Sunteți pe pagina 1din 19

Instrumentele de dezvoltare a sistemelor multiagent reprezint concepte relativ noi in inteligenta artificiala relativ la sistemele distribuite.

Scopul este de a nelege paradigma instrumentelor de dezvoltare a agenilor, studiind metodologia utilizat n modele de design a multiagenilor i aplicaiile pe care acetia le dezvolta. Este realizat astfel un studiu comparativ, care include instrumente de dezvoltare a agenilor. Toolkit urile cu ageni mobili !"# sunt aproape sisteme soft$are de sine stttoare, incapabile s comunice cu altele asemenea lor i de multe ori fiind doar prototipuri dezvoltate doar pentru rezolvarea unei anumite probleme de cercetare. E%ist astfel concepte diferite n ceea ce privete noiunile de baz ale agenilor mobili& ' ce ar trebui s fac un agent mobil din punct de vedere al programatorului( ' un obiect de un anumit tip definete mai multe funcii de baz pentru agenii mobili cum ar fi comunicarea i migraia, sau orice obiect serializabil( ' ce nivel de comunicaie este necesar& unul simplu ntre agenii care aparin aceleai agenii sau unul comple% care permite i comunicarea la distan( ' ce nivel de securitate este necesar& unul care prote)eaz *ost urile mpotriva unor ageni maliioi, sau este necesar a se prote)a i agenii mpotriva altor agenii maliioase( ' ce fel de mobilitate este necesar( +neori c*iar i cea mai general idee de cercetare nu poate fi adaptat pentru alt toolkit de ageni mobili din cauza diferenelor dintre conceptele de baz. E%cept,nd dou toolkit uri -.glets i /rass*opper0 care suport protocolul de migraie 1.SI2 propus ca standard 31/ n 4""5, este practic imposibil s faci alte dou toolkit uri interoperabile. n contrast cu implementarea unui mare numr de prototipuri, puine dintre acestea au fost dezvoltate ca toolkit uri mobile care ar putea fi folosite n aplicaii de nivel industrial. .cestea sunt dezvoltate ca sisteme monolitice, cu un mare numr de funcii i sunt greu de configurat i folosit. E aproape imposibil s modifici i s e%tinzi astfel de sisteme pentru a le utiliza cu succes n orice tip de scenariu de aplicaie soft$are. 6entru a remedia aceasta situaie, una dintre cele mai importante

provocri a fost ca prin proiectul Trac7 s se dezvolte un nou model de ar*itectur ca referine pentru alte toolkit uri mobile. .cest capitol realizeaz o revizuire a fiecrui agent, apoi efectueaz o comparaie pe mai multe niveluri de abstractizare, fc,nd o evaluare dintr o comparaie colectiva a instrumentelor de dezvoltare a agenilor, precum i a eficientei acestora. .u fost alei cinci ageni care vor fi investigai pentru a nelege diversitatea si dinamica sistemelor multiagent. 6entru a nelege ar*itectura agenilor sunt folosite urmtoarele caracteristici ale agenilor& autonomie, auto nvare, comunicare, cooperare, mobilitate i scop, caracteristici ce au fost incluse at,t in designul ar*itectural, precum i in metodologia instrumentelor. 8oiunea de agenie este o e%tensie a unei noiuni mai slabe, i susine proprieti suplimentare pe care agentul le incorporeaz. Te*nologia agentul a fost subiect de dezbatere, nelegerea diferitelor abordri din spatele fiecrui ar*itecturi de agent este fundamental, ns acest lucru a fost de asemenea un subiect de opinie cu privire la cea mai bun abordare. 6entru a putea face o comparaie eficace i care s evalueze instrumentele individuale folosind instrumente de dezvoltare a multiagentilor pentru a compara performana lor, va fi folosit o nelegere la nivel nalt a unei metodologii specifice de dezvoltare a agenilor orientai. Se va realiza un studiu amnunit al ar*itecturii agenilor. n cadrul acestui studiu comparativ vor fi analizate urmtoarele utilitare folosite pentru dezvoltarea sistemelor multiagent& ' 9.:; -9ava Intelligent .gent 2rame$ork0 ' 9.<E -9ava .gent <evelopment 2rame$ork0 ' 9.S38 -9ava +sing .gent Speak0 ' .gentTool ' :ougaar -:ognitive .gent .rc*itecture0 <efiniia unui agent difer datorita diversitii si funcionalitii acestora. :ei mai muli cercettori dau definiii pentru ageni slabi si ageni tari. 8oiunea de slab definete termenul de agent autonom, sociabil i reactiv -!4=#, !=>#0, n timp ce noiunea de tare este mult mai descriptiv, un agent este n general privit ca un sistem care pe l,ng proprietile unui agent slab este conceput sau implementat folosind concepte care folosesc noiuni cognitive, precum cunotinele, inteniile, obligaiile i emoiile.

1uli cercettori au considerat ca te*nologia agentului este un rezultat de convergen a mai multe te*nologii din domeniul tiinei calculatoarelor, precum programarea orientat obiect, programarea distribuit i inteligena artificial. +n aspect important al agentului este abilitatea de a oferi inteligen prin interaciune, care sugereaz unirea a dou flu%uri de cercetare diferite precum inteligena artificial i interaciunea om calculator. Inteligena artificial n mod tradiional presupune inteligena pur cu puine interaciuni e%terne. .ceast abordare produce bloca)e n special atunci c,nd problemele apar n sisteme dezvoltate prin te*nici din inteligena artificial. .pariia sistemelor .? i nevoia de interaciune au condus la un nou domeniu numit inteligena artificial distribuit. .genii raionali e%tind bazele de date inteligente la cadre care permit descrieri bazate pe logic i specificaii ale comunicrii dintre ageni. 6entru a nelege standardele pentru agenii, 2I6. dezvoltat o specificaie ar*itectural care a acoperit limba de comunicare a agenilor, structura mesa)elor, transportul mesa)elor. 2I6. este implementat cu a)utorul e%tensiilor 9ack i 9.<E n ceea ce privete interoperabilitatea ntre ageni, stabilirea comunicrii ntre ageni. .gentTool, :ougaar i 9ason a fost pus n aplicare interaciunea lor folosind protocoale diferite dezvoltate folosind abordri individuale, care are un alt set de abordri prevzute de caietul de sarcini standard de 2I6.. :omunicarea, utiliz,nd metodologia primara, nu este e%clus, si utilizeaz protocoale diferite de interaciune i de structur n cadrul ar*itecturilor individuale. n conformitate cu 9.<E ar*itectura standard a fost pe deplin compatibila cu standardul stabilit de ndeplinire a tuturor specificaiilor, prin spri)inul oferit pe toate nivelurile de ontologie. 9ack, de asemenea, respect n totalitate caietul de sarcini, prin spri)inul oferit de e%tindere, ca parte a serviciilor sale pentru ali ageni. .lte instrumente de creare a agenilor, 9ason i :ougaar, utilizeaz o metodologie diferit n ceea ce privete modul n care acestea comunica. 9.<E i 9ack respect standardul prin paradigma agent, pledeaz pentru autonomie n utilizarea mediilor sale, permi,nd interaciunea social cu alte ar*itecturi i utilizeaz cea mai mare a ontologiei 2I6. n procesarea pro activ. 6rocesul de comunicare permite agenilor s i ndeplineasc sarcinile prin utilizarea proprietii de interoperabilitate. :ooperarea permite desfurarea altor ageni in cadrul reelei.

:omunicarea permite mobilitatea, prin intermediul ontologiei, la diferite niveluri, pentru atingerea scopurilor.

9.<E -9ava .gent <evelopment 2rame$ork0 !5# a fost deplin codat n 9ava ca un agent de programare pentru a e%ploata cadrul de lucru. 9.<E este scris n limba)ul de programare 9ava i este format din mai multe pac*ete 9ava oferind programatorilor de aplicaii at,t piese funcionale gata fcute c,t i interfaa abstract pentru aplicaii ce depind de task uri. 9ava a fost limba)ul de programare ales datorit caracteristicilor sale atractive, n special orientate spre programarea orientat pe obiecte n medii distribuite eterogene. .r*itectura 9.<E este aceea a unei ar*itecturi principale care controleaz celelalte containere i care este distribuit n cadrul reelei. @iaa unui agent 9.<E ncepe cu elementul principal care distribuie servicii pentru *ostarea i e%ecuia agentului. .r*itectura principal 9.<E este punctul de pornire pe platform i permite managementul elementelor rmase astfel nc,t acestea s comunice cu ar*itectura principal, aceste containere au o legtura cu ar*itectura principal. Elementul duce responsabilitatea obiectelor referite i transport adresele tuturor containerelor noduri din care este compus ar*itectura principal. Informaiile pe care le dob,ndesc containerele sunt adrese de ageni, nume, etc. .ceste informaii sunt pstrate apoi n containerul principal, i atunci c,nd este depistata o identitate 9.<E se autorizeaz cone%iunea. :ontainerul principal 9.<E este entitatea care deine toate informaiile despre agent. :u toate acestea, dac n timpul proces de conectare cone%iunea returneaz eroare, ar*itectura principal urmeaz diferite etape pentru a obine referine adecvate

4. 1. A. .ndersson i T. Sand*olm, Contract types for satisfcing task allocation: Ii experimental results, ...I Spring S7mposium& Satiscing 1odels, 4""5. =. S. .knine, S. 6inson i 1. 2. S*akun, An extended multi-agent negotiation protocol, .utonomous .gents and 1ulti .gent S7stems, voi. 5-40, B CB, =DDC. E. 6. Fraun i G. Aossak, Mobile Agents: Concepts, Mobility Models, & the racy oolkit, Elsevier Inc., +S., =DDB. C. A. .. Frooks, A robust layered control systemfor a mobile robot, IEEE 9ournal of Aobotics and .utomation, A. =-40, 4C =E, 4"5H.

B. A. .. Frooks, Challenges for Complete Creature Architectures , 2irst International :onference on Simulation of .daptive Fe*avior, 6aris, 2rance, CEC CCE, 4""D. H. :. I. Frooks i E. I. <urfee, Congregation formation in multiagent systems, 9ournal of .utonomous .gents and 1ulti agent S7stems, >-4 =0, 4CB 4>D,=DDE. >. ;. Finmore i 8. @ulkan, .ppl7ing game t*eor7 to automated negotiation, 8etnomics, voi. 4-40, 4 ", 4""". 5. 2. ?. Fellifemine, /. :aire i <. /reen$ood, !e"eloping Multi-Agent #ystems $ith %A!&, Gile7, =DD>. ". A. I. Fordini, 1. <astani, 9. <i% i .. El 2alla* Seg*rouc*ni, MultiAgent 'rogramming: (anguages, ools and Applications, Springer <ordrec*t Ieidelberg ?ondon 8e$ Jork, =DD". 4D. @. :onitzer i T. Sand*olm, A)&#*M&: A general multiagent learning algorithm that con"erges in self-play and learns a best response against

stationary opponents, 6roceedings of t*e T$entiet* International :onference on 1ac*ine ?earning, =DDE. 44. @. :onitzer i T. Sand*olm, A )&#*M&: A general multiagent learning algorithm that con"erges in self-play and learns a best response against stationary opponents, 1ac*ine ?earning, =DDH. 4=. A. :onte i :. :astelfranc*i, he treatment of norms in modelling ra+ional agents: cogniti"e issues, 6roceedings of 1odel.ge "B, /eneral 1eeting of Esprit Gorking /roup, I8AI., 2rance, 4""B. 4E. 8. :onstantinescu i :. I. 6oprlan, Authentication model based on Multi-Agent #ystem, .nnals of t*e +niversit7 of :raiova 1at*ematics and :omputer Science Series, Editura +niversitaria, :raiova, Aom,nia, voi. E5, nr. =, Iunie, =D44, B" H5. 4C. 9. 6. <ela*a7e, #e,uence ransformations, Springer @erlag, 4"55. 4B. A. <avis i A. /. Smit*, -egotiation as a metaphor for distributed problem sol"ing, .rticial Intelligence, voi. =D, HE 4D", 4"5E. 4H. +. Endriss, 8. 1audet, 2. Sadri i 2. Toni, -egotiating socially optimal allocations ofresources, 9ournal of .rti cial Intelligence Aesearc*, voi. =B,E4B EC5,=DDH. 4>. 6. van Eck, A Compositional #emantic #tructure for Multi-Agent #ystems !ynamics, =DD4. 45. S. 2ranklin i .. /raesser, s it an agent, or .ust a program/: A taxonomy for autonomous agents, 6roceedings of t*e Gorks*op on Intelligent .gents III, .gent T*eories, .rc*itectures, and ?anguages, =4 EB, 4""H. 4". <. 2udenberg i <. ;. ?evine, he heory of (earning in 0ames, 1IT 6ress, 4""5. =D. <. 2udenberg i <. ;reps, (ectures on learning and e,uilibrium in strategic-form games, Tec*nical report, :3AE ?ecture Series, 4""D.

44. 6. 2aratin, :. Sierra i 8. A. 9ennings, -egotiation decision functions for autonomous agents, Aobotics and .utonomous S7stems, voi. =C-E C0, 4B" 45=, 4""5. 4=. S. S. 2atima, 1. Gooldridge i 8. 9ennings, *ptimal negotiation of multiple issues in incomplete information settings, 6roceedings of t*e T*ird International 9oint :onference on .utonomous .gents and 1ulti.gent S7stems, 4D5D 4D5", .:1, =DDC. 4E. /. /lass, Mobility: processes, computers, and agents, .:1 6ressK.ddison Gesle7 6ublis*ing :o., 8e$ Jork, +S., 4""". 4C. A. S. /ra7, <. ;otz, /. :7benko i <. Aus, A"ailable at1 d2agents: #ecurity in a multiple-language, mobile-agent system, 1obile .gents and Securit7, Springer @erlag, Ferlin, /erman7, 4""". 4B. 6. 9. /m7trasie$icz i E. I. <urfee, A rigorous, opera+ional formali3ation of recursi"e modeling, 6roceedings of t*e 2irst International :onference on 1ulti .gent S7stems, pages 4=B 4E=, ...IK1IT press, 4""B. 4H. 6. 9. /m7trasie$icz i E. I. <urfee, 4a+ional communication in multiagent systems, .utonomous .gents and 1ulti .gent S7stems 9ournal, C-E0, =EE =>=, =DD4. 4>. 1. A. /eneseret* i S. 6. ;etc*pel, #oft$are agents, :ommunications of t*e .ssociation for :omputing 1ac*iner7, E>->0, 4""C. 45. :. Ie$itt i 9. Inman, !ai bet$ixt and bet$een: 5rom intelligent agents to open systems science, IEEE Transactions on S7stems, 1an and :7bernetics, 4""4. 4". 9. Iu i 1. 6. Gellman, -ash ,-learning for general-sum stochastic games, 9ournal of 1ac*ine ?earning Aesearc*, 4DE" 4DH", =DDE. =D. :. I. Iommes, 5inancial markets as nonlinear adapti"e

e"olutionary systems, Luantitative 2inance 4, 4C" 4H>, =DD4.

44. 9. :. Iarsan7i, Approaches to the bargaining problem before and after the theory ofgames: A criticai discussion of3euthen2s, hicks2, and nash2s theories, Econometrica, voi. =C-=0, 4CC 4B>, 4"HB. 4=. 8. A. 9ennings, 6. 2aratin, .. A. ?omuscio, S. 6arsons, 1. Gooldridge i :. Sierra, Automated negotiation: 'rospects methods and challenges, /roup <ecision and 8egotiation, voi. 4D-=0, 4"" =4B, =DD4. 4E. E. ;alai i 1. Smorodinsk7, *ther solutions to nash2s bargaining problem, Econometrica, voi. CE, B4E B45, 4">B. 4C. S. ;raus, #trategic -egotiation in Multiagent &n"ironments, 1IT 6ress, =DD4. 4B. 1. ;lein, 6. 2aratin, I. Sa7ama i J. Far Jam, -egotiating complex contracts, /roup <ecision and 8egotiation, voi. 4=,444 4=B, =DDE. 4H. 2. ?eon, Agen+i inteligen+i cu capacit6+i cogniti"e, Te*nopress, Iai, =DDH. 4>. 1. ?. ?ittman, 5riend-or-foe ,-learning in general-sum games, 6roceedings of t*e Eig*teent* International :onference on 1ac*ine ?earning, E== E=5, =D44. 45. F. A. ?eaa, 1. :. /uptab i G. F. Ju, A prototype multi-agent &4' systern: an integrated architecture and a conceptual frame$ork, Tec*novation =B, =DDB, CEE CC4. 4". F. ?eFaron, Agent-based computa+ional finance: #uggested readings and early research, 9ournal of Economic <7namics and :ontrol =C, H>" >D=, =DDD. =D. I. ?in, Architectural !esign of Multi-Agent #ystems: echnologies and echni,ues, +niversit7 of Iouston <o$nto$n, +S., =DD>. =4. 6. 1aes, Modeling Adapti"e Autonomous Agents, .rtificial ?ife, .n 3vervie$, edited b7 :*ristop*er /. ?angton, 1IT 6ress, :ambridge, 4""B.

44. T. 1. 1itc*ell, Machine (earning, 1c/ra$ Iill, 4"">. 4=. T. 1agedanz i A. /lit*o, Mobile agent-based net$ork and ser"ice management, 9ournal of 8et$ork and S7stems 1anagement, voi. >, Springer, 8et*erlands, 4""". 4E. .. 1ut*oo, 7argaining heory $ith Applications, :ambridge +niversit7 6ress, 4""". 4C. 9. 2. 8as*, he bargaining problem, Econometrica, voi. 45, 4BB 4H=, 4"BD. 4B. :. :. 8evar, 9. 1c/innis, S. 1odgil, I. Aa*$an, :. Aeed, /. Simari, 1. Sout*, /. @rees$i)k i S. Gillmott, o$ards and argument interchange format, T*e ;no$ledge Engineering Aevie$, voi. =4-C0, ="E E4H, =DDH. 4H. T. <. 8gu7en i 8. 9ennings, Coordinating multiple concurrent negotiations, 6roceedings of t*e T*ird International 9oint :onference on .utonomous .gents and 1ulti.gent S7stems, 4DHC 4D>4, .:1, =DDC. 4>. 1. 3s*ima i <. ?ange, 'rogramming and !eploying %a"a Mobile Agents $ith Aglets, .ddison Gesle7, +S., 4""5. *ttp&KK$$$.trl.ibm.comKagletsK 45. :. I. 6oprlan i ?. Mtefanescu, A Multi-agent Approach for Adapti"e 8irtual *rgani3ation using %A!&, .<.6TI@E .8< I8TE??I/E8T SJSTE1S, ?ecture 8otes in :omputer Science, Springer Ferlin, Ieidelberg =D44, voi. H"CEK=D44, ECC EBB. 4". :. I. 6oprlan, A solution based on intelligent soft$are agents to impro"e the data searching in the contact centers, Ht* IEEE 9oint International Information Tec*nolog7 and .rtificial Intelligence :onference, IEEE IT.I: =D44, :*ongNing, :*ina,vol. =, =D == .ugust =D44, 4 B.

=4. :. I. 6oprlan i 1. <upac, An *ptimal 'ath Algorithm for Autonomous #earching 4obots, .nnals of +niversit7 of :raiova, 1at*ematics and :omputer Science Series, voi. EH-40, E> C5, =DD". ==. :. I. 6oprlan i :. 6oprlan, Mobile Agents communication for kno$ledge representation, llt* Gorld 1ulti :onference on S7stemics, :7bernetics and Informatics, -G1S:I =DD>0, =DD>, 3rlando, 2lorida, +S., "= "H. =E. :. I. 6oprlan i :. 6oprlan, Algorithms for Mobile Agents in -et$ork using racy9Mobile Agent oolkit:, B t* AoEdu8et IEEE International :onference, 4 E 9une =DDH, Sibiu, Aom,nia, EE> EE". =C. :. I. 6oprlan, Collaborati"e soft$are agents based on %A!& for distributed data processing, .nnals of t*e +niversit7 of :raiova 1at*ematics and :omputer Science Series, Editura +niversitaria, :raiova, Aom,nia, voi. E5, nr. E, =D44, CH BB. =B. :. I. 6oprlan, ;no$ledge 'rocessing in Contact Centers using a MultiAgent Architecture, GSE.S TA.8S.:TI38S 38 :316+TEAS, nr. 44, voi. ", 4E45 4E=>,=D4D. =H. :. I. 6oprlan, A Mobile Agents Architecture for 4obots Control, .nnals of +niversit7 of :raiova, 1at*ematics and :omputer Science Series, voi. EH-=0, 4D" 44>,=DD". =>. :. I. 6oprlan, ;no$ledge 7ases Management using Mobile Agents Architecture, 9ournal of ;no$ledge, :ommunications and :omputing Tec*nologies -9;::T0, voi. 4-40, B" H", =DD". =5. :. I. 6oprlan i 1 <pac, A Mobile Agent 8irtual 4eality Modeling of #earching 4obots, .nnals of +niversit7 of :raiova, 1at*ematics and :omputer Science Series, voi. EB, 4CB 4BC, =DD5. =". :. I. 6oprlan i :. .. Mtefanescu, A multi-agent solution for contact centers impro"ement, .nnals of <...1 for =D44 O 6roceedings of

T*e ==nd International <...1 SJ163SI+1, voi. ==, nr. 4, 6ublis*ed b7 <...1 International, @ienna, .ustria, =D44, 445E 445C. HD. :. I. 6oprlan, :. .. Mtefanescu i ?. Mtefanescu, Mul+i-Agent Approach for !ata Analysis in a ;no$ledge based #ystem for Contact Center, 6roceedings of I::SIT =D44 & International :onference on :omputer Science and Information Tec*nolog7, @enice, Ital7, nr. B" Gorld .cadem7 of Science Engineering and Tec*nolog7, =D44, 44=H 44E4. H4. :. I. 6oprlan, A Multi-Agent Approach for !istributed ;no$ledge 'rocessing in Contact Centers, 6roceedings of T*e 4Ct* GSE.S International :onference on :316+TEAS -part of t*e 4Ct* GSE.S :S:: 1ulticonference0, :orfu Island, /reece, ?atest trends on :omputers, voi. 4, =E =B Iulie, =D4D, =4C =4". H=. :. I. 6oprlan i ?. Mtefanescu, A Mobile Agents #ystem for Intelligent !ata Analysis, 6roceedings of GSE.S .pplied :omputing :onference =DD" -.:: =DD"0, September =5 ED, .t*ens, /reece, 1at*ematical 1et*ods and .pplied :omputing, voi. 4, GSE.S 6ress, HHE HH5. HE. :. I. 6oprlan i 1. <upac, A Mobile Agents Approach for #earching 4obots 8irtual 4eality Modeling and #imulation, 6roceedings of t*e =Dt* I.STE< International :onference on 1odelling and Simulation -1S =DD"0, =DD", Fanff, .lberta, :anada. HC. :. I. 6oprlan, A !istributed ;no$ledge-based 'rocessing using

Mobile Agents, Fuletin Mtiinific +niversitatea din 6iteti, Seria 1atematic i Informatic, nr. 4C, 4>" 45H. HB. S. Aussell i 6. 8orvig, Artificial Intelligence: A Modern Approach, 6rentice Iali, 4""B. HH. S. Aao i 1. /eorgeff, Modeling ra+ional agents $ithin a 7!I architecture, 6roceedings of t*e Second International :onference on 6rinciples of ;no$ledge Aepresentation and Aeasoning, C>E C5C 4""4.

HD. S. Aao i 1. /eorgeff, 7!I agents: 5rom theory to practice, 6roceedings of t*e 2irst International :onference on 1ulti .gent S7stems -I:1.S "B0, E4= E4", 4""B. H4. S.9. Aussell i 6. 8orvig, Artificial intelligence: A modern approach -=nd ed.0, 6rentice Iali, =DDE. H=. @. Aot*, Mutual protection of co-operating agents, ?ecture 8otes in :omputer Science, voi. 4HDE, Springer @erlag Inc., 8e$ Jork, +S., =DD4. HE. .. Aubinstein, 'erfect e,uilibrium in a bargaining model, Econometrica, voi. BD-40, "> 44D, 4"5=. HC. 9. S. Aosensc*ein i /. Plotkin, 4ules of &ncounter, T*e 1IT 6ress, :ambridge, 1., 4""C. >=.4. Aa*$an, S. <. Aamc*urn, 8. A. 9ennings, 6. 1cFurne7, S. 6arsons i ?. Sonenberg, Argumentation-based negotiation, T*e ;no$ledge Engineering Aevie$, voi. 45-C0, ECE E>B, =DDC. >E.6. Stone, (ayered(earning in Multiagent #ystems, 1IT 6ress, =DDD. >C. A. S. Sutton i .. /. Farto, 4einforcement (earning: An Introduction, 1IT 6ress, 4""5. >B.9. Suro$iecki, he )isdom ofCro$ds, .nc*or, =DDB. >H. A. /. Smit*, he contract net protocol: <igh-le"el communication and control in a distributed problem sol"er, IEEE Transactions on :omputere, voi. : ="-4=0, 44DC 444E, 4"54. >>. T. Sand*olm, Contract types for satiscing task allocation: I theoretical results, ...I Spring S7mposium& Satiscing 1odels, 4""5. >5. T. Sand*olm, An algorithm for $inner determination in combinatorial auctions, .rticial Intelligence, voi. 4EB-4 =0, 4 BC, =DD=.

>E. T. Sand*olm i @. ?esser, (e"eled-commitment contracting: A backtracking instrument for multiagent systems, .I 1agazine, voi. =E-E0, 5" 4DD, =DD=. >C. T. Sand*olm i J. P*ou, #urplus e,ui"alence of le"eled commitment contracts, .rticial Intelligence, voi. 4C=-=0, =E" =HC, =DD=. >B. /. Stoian i :. I. 6oprlan, A 'roposal for an &nhanced Mobile Agent Architecture 9&MA:, .nnals of +niversit7 of :raiova, 1at*ematics and :omputer Science Series, >4 >", =D4D. >H. /. Stoian i :. I. 6oprlan, =sing Mobile Agents for <andoff uning in Mobile )iMA> -et$orks, .nnals of t*e +niversit7 of :raiova 1at*ematics and :omputer Science Series, voi. E>, no. E, Septembrie, =D4D,== =5. >>. J. S*o*am i ;. ?e7ton Fro$n, M=( IA0&#?# &M#: Algorithmic, 0ame- heoretic, and (ogical 5oundations, :ambridge +niversit7 6ress, =DD". >5. ?. Mtefanescu i :. I. 6oprlan, Intelligent #oft$are Agents for !ata Analysis in ;no$ledge-based #ystems, :*apter for t*e book QIntelligent <ecision Support S7stems for 1anagerial <ecion 1akingQ, .SEAS 6ublis*ing =D44,=B C5. >". 6. Ta7lor i ?. 9onker, &"olutionary stable strategies and game dynamics, 1at*ematical Fiosciences, >H 5E, 4">5. 5D. 8. Rndreanu i :. I. 6oprlan, A Mobile Agents Approach for ;no$ledge 7ases 'rocessing, 6roceedings of t*e T$elft* International :onference on Intelligent S7stems and :ontrol -IS: =DD"0, =DD", => E=, :ambridge, 1assac*usetts, +S.. 54. 9. 1. @idai i E. I. <urfee, (earning nested models in an information economy, 9ournal of E%perimental and T*eoretical .rticial Intelligence,

4D-E0, ="4 ED5, 4""5.

>E. 9. 1. @idai i E. I. <urfee, he mo"ing target function problem in multi-agent learning, 6roceedings of t*e T*ird International :onference on 1ulti .gent S7stems, E4> E=C, ...IK1IT press, 4""5. >C. 9. 1. @idai i E. I. <urfee, 'redicting the expected beha"ior of agents that learn about agents: the C(4I frame$ork, .utonomous .gents and 1ulti .gent S7stems, H-40, >> 4D>, =DDE. >B. 9. 1. @idai, 5undamentals of Multiagent #ystems $ith -et(ogo &xampes, =DD". >H. 1. Gooldridge i 8. A. 9ennings, Intelligent Agents - heories, Architectures, and (anguages, ?ecture 8otes in .rtificial Intelligence, Springer @erlag, voi. 5"D, 4""B. >>. 9. G. Geibull, &"olutionary 0ame heory, 1IT 6ress, 4"">. >5. :. 9. Gatkins i 6. <a7an, @-learning, 1ac*ine ?earning, 5-E C0, =>" ="=, 4""=. >". <. I. Golpert i ;. Tumer, An introduction to collecti"e intelligence, Tec*nical report, 8.S., 4""". 5D. 1. Gooldridge, Intelligent agents, Multiagent #ystems: a Modern Approach to !istributed Artificial Intelligence, 1IT 6ress, :ambridge, 1assac*usetts, => >>, 4""". 54. /. Gei i S. Sen, Adaptation and (earning in Multiagent #ystems, Springer @erlag, Ferlin, /erman7, 4""H. 5=. 1. Gooldridge i 8. A. 9ennings, Agent heories, Architectures, and (anguages: a #ur"ey, Springer @erlag, Ferlin, /erman7, 4""B. 5E. <. Gong, 8. 6aciorek, T. Gals*, 9. <i:elie, 1. Joung i F. 6eet, Concordia: An infrastructure for collaborating mobile agents, 6roceedings of t*e 2irst International Gorks*op on 1obile .gents. Springer @erlag ?ondon, +;, 4"">.

"".<. Giller, -et$ork &xchange heory, 6raeger 6ublis*ers, Gestport, 4""". 4DD. 1. Gooldridge, An introduction to multiagent systems, 9o*n Gile7 O Sons ?T<, +;, =DD=. 4D4. S. P*ang, @. ?esser i S. .bdalla*, Eficient management of multilinked negotiation based on a formalized model. .utonomous .gents and 1ulti.gent S7stems, voi. 4D-=0, =DDB. 4D=. S. P*ang, @. ?esser i A. 6odoroz*n7, Multi-dimensional, multistep negotiation for task allocation in a cooperati"e system, .utonomous .gents and 1ulti .gent S7stems, voi. 4D-40, B CD, =DDB.