Simon D. Parsons - Böcker

This is a collection of papers from international researchers, that offers a broad view of the many ways game theory and decision theory can be applied in agent-based systems, from standard applications of the core elements of the theory to more cutting-edge developments. The range of topics discussed in this book should provide the reader with a comprehensive volume that reflects both the depth and breadth of work in applying techniques from game theory and decision theory to design agent-based systems.Chapters include: selecting partners; evolution of agents with moral sentiments in an IPD exercise; dynamic desires; emotions and personality; decision-theoretic approach to game theory; shopbot economics; finding the best way to join in; shopbots and pricebots in electronic service markets; polynomial time mechanisms; multi-agent Q-learning and regression trees; satisficing equilibria; investigating commitment flexibility in multi-agent contracts; pricing in agent economies using multi-agent Q-learning; using hypergames to increase planned payoff and reduce risk; bilateral negotiation with incomplete and uncertain information; and robust combinatorial auction protocol against false-name bids.

Game Theory And Decision Theory In Agent-Based Systems is a collection of papers from international leading researchers, that offers a broad view of the many ways game theory and decision theory can be applied in agent-based systems, from standard applications of the core elements of the theory to more cutting edge developments. The range of topics discussed in this book provide the reader with the first comprehensive volume that reflects both the depth and breadth of work in applying techniques from game theory and decision theory to design agent-based systems. Chapters include: Selecting Partners; Evolution of Agents with Moral Sentiments in an IPD Exercise; Dynamic Desires; Emotions and Personality; Decision-Theoretic Approach to Game Theory; Shopbot Economics; Finding the Best Way to Join in; Shopbots and Pricebots in Electronic Service Markets; Polynomial Time Mechanisms; Multi-Agent Q-learning and Regression Trees; Satisficing Equilibria; Investigating Commitment Flexibility in Multi-agent Contracts; Pricing in Agent Economies using Multi-agent Q-learning; Using Hypergames to Increase Planned Payoff and Reduce Risk; Bilateral Negotiation with Incomplete and Uncertain Information; Robust Combinatorial Auction Protocol against False-name Bids.

In this work, we provide a treatment of the relationship between two models that have been widely used in the implementation of autonomous agents: the Belief DesireIntention (BDI) model and Markov Decision Processes (MDPs). We start with an informal description of the relationship, identifying the common features of the two approaches and the differences between them. Then we hone our understanding of these differences through an empirical analysis of the performance of both models on the TileWorld testbed. This allows us to show that even though the MDP model displays consistently better behavior than the BDI model for small worlds, this is not the case when the world becomes large and the MDP model cannot be solved exactly. Finally we present a theoretical analysis of the relationship between the two approaches, identifying mappings that allow us to extract a set of intentions from a policy (a solution to an MDP), and to extract a policy from a set of intentions.

This volume is based on the Second Workshop on Argumentation in Multiagent Systems (Argmas). The workshop was held in conjunction with the 4th Int- national Conference on Autonomous Agents and Multiagent Systems (aamas), at the University of Utrecht in Utrecht, The Netherlands, in July 2005. The workshop itself took place on July 26. We are happy to report that the second workshop was just as popular and successfulasitspredecessor,heldtheprevioussummerinNewYork.Wereceived 17 submissions, eachof which wasreviewed by at least three experts in the ?eld, and ten of these papers were accepted for presentation at the workshop. Once again the workshop was graced by an invited lecture, this time by Frans van Eemeren of the University of Amsterdam, who talked on the subject of prag- dialectics. The workshop attracted 31 participants, ensuring many questions for the speakers, and a healthy exchange of views during the discussion periods. Following the practice established with the post-proceedings of the ?rst Argmasworkshop,weinvitedthepresentersofalltheacceptedpaperstoprepare revised versions of their papers for this volume.In addition we approached - thors of papers on directly related topics that had been presented in theaamas conference, and this gaveus an additionalseven papers. We further solicited one additional paper (details below) and were lucky enough that Prof. van Eemeren consented to send us a paper that covered the material of his invited talk.

An introductory review of uncertainty formalisms by the volume editors begins the volume. The first main part of the book introduces some of the general problems dealt with in research. The second part is devoted to case studies; each presentation in this category has a well-delineated application problem and an analyzed solution based on an uncertainty formalism. The final part reports on developments of uncertainty formalisms and supporting technology, such as automated reasoning systems, that are vital to making these formalisms applicable. The book ends with a useful subject index. There is considerable synergy between the papers presented. The representative collection of case studies and associated techniques make the volume a particularly coherent and valuable resource. It will be indispensable reading for researchers and professionals interested in the application of uncertainty formalisms as well as for newcomers to the topic.

Uncertaintyisanincreasinglyimportantresearchtopicinmanyareasofc- puterscience. Manyformalismsarebeingdeveloped,withmuchinterestatthe theoryleveldirectedatdevelopingabetterunderstandingoftheformalismsand identifyingrelationshipsbetweenformalisms,andatthetechnologyleveldirected atdevelopingsoftwaretoolsforformalismsandapplicationsofformalisms. ThemainEuropeanforumforthesubjectistheEuropeanConferenceon SymbolicandQuantitativeApproachestoReasoningandUncertainty(EC- SQARU). FollowingthesuccessofthepreviousECSQARUconferences,heldin Marseilles(1991),Granada(1993),Fribourg(1995),andBonn(1997),the?fth conferenceintheserieswasheldatUniversityCollegeLondoninJuly1999. ThisvolumecontainspapersacceptedforpresentationatECSQARU'99. In additiontothemainconference,twoworkshopswereheld. The?rstwason DecisionTheoreticandGameTheoreticAgents,chairedbySimonParsonsand MikeWooldridge,andthesecondwasonLogicalandUncertaintyModelsfor InformationSystems,chairedbyFabioCrestaniandMouniaLalmas. Selected papersfromtheworkshopsarealsoincludedintheseproceedings. Weareindebitedtotheprogrammmecommitteefortheire?ortinorganising theprogramme,totheinvitedspeakers,andtothepresentersofthetutorials.Furthermore,wegratefullyacknowledgethecontributionofthemanyreferees whowereinvolvedinthereviewingprocess. Finallywewouldliketothankthe DepartmentofComputerScienceatUniversityCollegeLondonforadminist- tivesupport. ProgrammeCommittee TheprogrammecommitteewaschairedbyAnthonyHunter(UniversityCollege London),andcomprisedDovGabbay(King'sCollegeLondon),FinnJensen (AalborgUniversity),RudolfKruse(UniversityofMagdeburg),SimonParsons (QueenMary,UniversityofLondon)HenriPrade(IRIT,Toulouse),Torsten Schaub(UniversityofPotsdam),andPhilippeSmets(ULB,Bruxelles). Reviewers Theprogrammecommitteeisverygratefulforallthehardworkcontributed bythereviewers. Hopefully,wehavenotmissedanyonefromthefollowing list:BruceD'Ambrosio,FlorenceBannay,SalemBenferhat,PhilippeBesnard, HuguesBersini,ChristianBorgelt,RachelBourne,StefanBrass,LaurenceCholvy, RogerCooke,AdnanDarwiche,YannisDimopoulos,JurgenDix,DidierDubois, UweEgly,LindavanderGaag,JoergGebhardt,SiegfriedGottwald,RolfHaenni, Jean-YvesJa?ray,RadimJirousek,RuthKempson,U?eKjaerulf,FrankK- wonn,AljoschaKlose,JuergKohlas,PaulKrause,GerhardLakemeyer,Mounia Lalmas,JeromeLang,KimG. Larsen,NorbertLehmann,T. Y. Lin,Thomas Linke,KhalidMellouli,JeromeMengin,J. -J. Ch.Meyer,SanjayModgil,Yves Moinard,Sera?nMoral,DetlefNauck,AnnNicholson,PascalNicolas,Dennis VI Preface Nilsson,KristianG. Olesen,RainerPalm,ZdzislawPawlak,VincentRisch,Regis Sabbadin,CamillaSchwind,PrakashP. Shenoy,MilanStudeny,HeikoTimm, HansTompits,MarcoValtorta,andCeesWitteven. April1999 AnthonyHunterandSimonParsons TableofContents Onthedynamicsofdefaultreasoning GrigorisAntoniou...1 Non-monotonicandparaconsistentreasoning:Frombasicentailmentsto plausiblerelations OferArieliandArnonAvron ...11 AcomparisonofsystematicandlocalsearchalgorithmsforregularCNF formulas Ram'onB'ejarandFelipMany'a...22 Query-answeringinprioritizeddefaultlogic FaridBenhammadi,PascalNicolasandTorstenSchaub...32 Updatingdirectedbeliefnetworks BoutheinaBenYaghlaneandKhaledMellouli...43 Inferringcausalexplanations PhilippeBesnardandMarie-OdileCordier...55 Acritiqueofinductivecausation ChristianBorgeltandRudolfKruse...68 Connectinglexicographicwithmaximumentropyentailment RachelA. BourneandSimonParsons...80 Avoidingnon-groundvariables StefanBruning .. andTorstenSchaub...92 Anchoringsymbolstovisiondatabyfuzzylogic SilviaCoradeschiandAlessandroSa?otti...104 Filteringvsrevisionandupdate:Letusdebate!CorineCossartandCatherineTessier...116 Irrelevanceandindependenceaxiomsinquasi-Bayesiantheory FabioG. Cozman...128 Assessingthevalueofacandidate:Aqualitativepossibilisticapproach DidierDubois,MichelGrabischandHenriPrade...137 VIII TableofContents Learningdefaulttheories B'eatriceDuvalandPascalNicolas...148 Knowledgerepresentationforinductivelearning PeterA. Flach...160 Handling inconsistency e?ciently in the incremental construction of strati?ed beliefbases EricGr'egoire...168 Roughknowledgediscoveryandapplications J. W. GuanandD. A. Bell...179 GradientdescenttrainingofBayesiannetworks FinnV. Jensen...190 Opendefaulttheoriesovercloseddomains:Anextendedabstract MichaelKaminski...201 Shopboteconomics Je?reyO. KephartandAmyR. Greenwald...208 OptimizedalgorithmforlearningBayesiannetworkfromdata F'ediaKhalfallahandKhaledMellouli...221 Mergingwithintegrityconstraints S'ebastienKoniecznyandRam'onPinoP'erez...233 Boolean-likeinterpretationofSugenointegral IvanKramosil...245 AnalternativetooutwardpropagationforDempster-Shaferbelieffunctions NorbertLehmannandRolfHaenni...256 Onbottom-uppre-processingtechniquesforautomateddefaultreasoning ThomasLinkeandTorstenSchaub...268 Probabilisitclogicprogrammingundermaximumentropy ThomasLukasiewiczandGabrieleKern-Isberner...279 Lazypropagationandindependenceofcausalin?uence AndersL. MadsenandBruceD'Ambrosio...293 TableofContents IX AMonteCarloalgorithmforcombiningDempster-Shaferbeliefbasedon approximatepre-computation Seraf'?nMoralandAntonioSalmer'on...305 Anextensionofalinguisticnegationmodelallowingustodenynuanced propertycombinations DanielPacholczyk...316 Argumentationandqualitativedecisionmaking SimonParsonsandShawGreen...328 Handling di?erent forms of uncertainty in regression analysis: A fuzzy belief structureapproach SimonPetit-RenaudandThierryDenoeux...340 State recognition in discrete dynamical systems using Petri nets and evidence theory Mich'eleRombaut,ImanJarkassandThierryDenoeux...352 Robotnavigationandmapbuildingwiththeeventcalculus MurrayShanahanandMarkWitkowski...362 Informationfusioninthecontextofstockindexprediction StefanSiekmann,J..orgGebhardtandRudolfKruse...363 Defeasiblegoals LeendertvanderTorre...374 Logicaldeductionusingthelocalcomputationframework NicWilsonandJ'eromeMengin...386 AuthorIndex...397 OntheDynamicsofDefaultReasoning Grigoris Antoniou Gri?thUniversity,QLD4111,Australia UniversityofMacedonia,Thessaloniki,Greece ga@cit.gu. edu. au Abstract. Defaultlogicisaprominentrigorousmethodforreasoning withincompleteinformationbasedonassumptions. Itisastaticreas- ingapproach,inthesensethatitdoesn'treasonaboutchangesandtheir consequences. Ontheotherhand,itsnonmonotonicbehaviourappears whenchangestoadefaulttheoryaremade.

Argumentation provides tools for designing, implementing and analyzing sophisticated forms of interaction among rational agents. It has made a solid contribution to the practice of multiagent dialogues. Application domains include: legal disputes, business negotiation, labor disputes, team formation, scientific inquiry, deliberative democracy, ontology reconciliation, risk analysis, scheduling, and logistics.This book constitutes the thoroughly refereed post-proceedings of the Third International Workshop on Argumentation in Multi-Agent Systems held in Hakodate, Japan, in May 2006 as an associated event of AAMAS 2006, the main international conference on autonomous agents and multi-agent systems.The volume opens with an original state-of-the-art survey paper presenting the current research and offering a comprehensive and up-to-date overview of this rapidly evolving area. The 11 revised articles that follow were carefully reviewed and selected from the most significant workshop contributions, augmented with papers from the AAMAS 2006 main conference, as well as from ECAI 2006, the biennial European Conference on Artificial Intelligence.

This volume presents the latest developmentsinthegrowingareaofresearch at the interface of argumentation theory and multiagent systems. This area has grown tremendously with many papers appearing in the recent special issue of the Arti?cial Intelligence Journal on "Argumentation" and the special issue of IEEE Intelligent Systems on "Argumentation Technologies. " Over the last few years, argumentation has been gaining increasing imp- tance in multiagent systems, mainly as a vehicle for facilitating rational int- action (i. e. , interaction which involves the giving and receiving of reasons). This is because argumentation provides tools for designing, implementing, and a- lyzing sophisticated forms of interaction among rational agents. Argumentation hasmadesolidcontributionstothe practiceofmultiagentdialogues. Application domains include: legal disputes, business negotiation, labor disputes, team f- mation, scienti?c inquiry, deliberative democracy, ontology reconciliation, risk analysis, scheduling, and logistics.A single agent may also use argumentation techniquestoperformitsindividualreasoningbecauseitneedstomakedecisions under complex preferences policies, in a highly dynamic environment. Most papers in this volume appeared in the proceedings of the 4th Int- national Workshop on Argumentation in Multiagent Systems (ArgMAS 2007), which took place in Honolulu, USA, in conjunction with the International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS). This continues the success of the ArgMAS workshop series, which took place in t- dem with AAMAS in New York in 2004,Utrecht in 2005,and Hakodatein 2006.

This book constitutes the thoroughly refereed post-conference proceedings of the 9th International Workshop on Agent-Mediated Electronic Commerce, AMEC IX, co-located with the Sixth International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS 2007, held in Honolulu, Hawai, in May 2007, and the 5th Workshop on Trading Agent Design and Analysis, TADA 2007, co-located with the Twenty-Second AAAI Conference on Artificial Intelligence, AAAI 2007, held in Vancouver, Canada, in July 2007. This volume presents 15 carefully revised and selected papers from these workshops. The primary and complementary goal of both workshops was to continue to bring together novel work from diverse fields on modeling, implementation and evaluation of computational trading institutions and/or agent strategies. The papers originating from AMEC focus on a large variety of issues on auctions, negotiation, and strategic behavior in electronic marketplaces. The papers originating from TADA reflect the effort of the community to design scenarios where trading agent designers and market designers can be pitched against one another.

This volume contains revised versions of the papers presented at the sixth e- tion of the International Workshop on Argumentation in Multi-Agent Systems, (ArgMAS2009),heldinBudapest,Hungary,inassociationwiththe8thInter- tional Conference on Autonomous Agents and Multi-Agent Systems (AAMAS 2009) in May 2009. Previous ArgMAS workshops have been held in New York City, USA (2004),Utrecht, The Netherlands (2005),Hakodate, Japan (2006), Honolulu, USA (2007) and Estoril, Portugal (2008). The event has now clearlyestablisheditselfonthe internationalcalendaramongresearchersinc- putational argument and dialectics. A brief word to explain these topics is in order. Di?erent agents within a multi-agent system (MAS) potentially have di?erential access to information and di?erent capabilities, di?erent beliefs, di?erent preferences and desires, and di?erent goals. A key aspect of the scienti?c and engineering study of mul- agentsystemsthereforehasbeenthedevelopmentofmethodsandproceduresfor identifying, assessing, reconciling, arbitrating between, managing, and mitig- ing suchdi?erences. Marketmechanismsandvotingproceduresaretwo methods fordealing with thesedi?erences. Argumentationisanother.Argumentationcan be understood as the formal interaction of di?erent arguments for and against some conclusion (e. g. , a proposition, an action intention, a preference, etc. ). An agent may use argumentation techniques to perform individual reasoning for itself alone, in order to resolve con?icting evidence or to decide between c- ?icting goals it may have. Two or more agents may also jointly use dialectical argumentationto identify, express,andreconciledi?erencesbetweenthemselves, by means of interactions such as negotiation, persuasion, inquiry, and joint - liberation.