PMA-4	Rationality of Reward Sharing in Multi-agent Reinforcement Learning
Authors	Kazuteru Miyazaki	Tokyo Institute of Technology, Department of Computational Intelligence and Systems Science, Interdisciplinary Graduate School of Science and Engineering, 4259, Nagatsuta, Midori-ku, Yokohama, 226-8502 JAPAN
	Shigenobu Kobayashi	Tokyo Institute of Technology, Department of Computational Intelligence and Systems Science, Interdisciplinary Graduate School of Science and Engineering, 4259, Nagatsuta, Midori-ku, Yokohama, 226-8502 JAPAN
Abstract	This abstract contains many mathematical expressions. Please refer to the printed matter.
Keywords	Reinforcement Learning, Profit Sharing, multi-agent system, Rationality Theorem, reward sharing
Contact Author:	Kazuteru Miyazaki

PMA-6	The Security Mechanism In Multiagent System AOSDE
Authors	Zhongzhi Shi	Institute of Computing Technology, Chinese Academy of Sciences,Beijing 100080, China
	Ju Wang	Institute of Computing Technology, Chinese Academy of Sciences,Beijing 100080, China
	Hu Cao	Institute of Computing Technology, Chinese Academy of Sciences,Beijing 100080, China
Abstract	In the paper, we discuss the requirements on security of agent communication and its implementation in multiagent system. Multiagent system must run on security encrypting channel and provide multilevel check mechanism in order to cope with illegal intruder in distributed application environment. We propose an encrypting channel based on RSA and Rabin algorithm, providing signature and encrypting service is low layer, and present an authority-control mechanism DSM in high layer. In DSM, we design a method of hybrid authority check that agent can visit a kind of default agent service by the right of its identity or by agent ID. In terms of above mechanism a flexible security configuration is provided.
Keywords
Contact Author:	Zhongzhi Shi

PMA-8	Mobile Robot Navigation by Distributed Vision Agents
Authors	Takushi Sogo	Kyoto University, Japan
	Hiroshi Ishiguro	Kyoto University, Japan
	Toru Ishida	Kyoto University, Japan
Abstract	A Distributed Vision System (DVS) is an infrastructure for mobile robot navigation. The system consists of vision agents embedded in an environment and connected with a computer network, observes events in the environment, and provides various information to robots. We have developed a prototype of the DVS which consists of sixteen vision agents and simultaneously navigates two robots. This paper describes the detail of the prototype system, shows the robustness through experimentation, and considers problems in applying a multi-agent system to a real robot system through development of the DVS.
Keywords	distributed vision, vision agent, mobile robot navigation
Contact Author:	Takushi Sogo

PMA-13	Designing Multi-Agent Reactive Systems. A specification method based on Reactive Decisional Agent
Authors	Bouchaib Bounabat	ENSIAS, Rabat, Morrocco
	Rahal Romadi	ENSIAS, Rabat, Morrocco
	Salah Labhalla	FSM, Marrakech, Morrocco
Abstract	A Reactive system is one that is in continual interaction with its environment, and executes at a pace determined by that environment. Examples of such systems are network protocols, industrial-process control systems etc. The use of rigorous formal method in specification and validation, can help designers to limit the introduction of potentially faulty components during the construction of the system. Due to their complex nature, reactive systems are extremely difficult to specify and validate. In this paper, we propose a new formal model for the specification and the validation of such systems. This approach considers a Reactive System as a Reactive Multi-Agent System consisting of concurrent reactive agents that cooperate with each other to achieve the desired functionality. In addition, this approach uses formal synchronous specification and verification tools in order to specify and to verify the systems behaviors. Finally an example of application of the approach is mentioned.
Keywords	Reactive Agent, Reactive System, Specification, Verification, Formal Methods
Contact Author:	Bouchaib Bounabat

PMA-15	Distributed Fault Location in Networks Using Learning Mobile Agents
Authors	Tony White	Systems and Computer Engineering,Carleton University, 1125 Colonel By Drive, Ottawa,Ontario,Canada KIS 5B6
	Bernard Pagurek	Systems and Computer Engineering,Carleton University, 1125 Colonel By Drive, Ottawa,Ontario,Canada KIS 5B6
Abstract	This paper describes how multiple interacting swarms of adaptive mobile agents can be used to locate faults in networks. The paper proposes the use of distributed problem solving using learning mobile agents for fault finding. The paper uses a recently described architectural description for an agent that is biologically inspired and proposes chemical interaction as the principal mechanism for inter-swarm communication. Agents have behavior that is inspired by the foraging activities of ants, with each agent capable of simple actions; global knowledge is not assumed. The creation of chemical trails is proposed as the primary mechanism used in distributed problem solving arising from the self-organization of swarms of agents. Fault location is achieved as a consequence of agents moving through the network, sensing, acting upon sensed information, and subsequently modifying the chemical environment that they inhabit. Elements of a mobile code framework that is being used to support this research, and the mechanisms used for agent mobility within the network environment, are described.
Keywords	mobile agents, swarm intelligence, reinforcement learning, fault location
Contact Author:	Tony White

PMA-24	Remote Messaging Mechanism Supporting Agent Mobility based on CORBA
Authors	Byung-Rae, Lee	Dept. of Computer Science & Engineering, Korea University,Korea
	Kyung-Ah, Chang	Dept. of Computer Science & Engineering, Korea University,Korea
	Tai-Yun, Kim	Dept. of Computer Science & Engineering, Korea University,Korea
Abstract	Mobile agents[1] are active entities, which may migrate to meet other agents and access the place`s services. For agent collaboration in distributed environment, mobile agents should be able to exchange messages with each other, even if they are moving across a network. The ability to send messages to a moving agent is important mechanism for agent collaboration. In this paper, we propose CORBA[2] based remote messaging mechanism with a binding table. This mechanism allows messages in flight when a mobile agent moves, and messages sent based on an out-of-date table binding, to be forwarded directly to the mobile agents new location. In this mechanism, an agent can transfer messages directly to the target agent based on the binding table.
Keywords	agent, CORBA, messaging, MASIF, Mobility
Contact Author:	Byung-Rae, Lee
File Format:	PDF

PMA-25	Hierarchical Multi-Agent Organization for Text Database Discovery
Authors	Yong S. Choi	Seoul National University,Korea
	Jaeho Lee	The University of Seoul,Korea
Abstract	Agent approaches has been increasingly used within information technology to describe various computational entities. Especially, due to the proliferation of readily available text databases on the Web, agents have been often developed as the computational entities for discovering useful text databases on the Web. In this paper, we motivate the need for the hierarchical organization of those agents. The motivation is based on our experiences with the neural net agents for the text database discovery and an analysis of the tradeoff between the benefit of the hierarchical organization of agents and multi-agent coordination overhead. We first introduce the neural net agent and then motivate our multi-agent approach based on the hierarchical organization of neural net agents both analytically and experimentally.
Keywords	Information Retrieval, Multi-Agent Systems, Hierarchical Organization, Text Database Discovery, Neural Net Agent, Information Technology
Contact Author:	Jaeho Lee

PMA-27	Flexible Multi-Agent Collaboration using Pattern Directed Message Collaboration of Field Reactor Model
Authors	Tadashige Iwao	Fujitsu Laboratories,Japan
	Makoto Okada	Fujitsu Laboratories,Japan
	Yuji Takada	Fujitsu Laboratories,Japan
	Makoto Amamiya	Kyushu University,Japan
Abstract	In this paper, we propose a flexible multi-agent model, called the Field Reactor Model (FRM) for open system environment such as ubiquitous computing. The FRM unifies indirect communication with an abstract medium and pattern-oriented message communication. A field is an abstract medium, which is independent of the physical media, and dispatches messages to all agents in the field. Reactors are agents, which have transformation rules for messages using pattern matching. The collaboration method among reactors is pattern directed message collaboration that yields functional relations between patterns of agents. The pattern directed message collaboration provides a method for flexible multi-agent collaboration on heterogeneous platforms using patterns and supports to change collaboration dynamically. Several fields on various networks can be combined into one logical field. A logical field enables agents to collaborate across various media and heterogeneous platforms. We describe how to apply the computation scheme originated from dataflow to the pattern directed message collaboration. Also, we show the flexibility of the model with an example of file format translations.
Keywords	Agent-oriented system, Multi-agent, Distributed system
Contact Author:	Tadashige Iwao

PMA-29	Formal Semantics of Acknowledgements, Agreements, and Disagreements
Authors	Norihiro Ogata	Assistant Professor, Dept. of Language and Culture, Osaka University, Japan
Abstract	This paper investigates formal semantics of acknowledgements, agreements, and disagreements as a formal basis of associating information sharing among agents with communications, especially dialogues. Shared information is modeled as circular objects using hypersets. Circular objects have infinite information but have finite representations, so their operations are also defined finitely, which are called corecursive definitions. This property helps to solve the problem of finiteness of acknowledgements. Moves in communications or acts in group actions are formalized as maps from a bit of shared information to a bit of shared information. Acknowledgements, agreements, and disagreements are formalized as functions by corecursion. Although we can assume many courses of communications or group actions to shared information, the resulting states, i.e., shared information is unique. Our semantics will assure such a property.
Keywords	Dialogue, Acknowledgement, Agreement, Formal Semantics, Hyperset, Corecursion, Information Sharing
Contact Author:	Norihiro Ogata

PMA-30	A Class of Isomorphic Transformations for Integrating EMYCIN-Style and PROSPECTOR-Style Systems into a Rule-Based Multi-Agent System
Authors	Xudong Luo	Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong
	Chengqi Zhang	School of Computing and Mathematics, Deakin University, Australia
	Ho-fung Leung	Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong
Abstract	The motivation for investigating the transformation between the EMYCIN model and the PROSPECTOR model lies in a realistic consideration. In the past, expert systems exploited mainly the EMYCIN model and the PROSPECTOR model to deal with uncertainties. In other words, a lot of stand-alone expert systems which use these two models are available. If there are reasonable transformations of uncertainties between the EMYCIN model and the PROSPECTOR model, we can use the Internet to couple them together so that the integrated systems are able to exchange and share helpful information with each other, thereby improving their performance through cooperation. In this paper, we discovered a class of exactly isomorphic transformations between uncertain reasoning models, as used by EMYCIN and PROSPECTOR. More interestly, among the class of isomorphic transformation functions, different ones can handle different degrees to which domain experts are optimistic or pessimistic if they perform such a transformation task.
Keywords	Multi-agent, uncertainty, distributed expert systems, algebra
Contact Author:	Xudong Luo

PMA-31	Reliable Agent Communication - a Pragmatic Perspective
Authors	David Kinny	University of Melbourne, Australia
Abstract	It is well recognized that Agent Communication Languages (ACL's) are a critical element of Multi-Agent Systems and a key to their successful application in commerce and industry. The field of Protocol Engineering, which addresses the problems of specifying and verifying machine communication languages and testing implementations, has developed powerful theoretical and automated techniques for doing this, and more importantly, a mature understanding of the requirements that communication language and protocol specifications should meet. Unfortunately, those developing and promulgating ACL's appear not to have taken advantage of this body of knowledge. An examination of the current ACL specifications being developed by the Foundation for Intelligent Physical Agents (FIPA) reveals a confusing amalgam of different formal and informal specification techniques whose net result is ambiguous, inconsistent and certainly underspecified. Allowances must be made, as these are draft specifications, but rather than providing a verified foundation for reliable communication between heterogeneous agents, they seem likely to lead to a host of unreliable and incompatible implementations, or to be ignored in favour of more pragmatic and robust approaches. In this paper, we propose a set of requirements and desiderata against which an ACL specification can be judged, briefly explore some of the shortcomings of the FIPA ACL and their origins, and contrast it with a small ACL which was designed with reliability and ease of verification as prime objectives.
Keywords	ACL, MAS, protocols, standards
Contact Author:	David Kinny

PMA-34	Persisting Autonomous Workflow for Mobile Agents Using a Mobile Thread Programming Model
Authors	Minjie Zhang	Department of Computer Science and Software Engineering The University of Newcastle, Australia
	Wei Li	Department of Information Management Capital University of Economics and Business, China
Abstract	In this paper, we present a Mobile Thread Programming Model (MTPM),a model to simulate the persistence of a migratory thread, to overcome the problem of coexistence of mobility, persistence and autonomy for mobile agents. An advantage of MTPM over other code mobility paradigms is that the model simulates strong mobility at the application-level rather than at the system-level as used in many strong mobility-supporting systems. It is runtime dependent to migrate threads at system-level. However, MTPM is constructed on Java Virtual Machine (JVM) by using Serialization and Remote Method Invocation (RMI), thus it is suitable to heterogeneous environments without introducing new spatial and time complexities in the implementation. Distributed Task Plan (DTP), which is detailed in this paper, is a flexible implementation model of MTPM used to simulate the persistence of an agent thread. Also, a DTP is embedded with navigational and computational autonomies, so that a mobile agent can obtain a continuous and autonomous workflow only by executing a DTP.
Keywords	code mobility, thread migration, autonomous agent, mobile agent
Contact Author:	Minjie Zhang

PMA-36	Making Rational Decisions in N-by-N Negotiation Games with a Trusted Third Party
Authors	Shih-Hung Wu	National Tsing Hua University, Taiwan
	Von-Wun Soo	National Tsing Hua University, Taiwan
Abstract	The optimal decision for an agent to make at a given game situation often depends on the decisions that other agents make at the same time. Rational agents will try to find a stable equilibrium before taking an action. Rational agents can use the negotiation mechanism to reach the equilibrium. In the previous work, we proposed the communication actions of paying guarantee or compensation to convince or persuade other agents and a negotiation protocol for agents to communicate with a trusted third party. In this paper, we extend the negotiation mechanism to deal with n-by-n games and justify its optimality with the underlying assumptions. During the negotiation process, each agent makes suggestions on how they can reach equilibrium while maximizing its own payoff. The mechanism can deal with all the game situations and find an acceptable equilibrium that gives optimal payoffs.
Keywords	Negotiation, Game thoery, Multi-agent coordination
Contact Author:	Shih-Hung Wu

PMA-37	Predicting User Actions Using Interface Agents with Individual User Models
Authors	Jung-Jin Lee	University of Connecticut, USA
	Robert McCartney	University of Connecticut, USA
Abstract	The incompleteness and uncertainty about the state of the world and about the consequences of actions are unavoidable. If we want to predict the performance of multiuser computing systems, we have the uncertainty of what the users are going to do, and how that affects system performance. Intelligent interface agent development is one way to mitigate the uncertainty about user behaviors by perdicting what users will do based on learned users' behaviors, preferences, and intentions. This work focuses on developing user models that can analyze and predict user behavior in multi-agent systems. We have developed a formal theory of user behavior prediction based on hidden Markov models. This work learns the user model through a time-series action analysis and abstraction by taking users' preferences and intentions into account in order to formally define user modeling.
Keywords	Interface, Agents, User, Modeling, Prediction, Abstraction, Probabilistic, Reasoning
Contact Author:	Jung-Jin Lee

PMA-38	How to design good rules for multiple learning agents in scheduling problems?
Authors	Keiki Takadama	ATR HIP Research Labs.
	Masakazu Watabe	ATR HIP Research Labs.
	Katsunori Shimohara	ATR HIP Research Labs.
	Shinichi Nakasuka	Univ. of Tokyo
Abstract	This paper explores how to design good rules for multiple learning agents in scheduling problems, and investigates what kind of factors are required to find good solutions with small computational costs. Through intensive simulations of crew task scheduling in a space shuttle/station, the following experimental results have been obtained: (1) an integration of a solution improvement factor, an exploitation factor, and an exploration factor contributes to finding good solutions with small computational costs; (2) an introduction of these three factors into rules results in a good rule designs for multiagent environments; and (3) this kind of rule design can be applied to scheduling problems as well as CSPs (constraint satisfaction problems).
Keywords	rule design, scheduling problem, multiple learning agents, organizational learning, learning classifier system
Contact Author:	Keiki Takadama

PMA-39	An Agent Architecture for Strategy-centric Adaptive QoS Control in Flexible Videoconference System
Authors	Takuo Suganuma	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University
	SungDoke Lee	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University
	Takuji Karahashi	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University
	Tetsuo Kinoshita	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University
	There is/are more author(s) in this paper.
Abstract	In this paper, we propose an agent architecture to improve the flexibility of a videoconference system with strategy-centric adaptive QoS (Quality of Service) control mechanism. The proposed architecture realizes more flexibility by changing their QoS control strategies dynamically. To switch the strategies, system considers the properties of problems occurred on QoS and status of problem solving process. This architecture is introduced as a part of knowledge base of agent that deals with cooperation between software module of videoconference systems. We have implemented the mechanism, and our prototype system shows its capability of flexible problem solving against the QoS degradation, along with other possible problems within the given time limitation. Thus we confirmed that the proposed architecture can improve its flexibility of a videoconference system compare to traditional systems.
Keywords	videoconference system, multiagent, adaptive QoS control, ADIPS, agent architecture, flexible system
Contact Author:	Takuo Suganuma

PMA-7	Belief-Goal-Role Agents versus BDI Agents
Authors	Walid Chainbi	FSEG,Tunisia
Abstract	This paper proposes a particular type of rational agent : a Belief-Goal-Role agent. The primary aim of this paper is to show a new approach to designing an agent which is different from the most known in DAI field. Indeed, beliefs, goals, roles are relevant to our study of cooperation which have lead to the identification of communication concepts and organization concepts. The behavioral semantics as well as the specification of our agent model are presented.
Keywords	BDI agents, Multiagent systems, Temporal logic
Contact Author:	Walid Chainbi

PMA-9	Intention Spreading: an extensible theme to protect mobile agents from read attack hoisted by malicious hosts
Authors	Sau-Koon Ng	The Chinese University of Hong Kong, Hong Kong
	Kwok-Wai Cheung	The Chinese University of Hong Kong, Hong Kong
Abstract	An approach called Intention Spreading is recently proposed to protect mobile agents from read attack. Two terms are involved: Intention spreading protects a mobile agent from read attack by adding noisy code to spread its original intention; read attack helps the malicious hosts to win over other hosts in a competitive environment by analyzing the content of the mobile agents, without modification of the agents' static code or the dynamic state. We have three main categories to distribute mobile agents: source code, intermediate code and compiled native binary code. The source code category suffers the direct read attack because this category is readily readable. The remaining categories suffer a certain degree of read attack after software reverse engineering. Besides, read attack could be simply obvious in some mobile agent systems in which the system calls are monitored for host security. In this paper, ... (please refer to paper)
Keywords	Intention Spreading, read attack, code addition and reconfiguration, chaffing and winnowing, multi-agent approach, classical cryptography
Contact Author:	Sau-Koon Ng

PMA-10	A Trust-Level Exchanging Protocol in Mobile Agent Systems for Security and Performance benefits
Authors	Sau-Koon Ng	The Chiense University of Hong Kong, Hong Kong
Abstract	Security and Performance in mobile agent systems contrast to each other. To manage both host security and agent security, the mobile agent is expanded for many reasons, and the expanded mobile agent consumes communication and computation resources many times more than the original mobile agent that operates in a network of trusted nodes. This is always considered as a trade-off factor in deploying secure mobile agent systems. However, this trade-off factor challenges one of the primitive ideas of the mobile agent paradigm: allow the mobile devices, which are bottlenecked by the intermittent transmission-link, to dispatch the small-sized mobile agent, then go offline for a while, and retrieve the useful and filtered information brought by the mobile agent. In this paper, a trust-level exchanging protocol is described to address both security and performance factors, and to preserve the essence of the mobile agent paradigm. The security schemes proposed, both host security and agent security, in literature are surveyed to weigh their communication and computation redundancies. The protocol is illustrated in two cases: two-hop boomerang agents and multi-hop agents ... (please refer to paper)
Keywords	Host security, Agent security, trust-level exchanging protocol
Contact Author:	Sau-Koon Ng

PMA-14	MASCAN: A Multi-Agent System for Transmission Cost Allocation
Authors	Yonghe Yan	Department of Computer Science & Information Systems, the University of Hong Kong, Pokfulam Road, Hong Kong
	Jerome Yen	Department of System Engineering & Engineering Management, Chinese University of Hong Kong, Shatin, New Territories, Hong Kong
Abstract	Abstract. In the paper we proposed a multi-agent system, MASCAN-Multi-Agent System for Cost Allocation on Network, to solve the cost allocation problem on network flow based on minimum cost flow problem. The agent in MASCAN, representing a node of network, does not receive any centralized control and centralized information source, and make decisions in his own behalf to compete lowest cost with other agents. The multi-agent system generates insights and builds intuition of each agent during the cost allocation, and each agent try to find a possible cheapest path under a fair-play practice. Under the circumstance that all the agents are rational, the cost allocated to each agent is the equilibrium point of a zero-sum game subject to the constraints of the given network. MASCAN can be used to model and to support cost allocation problem of power transmission network under the deregulation of electric utility industry.
Keywords	multi-agent system, transmission cost allocation, zero-sum game
Contact Author:	Yonghe Yan

PMA-17	Shine: a Cyber-community Application Platform - A Proposal -
Authors	Sen Yoshida	NTT Communication Science Laboratories, Japan
	Takeshi Ohguro	NTT Communication Science Laboratories, Japan
	Koji Kamei	NTT Communication Science Laboratories, Japan
	Kaname Funakoshi	NTT Communication Science Laboratories, Japan
	There is/are more author(s) in this paper.
Abstract	We are developing an environment in cyber-society for supporting the communication utilized by communities and for facilitating formation, maintenance and evolution of communities. This paper proposes a common platform that provides the foundation for the various applications used in this environment. The platform, named Shine, is designed as a multi-agent architecture. We will discuss design requirements such as organization of modules, communication protocols for agents and user modeling.
Keywords	socialware, multi-agent system, architecture, communityware
Contact Author:	Sen Yoshida

PMA-18	Adaptive Tutoring Agent System
Authors	Youngmee Choi	Department of Multimedia, University of SungKyul, Korea
	Moonwon Choo	Department of Multimedia, University of SungKyul, Korea
	Minkoo Kim	Department of Information & Computer Engineering, University of Ajou, Korea
Abstract	Many advisory tutoring systems have been developed in an attempt to achieve a high-level discovery learning performed in formal as well as informal environments. These systems usually try to improve localized weak points of learning process based on current interactions. However, they often fail in decision making to capture the procedures of a student's own way of learning. This paper proposes an Adaptive Tutoring Agent System(ATAS) that assists a user's learning activity through utilizing the history of user's interactions, thus, exploiting student-oriented discovery learning. The agents adaptively build a student model and interact with a student as well as other agents to provide useful advices at a proper time. The appropriateness of advice depends on the agents' reasoning, interacting, and referring to user model, domain knowledge, teaching strategy, and interfacing objects. ATAS was applied to a simple English learning environment.
Keywords
Contact Author:	Youngmee Choi

PMA-19	A Society of Natural Language Agents on the Internet
Authors	Wen-Lian Hsu	Institute of Information Science, Academia Sinica ,Taipei, Taiwan, Republic of China
	Yi-Shiou Chen	Institute of Information Science, Academia Sinica ,Taipei, Taiwan, Republic of China
	Yuan-Kai Wang	Institute of Information Science, Academia Sinica ,Taipei, Taiwan, Republic of China
Abstract	As the Internet is flooded with different types of documents (unstructured, semi-structured texts) and heterogeneous databases, information retrieval and extraction become increasingly difficult. To deal with the information service problem on the Internet, we are experimenting with an agent society in which agents act as spokesperson for various web pages, databases and provide information integration services in a cooperative manner. Furthermore, these agents can communicate with restricted natural language. These agents are particularly useful in information retrieval, extraction and integration services where interfaces in natural language are appropriate.
Keywords	natural language, agent communication language, knowledge base, information extraction
Contact Author:	Wen-Lian Hsu

PMA-21	An Intelligent Telemedicine Application Server toward a Digital Hospital Using Multi-Agent Systems
Authors	Il Kon Kim	Department of Computer Science, Kyungpook National University, Taegu 702-701, Republic of Korea
	Seong K. Mun	Imaging Science and Information Systems Center, Georgetown University Medical Center, Washington DC, 20007 USA
Abstract	Telemedicine applications are becoming widespread in clinical applications such as nephrology, urology, radiology, pathology, and surgery. Such increased access to telemedicine applications will improve the quality of care but also increase the complexity of interoperability and the volume of services provided. Internet application server technologies, browsers, DICOM (Digital Imaging and Communication in Medicine) server, and multi-agents system platforms make it possible to design and develop a digital hospital that will advance in an intelligent telemedicine application server structure, providing a continuum of on-line/telemedicine care while reducing the complexity for the end user. To build many types of "write-once-run-anywhere" digital hospital applications and allow them to dynamcally adapt to them to dynamically adapt to a variety of multi-agents platform restrictions, we provide a framework to support agents' services for digital hospital and intelligent telemedicine applications. In order for our framework to be clinically relevant, we draw on experience and knowledge acquired in the field of urology, dialysis, emergency, hospice telemedicine, and PACS (Picture Archiving and Communications System).
Keywords	Telemedicine, Multi-Agents, Agent Space, FIPA, DICOM, Intelligence, Multimedia, Tailorability
Contact Author:	Il Kon Kim

PMA-22	Coordinating Multiple Collaborative Strategies in Simulated Robotic Soccer Teams
Authors	Tung-Ing Jiang	Department of Computer Science National Tsing Hua University,Taiwan
	Von-Wun Soo	Department of Computer Science National Tsing Hua University,Taiwan
Abstract	To coordinate each player's actions at different levels of individual, group, and team tactics in a dynamic real time soccer game is a challenging problem for multi-agent study. In this paper we investigate mechanisms of coordinating multiple collaborative strategies in playing the simulated RoboCup soccer games. We propose the notion of using attractive and repulsive force vectors to represent a movement function of a given player. The movement function can determine the direction as well as the speed of movement of the player. The collaborative strategies and sub-strategies can be expressed in terms of different weighted formula of these force vectors. We show how the collaborative strategies and sub-strategies can be coordinated by superposition of force vectors, role assignment as well as differentiation of different game situations.
Keywords	coordinate, strategy, RoboCup
Contact Author:	Tung-Ing Jiang

PMA-23	Multi-agent Organization for Global Information Gathering
Authors	Timothy Lynn	Department of Computer Science National Tsing Hua University, Taiwan
	Jyi-Shane Liu	Department of Computer Science, National Chengchi University, Taiwan
	Von-Wun Soo	Department of Computer Science National Tsing Hua University, Taiwan
Abstract	Internet has emerged as an important medium for obtaining information as well as disseminating information. Yet, it has not been easy to harness the information power from this gigantic data source. In this paper, we propose a multi-agent organization for global information gathering and customized delivery service. The target application domain concerns with time-critical, geographically remote and global information sources, such as commodity trade market. We adopt an organizational approach with functional decomposition and role-obligation coordination. In order to cope with task requirements, agents are dispatched as teams for regional information gathering and digesting. We emphasize task coordination in regional information gathering team and their operational autonomy. We also discuss trade-off in certain design issues.
Keywords	Multi-agent, coordination, information gathering, organization
Contact Author:	Timothy Lynn

PMA-26	Unified Agent Cooperation Interface - a Critical-Mass Success Factor for MAS Design
Authors	Soe-Tsyr Yuan	Information Management Department, Fu-Jen University, Taiwan
	Zeng-Lung Wu	Information Management Department, Fu-Jen University, Taiwan
Abstract	Conversation-based multi-agent system design had been dominating most current multi-agent systems for a few years, which possess the capabilities of autonomy, adaptation, and cooperation. Supposedly, multi-agent systems should be prevailing enough in order to tackle dynamic aspects of problems in a variety of domains. However, as for industries, multi-agent systems are still found to be in its 'birth' stage where they only show their new values but yet await for further explorations and improvements in order to attract critical mass of information executives or software developers. Nevertheless, what are the success factors that can result in a critical mass of multi-agent system designers? This paper shows a possible success factor - the bottom-up design. The bottom-up design make it possible that agents are reassembled into multi-agent systems and agents are reused as needed. However, what do we need to successfully support the bottom-up design? The key is something that wraps each agent so that it exempts the designers from the careful detailed deployment of cooperation knowledge inside the agent and automates the cooperation process. This thing should be independent of the functions of agents. We name this thing as a Unified Agent Cooperation Interface. This paper then presents a novel methodology of implementing the Interface.
Keywords	multi-agent, critical-mass success factor, bottom-up design, unified cooperation wrapper
Contact Author:	Soe-Tsyr Yuan

PMA-28	Collaborating Heterogeneous Agents of Valued ConstraintSatisfaction Problems
Authors	Xudong Luo	Department of Computer Science and Engineering The Chinese University of Hong Kong,Hong Kong
	Ho-fung Leung	Department of Computer Science and Engineering The Chinese University of Hong Kong,Hong Kong
	Jimmy Ho-man Lee	Department of Computer Science and Engineering The Chinese University of Hong Kong,Hong Kong
Abstract	This paper develops an agent-based methodology for valued constraint satisfaction problems (VCSPs), in which each agent is associated with a VCSP and an agreement among all agents is reached by their negotiation. A negotiation procedure concerns three key issues: protocol used to organize a negotiation, synthesis of agents' local evaluations to a proposal, and transformation of agents' local evaluations when different agents are associated with heterogeneous instances of VCSPs. In particular, we (1) propose a protocols for negotiation among agents: the greedy protocol; (2) present a gray model for synthesizing agents' local evaluations to a proposal; (3) introduce two special kinds of VCSPs: the certainty factor constraint satisfaction problems (CFCSPs) and the probability constraint satisfaction problems (PCSPs); and (4) formulate criteria for a reasonable transformation between two heterogeneous VCSPs, and construct two transformation functions between a CFCSP and an FCSP, and between a PCSP and an FCSP.
Keywords	Constraint Satisfaction Problem, Fuzzy Set, Multi-Agent, Algebra
Contact Author:	Xudong Luo

PMA-33	Learning Improves Mobile Agent Efficiency
Authors	Sam Joseph	Toshiba, Japan
	Masanori Hattori	Toshiba, Japan
	Naoki Kase	Toshiba, Japan
Abstract	Mobile agent systems incur resource overheads as they acquire and maintain knowledge about their network environment. This paper presents a technique that, for a specic set of cases, optimises network resource consumption through dynamic selection of agent interaction protocols. This process is made possible through learning on the part of static knowledge agents, which can then predict the likely length of a given search procedure. It is these predications combined with information about the network environment that allow quantitative comparison between different search protocols.
Keywords
Contact Author:	Sam Joseph

PMA-35	A Model of Coalition Formation based on Agents' Mental States
Authors	Takashi KATOH	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University, Japan
	Tetsuo KINOSHITA	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University, Japan
	Norio SHIRATORI	Research Institute of Electrical Communication / Graduate School of Information Sciences, Tohoku University, Japan
Abstract	In this paper, we propose a method of coalition formation for allocating tasks to appropriate agents for the sake of improvement of the efficiency of multiagent systems. To realize this idea, we introduce the mental states to agents, which are the internal information of the agents. The mental states reflect the agents' cooperative behavior of the past. Next, we introduce loose coalition, a concept of a coalition of agents based on the mental states. Using the agents' value of sense defined by their mental states, each agent can give priority to the loose coalitions to ask the working status or allocating tasks. Thus loose coalitions with higher priority will be a candidate to cooperate with. Furthermore loose coalitions enable agents to collect information for task allocation, e.g. busyness of loose coalitions efficiently. Therefore, the agents on the system can decide its behavior properly, depending on the current status of the system, and the efficiency of the system is improved. We demonstrate the effectiveness of our proposal with empirical results.
Keywords	loose coalition, multiagent systems, coalition formation, mental states
Contact Author:	Takashi KATOH

PMA-40	CM-RELVIEW: A Tool for Causal Reasoning in Multiagent Environments
Authors	Brahim Chaib-draa	Laval University, Canada
Abstract	Analytical techniques are generally inadequate for dealing with interrelationships or causalities among a set of individual and social concepts. Usually, cognitive maps are used to cope with this type of causalities. However, the classical view of cognitive maps is based on an intuitive view with ad-hoc rules and no precise semantics of the primitive concepts, neither a sound formal treatment of relations between concepts. In this paper, we first present a formal model for cognitive maps with a precise semantics based on relation algebra and the software tool, CM-RELVIEW, in which it has been implemented. Then we investigate the issue of using CM-RELVIEW in multiagent environments, particularly: (1) the qualitative distributed decision making and; (2) the organization of agents considered as a wholistic approach. For each of these aspects, we focus on the computational mechanisms developed within CM-RELVIEW to support it. Keywords: Causal reasoning, cognitive maps, interrelationships, relation algebra, qualitative distributed decision making, organization of agents considered as a wholistic approach.
Keywords	Causal reasoning, cognitive maps, interrelationships, relation algebra, qualitative distributed decision making, organization of agents considered as a wholistic approach.
Contact Author:	Brahim Chaib-draa

PMA-43	Self-organization and Collaborative Learning in Multi-Agent Systems
Authors	Concepcion L. Khan	Computer Science and Information Management Program School of Advanced Technologies Asian Institute of Technology, Thailand
	R. Sadananda	Computer Science and Information Management Program School of Advanced Technologies Asian Institute of Technology, Thailand
Abstract	The volume and diversity of applications involving software agents require multiple agents to group together and coordinate their tasks. However, the proliferation of software agents in the network can be prohibitively large for all available agents to be tracked or to identify peers of any given agent. In this research, we are exploring methodologies that could hasten the learning process of agents from both user cues and other similar agents using organizational hierarchies. We demonstrate the viability of self-organization in multi-agent systems in defining the organization and functional inheritance of hierarchies that could be used by agents for sharing domain specific knowledge, reusing of existing architectures and designing future agent-paradigms.
Keywords
Contact Author: