûȣ״̬¼Ϣȣηֽ⣨ͨ򣩣ʹЭͬĹԷֲڶAgentγɷֲļйĹ.
ۣAgentԶΪԪ飺
intentionsactivities-1activities-2eventsassigningactivatinggener ating
assigningintentions   activities-2
activatingactivities-1intentions
generatingactivities-1activities-2   events
Agent֪ʶͨDistributorɶΪԪ飺
intentionsactivities-1activities-2eventsbindinginforming
bindingactivities-2  intentions
informingevents   activities-1
Уactivities-1activities-2ֱָʾintentionsɵĺ¼eventsļ.AssigningactivatinggeneratingbindinginformingָʾӳϵţƫȫϵPݼZԵԼ.
ЭĽ̶ܳȣǿ԰AgentЭͬΪ3룺
(1)AgentһAgentΪЭͬĹߣΪmaȨЭ̾ͷɼԱAgent.maжļҪµļ.
(2)AgentЭ顪maѰˣAgentЭȾЭ˵ܷͬɼma˹Э.
(3)żЭAgentAgentļ¼ʱAgentӦ裨ݷ󣩻Agentļṩ.
AgentɶΪԪ飺
mamembersacquaintancedistributors
һmaɳԱˣӦ֧ͨ.maйڵЭͬԱ˼以ͨ.AgentſɶΪAgentmaЭ飩һAgentЭ顢żЭֵ֧һAgentżЭdistributorsɵԪ飺
agent_federationssingle_agentother_distributors
distributors֪ʶͨŹ淶֧Agent.
ϵṹʽAgentAgentִл򣬹ɶIASϵṹ͸.ΪŦIASϵṹɰͬ𲽾
(1)ֽ⡪AgentͨһֽΪӻеӻAgentֿͨӦӻֽΪӻӶؿ̻ֽṹ.
(2)ֵAgentmaйӻķɣеӻAgentֿӦйӻķɣӶؿ̻ֵṹ.
IASϵṹ͸ÿζ漰²ӻʵϸڣעڼķֽͷֵ.ͼ1ͼ2ֱʾ˿ܵǶAgentͻĲνṹ.AgentAgentִͬһ»жַֽⷽʽ.£ϲAgent֯²Agent֯Ա.ӶͨԱϢϲϵṹΪAgentԿٽ֧֯IAS.

UpAFʾϲAFiʾ²ʵԲֱʾЭ飬
ԲȦʾmaСԲȦʾͨAgent
ͼ1ܵǶAgent

ԲȦʾʾɳеϻmaiɵӻ
ͼ2ܵĻνṹ
ִлɻΪܣͨʵֻṩļ.Ŀǰṩ3ַʽĵȣ̶У״̬-Ӧ鶯̬滮УԼģͶ̬滮.
ν̬滮ָݵǰAgent״̬ͻ״̬ѡִ˳.Լõ񣨼ιԼֽ⣩ͼУЧصȸִ.
IFVOУṩͳһĽӿڣӳ䵽ɷֲֵ֧ṩļ㺯.ӶʹIASϵṹ޷ؽڷֲ.
2.2AgentЭ
AgentЭIFVO֧IASٽЭתĻֲܷ˹ܺͶAgentϵͳоѶAgentЭ˴о̵֧ĽǶ͸ЭϵЭάȴ.
JenningsͨΪϴrecipeԤ滮ȥAgentжĽṹ֯Ӷ͸AgentЭһ[8].ִڼ򵥣˼ķֽӻķɣδԼṩļ˵ʹAgentЭ̺Эһάȱ.
IFVO֧Э̺ЭһάǶȣԼṩļȫĹ˵ͨ򽫼ṩΪڻֵĹ滮Ӷ͸ЭϵЭά.
ĶģBNFʽʾ£
Activity=ơ[Description][Info-Model]Sc heduling
Negotiating Maintaining
Description=  text 
Info-Model= Ϣģ
Scheduling=[Fire-Pattern][PreCondition][ Execute-Input]
[Assignment-Constraint][Execute-Output][PostCondition]
Negotiating=[Solicit-Input][Solicit-Output]BidTime][PostBidProcessing ]
Maintaining=[ExpireTime][RealTime][Exce ption-Processing]
[End-Processing]
ģṩĹ˵Ϊ3棺
(1)ȡΪɡִкͽṩݣģʽȾAgentͻִлԼִнӦĺ.
(2)Э̡ΪAgentma˵ЭṩݣͶʱͶϢͶֹʱбߴ.һб꣬ͱΪʱԱʽԱһmaĵ.
(3)ά֡ΪmaάֻִеһṩݣɵĽֹʱ䡢ִʵʱ䡢쳣.ǰֳ֧ʱ洦쳣ִ쳣͵쳣ִнΪĺ׼ڱҪʱִнΪ״̬¼ʵAgent.
ĶΪma滮Ⱥάֻִṩ˿ɿ.ִе״̬תΪɱʾΪԪ飺
activity-statesconditionsa-operatorsinitial-statetransitions
initial-stateactivity-states
transitionsactivity-statesconditions a-operators activity-states
ͨվ̬ƶĵȼƻҲ֮ΪRecipe滮͵Ȼķֵ.RecipeԤ趨˿ܵļƻƻͼƻаӻBNFʽʾ£
Scheduling-Plan={Plan-Step|LoopScheduling-Plan}+
Plan-Step=Return|Activity-Set[Condition]|or{Activity-Set[Condition]}+
Activity-Set=Activity|And{Activity}+
ʵʱ״̬תͼ3ʾ״̬תΪɱʾΪԪ飺

ͼ3ʵʱ״̬תͼ
intentions-statesconditionsi-operatorsinitial-statetransition
initial-stateintentions-states
transitionsintentions-statesconditions
i-operatorsintentions-states
Уĳʼ״̬Ϊֹÿ״̬õĲһ״̬.򼤻maRecipeĹ滮ȷͨЭ̣Щ˳еӻԼ˳еͬһӻʱȼ.Recipeִкͻ״̬ƻִвӻѡãӻAgentʵĳԱ.
Э̺ЭһάһֱAgentЭٵ.ЭЭϸģṩõԭԽЭ̺άЭϸڽضͨ.ͨرЭ̺ά֪ʶṩͨЭϸģ͵ӦЧֶΣ
(1)IFVOáб-ͶꡱЭЭ飬ÿʹòͬбͶֹʱбߴ
(2)IFVOáϸģάЭһԣÿе֪ʶʹͨõһάӻһԵضά޷
(3)Э̺ά֪ʶAgentmaȥ滮͵Ȼ͸ȣAgentЭĿԺͿά޸ģ.
֯Ӫ˷ʽԱڱֶԵͬʱЭĿ꣬Ϣĸ߲ͨʵAgentЭʮֺ.ΪԱ˵AgentֻѭӻĶṩmaĶӻЭй滮͵ȣ˽ʵֵϸ.
2.3廥
ֲ㻷廥ϰͨݵʧ.ԭڸеıۡ󡢹ϵԣĶشʽһ[910].Ϊˣ۳ΪʧĹؼĸӴֻݹۼĲշͨ䴦.
ǵĸԣʵı۲ܣʹĳӦҲ.磬ݹ淶STEPΪһֹۣƶķ˴ʱ䡢Բȫ.ӦǵĹ۵Ҳı.
IFVOУķֵЭ̡ɡصȣͨҪ֣ڻֵΪʵ廥ṩ˽Ϻõ֧֣
򽫹۵ĶȵӻΧڣͼ4

ͼ4ȵӻѭͬһۣ
(a1)ѭһ
ڼ۵ĶִиûAgent֮䣨ͼ5

ͼ5ķɺͳе߹
ۼӻѭͬĹ.
ڹ۵ĶСΧڣҸӻʱѭͬıۣƹ۵ĹԺͿά.⣬շ״̬¼AgentӦӦۣΪƹ۵һ.
IFVOṩʾCRLEconcept representation language for exchangeȥ干.ÿAgentִмʱշ¼οļϣΪЭAgentͨݵĹͬ׼.ǶࡢϵͺΪBNFʽʾ£
Conceptơ
[Super:{}+];//3֣ࡢϵ
{:{ֵݡ,}*;}*
[Constraint{ֵʽ}+];
End[ơ]
ͨȷضĽṹ壨۶棩ʾʽֵṩʽʾԼ۵ȡֵԼƣ۵AgentȷؼбϢԼ¼.
ʱԤͺμͼ3ΪͬۼĲ𣨲μͼ4ṩ䴦ĳֶΣ䴦ʾΪ
{ʽ||页}*
ѡԴ.
ĿǰֲAgentֱ䴦scriptԺAgent״̬ͬʽʾ䴦ǿԺͿά.
ִлAgentķʽԻѭıۺеı۲䴦Լģһ䴦. 
3IFVOĿ
IFVOĿһԪ۽AgentAgentͼܣΪȥIAS߲ϵṹЩİϵṹ.ֳ֧Ŀ֯ԱϢеӦƵɽ֧֯ӪIAS.ĿǰÿDCOM/OLEƽ̨֮ϣһ׼CORBA淶ҲɽAgent.
Ԫֱ֣֧IASĿУͼ6ʾ.в֣Ϊģͼ֧AgentԱУͨŻ֧AgentAgent廥.ÿmaΪݻģͼģͼ򡢼ܺԱ˼Э顢滮ִкͼִ.ͨŻƷ3ΣCRLE͹ۣڶͨݣϢʾԣЭװͨݣͨŴٽAgent֧Agent֪ʶ廥.֪ʶָÿAgentɾϢͣӲAgentӵеϢ֪ʶʵʲôʽʾʹ.Agentӵе֪ʶϹɹ֪ʶ.

aa,safaֱʾ桢ͨŴٽAgent
ͼ6IFVOԪ
ڿ֣ṩЭͺϳCCLcoordination and composition languageȥAgentЭIASϵṹṩCRLEȥ干.һʽӻڸIASĿߣ֧ģͿͱۿĹά.
IFVOڿʵûIAS׿շ豸ıϵͳ͹·ϵͳ.ƪͲٲ.
4ᡡ
AgentļɿIFVO3棺ϵṹʽAgentЭ廥֧֯ϢIASĿٽ.
ͨضڻֵIFVO˴ͳAgentAgentAgentЭ⣬ʹ͸ؿ̻AgentЭAgentΪİϵṹΪ.Ĺ²ηֽ⣬ʹIASĸ߲ϵṹɰֽͻֵ𲽾.ͨԼķֵ滮͵ȣԼͨڱۺ䴦ʵ廥֯ԱЭͬAgentΪн֪ʶһµؽ.
IFVOֱ֧֯ģͽͻѡԼӪ̵ƺŻּΪѳｨ֯ٽ֧ЧӪIAS.ͨԼAgentװԱҵӪҵЭ֯ϢIASΪɿϺͲжǶAgentӶȷԱҵıӪܵǰЧ֧֯Ķ̬Ӫ̵Э̵Э.
оƻAgentļۺԪ.
õȻѧĿĿ6958300769773019.
߼飺߼ãУ19462ڣʿʦҪоΪ˹ܡ
̡ϵͳɺCIMS.
ֶУ19737ʿоҪоΪ̡CIMS.
ߵλ㽭ѧ˹оݡ310027
ο
1O'Leasy D E et al. Artificial intelligence and virtual org anization. Communication of the ACM, 1991, 40(1): 5259
2Goldman S et al. Agile Competitors and Virtual Organization. New Y ork: Van Nostrand Reinhold, 1995
3Newell A. The knowledge level. Artificial Intelligence, 1982, 18(1): 8 2127
4Nierstrasg O et al. Directions in software composition. ACM Compu ting Surveys, 1995, 27(2): 262264
5Garlan D. Research direction in software architecture. ACM Computing Surveys, 1995, 27(2): 257261
6Wooldridge et al. Intelligent agent: Theory and practice. The Kno wledge Engineering Review, 1995, 110(2): 115152
7Gek Woo Tan et al. An intelligent-agent framework for concurrent product design and planning. IEEE Trans on Engineering Management, 1996, 43(3): 297306
8Jennings N R. Controlling cooperative problem solving in industrial mu lti-agent systems using joint intentions. Artificial Intelligence, 1995, 75(2): 195240
9߼. ڱʾ۵ϵͳ. о뷢չ, 1996, 33(11): 8 01807
(Gao Ji. Development of intelligent systems based on representation ont ologies. Computer Research & Development(in Chinese), 1996, 33(11): 801807)
10Mike Vschold et al. Ontologies: Principles methods and application s. The Knowledge Engineering Review, 1996, 11(2): 93136
ԭյڣ1998-12-09޸ĸյڣ1999-09-21.
