Showing papers in "Autonomous Agents and Multi-Agent Systems in 2002"

Larks: Dynamic Matchmaking Among Heterogeneous Software Agents in Cyberspace

Abstract: Service matchmaking among heterogeneous software agents in the Internet is usually done dynamically and must be efficient. There is an obvious trade-off between the quality and efficiency of matchmaking on the Internet. We define a language called Larks for agent advertisements and requests, and present a flexible and efficient matchmaking process that uses Larks. The Larks matchmaking process performs both syntactic and semantic matching, and in addition allows the specification of concepts (local ontologies) via ITL, a concept language. The matching process uses five different filters: context matching, profile comparison, similarity matching, signature matching and constraint matching. Different degrees of partial matching can result from utilizing different combinations of these filters. We briefly report on our implementation of Larks and the matchmaking process in Java. Fielded applications of matchmaking using Larks in several application domains for systems of information agents are ongoing efforts.

Game Theory and Decision Theory in Multi-Agent Systems

Abstract: In the last few years, there has been increasing interest from the agent community in the use of techniques from decision theory and game theory Our aims in this article are firstly to briefly summarize the key concepts of decision theory and game theory, secondly to discuss how these tools are being applied in agent systems research, and finally to introduce this special issue of Autonomous Agents and Multi-Agent Systems by reviewing the papers that appear

Pricing in Agent Economies Using Multi-Agent Q-Learning

Abstract: This paper investigates how adaptive software agents may utilize reinforcement learning algorithms such as Q-learning to make economic decisions such as setting prices in a competitive marketplace. For a single adaptive agent facing fixed-strategy opponents, ordinary Q-learning is guaranteed to find the optimal policy. However, for a population of agents each trying to adapt in the presence of other adaptive agents, the problem becomes non-stationary and history dependent, and it is not known whether any global convergence will be obtained, and if so, whether such solutions will be optimal. In this paper, we study simultaneous Q-learning by two competing seller agents in three moderately realistic economic models. This is the simplest case in which interesting multi-agent phenomena can occur, and the state space is small enough so that lookup tables can be used to represent the Q-functions. We find that, despite the lack of theoretical guarantees, simultaneous convergence to self-consistent optimal solutions is obtained in each model, at least for small values of the discount parameter. In some cases, exact or approximate convergence is also found even at large discount parameters. We show how the Q-derived policies increase profitability and damp out or eliminate cyclic price “wars” compared to simpler policies based on zero lookahead or short-term lookahead. In one of the models (the “Shopbot” model) where the sellers' profit functions are symmetric, we find that Q-learning can produce either symmetric or broken-symmetry policies, depending on the discount parameter and on initial conditions.

Reliable Communication for Highly Mobile Agents

Abstract: The provision of a reliable communication infrastructure for mobile agents is still an open research issue. The challenge to reliability we address in this work does not come from the possibility of faults, but rather from the mere presence of mobility, which complicates the problem of ensuring the delivery of information even in a fault-free network. For instance, the asynchronous nature of message passing and agent migration may cause situations where messages forever chase a mobile agent that moves frequently from one host to another. Current solutions rely on conventional technologies that either do not provide a solution for the aforementioned problem, because they were not designed with mobility in mind, or enforce continuous connectivity with the message source, which in many cases defeats the very purpose of using mobile agents. In this paper, we propose an algorithm that guarantees delivery to highly mobile agents using a technique similar to a distributed snapshot. A number of enhancements to this basic idea are discussed, which limit the scope of message delivery by allowing dynamic creation of the connectivity graph. Notably, the very structure of our algorithm makes it amenable not only to guarantee message delivery to a specific mobile agent, but also to provide multicast communication to a group of agents, which constitutes another open problem in research on mobile agents. After presenting our algorithm and its properties, we discuss its implementability by analyzing the requirements on the underlying mobile agent platform, and argue about its applicability.

MySpiders : Evolve Your Own Intelligent Web Crawlers

Abstract: The dynamic nature of the World Wide Web makes it a challenge to find information that is both relevant and recent. Intelligent agents can complement the power of search engines to meet this challenge. We present a Web tool called MySpiders, which implements an evolutionary algorithm managing a population of adaptive crawlers who browse the Web autonomously. Each agent acts as an intelligent client on behalf of the user, driven by a user query and by textual and linkage clues in the crawled pages. Agents autonomously decide which links to follow, which clues to internalize, when to spawn offspring to focus the search near a relevant source, and when to starve. The tool is available to the public as a threaded Java applet. We discuss the development and deployment of such a system.

Adaptive Web Site Agents

Abstract: We discuss the design and evaluation of a class of agents that we call adaptive web site agents. The goal of such an agent is to help a user find additional information at a particular web site, adapting its behavior in response to the actions of the individual user and the actions of other visitors to the web site. The agent recommends related documents to visitors and we show that these recommendations result in increased information read at the site. It integrates and coordinates among different reasons for making recommendations including user preference for subject area, similarity between documents, frequency of citation, frequency of access, and patterns of access by visitors to the web site. We argue that this information is best used not to change the structure or content of the web site but rather to change the behavior of an animated agent that assists the user.

Satisficing Equilibria: A Non-Classical Theory of Games and Decisions

Abstract: Satisficing, or being “good enough,” is the fundamental obligation of rational decision makers. We cannot rationally choose an option, even when we do not know of anything better, unless we know it is good enough. Unfortunately, we are not often in the position of knowing that there could be no better option, and hence that the option must be good enough. A complete search through all logical possibilities is often impractical, particularly in multi-agent contexts, due to excessive computational difficulty, modeling complexity, and uncertainty. It can be equally impractical, if it is even possible, to determine the cost of the additional required search to find an option that is good enough. In a departure from the traditional notion of satisficing as a species of bounded rationality, satisficing is here redefined in terms of a notion of intrinsic rationality. Epistemic utility theory serves as the philosophical foundation of a new praxeological decision-making paradigm of satisficing equilibria that is applicable to both single- and multiple-agent scenarios. All interagent relationships are modeled by an interdependence function that explicitly accommodates both self and group interest, from which multilateral and unilateral selectability and rejectability mass functions can be derived and compared via the praxeic likelihood ratio test.

Using Hypergames to Increase Planned Payoff and Reduce Risk

Abstract: Hypergames are an emerging tool for combining decision theoretic and game theoretic information. We explain a new approach to the standard two-player zero-sum single-stage normal game that maximizes expected gain while quantifying possible loss. A DTGT agent can use this formulation to select a plan based on its assessment of an opponent's intent, its assessment of an opponent's unpredictability and its utility model of the situation.

Interpreting Information Requests in Context A Collaborative Web Interface for Distance Learning

Abstract: We describe the use of theories of agent collaboration and human dialogue processing in providing a principled basis for the design of web interfaces to multimedia information stores. The DIAL system, an implementation in the domain of information support for distance learning by students in an introductory programming class, is used to illustrate the efficacy of this approach. DIAL builds a representation of context that is based on the collaborative plans of the system and its user and uses this contextual information to reduce the communication burden. Context is represented by a structure of intentions that a user is attempting to satisfy. This structure is modified as tasks are completed or task descriptions are refined. DIAL interprets information requests relative to the prevailing context as it is represented by this structure. As a result, requests may be expressed more economically; contextual information is added by the system. Furthermore, DIAL uses information about the intentional context to respond and act collaboratively, rather than in the master-slave style typical of most current human-computer interfaces. DIAL and the access method it supports provide a unique support tool for distance learning environments as well as a demonstration of a general way in which agent models can be used to improve human-computer communication.

A Scheduling Scheme for Controlling Allocation of CPU Resources for Mobile Programs

Abstract: There is considerable interest in developing runtime infrastructures for programs that can migrate from one host to another. Mobile programs are appealing because they support efficient utilization of network resources and extensibility of information servers. In this paper, we present a scheduling scheme for allocating resources to a mix of real-time and non real-time mobile programs. Within this framework, both mobile programs and hosts can specify constraints on how CPU should be allocated. On the basis of the constraints, the scheme constructs a scheduling graph on which it applies several scheduling algorithms. In case of conflicts between mobile program and host specified constraints, the schemes implements a policy that resolves the conflicts in favor of the host. The resulting scheduling scheme is adaptive, flexible, and enforces both program and host specified constraints.

Linking Agent Concepts and Methodology with CAN

Abstract: The design, implementation and application of agent technology to significant problems is an inherently complex task. In order to enable programmers who are not agent experts to develop such systems, it is necessary not only to simplify current agent programming systems (such as JACK) but also to integrate such systems into a well-understood framework. Our vision for achieving this involves the three pillars of concepts, notation and methodology. These are supported by a formal semantics and by various system tools. In this paper we give a brief overview of our concepts and methodology, and show how a notation, together with a formal semantics for it, can be designed as a bridge between concepts and methodology, thus yielding a comprehensive and holistic picture of agent development. The notation is in the tradition of AgentSpeak and Kinny’s Ψ but is closer to BDI theories than its predecessors – in particular it has an explicit notion of goal. Paper ID: 165

On the Natural Selection of Market Choice

Abstract: An evolutionary approach to the problem of economic mechanism choice is presented. It demonstrates the power that a single participant has on the choice of a preferred market mechanism. A population of sellers, each with one of two possible economic mechanisms, facing potential buyers, is presented as a test model. It is shown that if two auctions, such as first- and second-price auctions, are to attract an identical number of buyers, although under the model assumptions their expected revenues are identical, sellers using a first-price auction mechanism will be selected for. However, if a second-price auction attracts one additional buyer, then it will be selected for by the evolutionary process. These results are extended to the choice between an arbitrary k- and l-price auctions.

Handshaking Mechanism for Conversation Policy Agreements in Dynamic Agent Environment

Abstract: Utilizing conversation policies, also called as interaction protocols, is one of the various efforts that have been devoted to the improvement of agent interoperability. The research on conversation policies so far has focused on developing conversation policy models and standard conversation policies. However, in a dynamic agent society where new agents can be introduced and transaction sequences can be changed, there is a need for agents to handle ad-hoc conversation policies. For this reason, this paper suggests a handshaking mechanism for conversation policy agreements that enables agents to exchange and agree to new conversation policies in run-time. A conversation policy model was designed for dynamic exchange and interpretation in the handshaking procedure. The issue of interpreting the conversation policies at run-time was addressed based on the FIPA ACL semantics. By utilizing the handshaking mechanism of this paper, agents can interoperate with each other more flexibly in a dynamically changing environment without being re-implemented for unknown ad-hoc conversation policies. A prototype agent platform was implemented utilizing the conversation policy model and the handshaking mechanism.

M&M's: Building Binary Software Components for Supporting Mobile-Agent Enabled Applications

Abstract: Mobile Agents provide a new promising paradigm for developing distributed applications. Nevertheless, although the basic concept has been around for some years and many agent platforms are available both from the industry and research community, there are currently few examples where the technology has been deployed in the real world. One important reason for this is that using the current available agent frameworks it is quite difficult to develop applications without having to center them completely on the agents and on the agent infrastructure. In this paper, we present the M&M project, taking place at the University of Coimbra. In this project, we are developing an extensive component-based framework that enables ordinary applications to use mobile agents in a flexible and easy way. By using this approach, applications can be developed using current object-oriented approaches and become able of sending and receiving agents by the simple drag-and-drop of mobility components. The framework was implemented using the JavaBeans component model and provides integration with ActiveX, which allows applications to be written in a wide variety of programming languages. By using this framework, the development of applications that can make use of mobile regents is greatly simplified, which can contribute to a wider spreading of the mobile agent technology.

Behavioral Self-Organization in Lifelike Synthetic Agents

Abstract: Modern computer graphics technology has enjoyed rapid development in recent years, attracting researchers and practitioners to explore a wide spectrum of applications ranging from computer-aided graphical design to artificial life and virtual reality This paper is concerned with the animation-based entertainment use of computer graphics, ie, to create digitally synthetic agents that can self-animate themselves, adapt to their virtual environments, and learn new behaviors to attain some specific goals Here we propose a synthetic agent computational architecture called inter-threaded motif-based behavioral self-organization architecture, in which one motif acquires a conditioned association from the presently sensed state of the environment to the requirement of a desired motion as well as a plausible behavioral pattern to enable such a motion, whereas another computes the optimal parameters for the identified behavior in fulfilling the motion requirement This architecture will enable animated behaviors to be automatically programmed based on the concurrent self-organization of individual motifs as well as their crisscrossing interactions

Case Studies in Security and Resource Management for Mobile Object Systems

Abstract: Mobile objects have gained a lot of attention in research and industry in the recent past, but they also have a long history. Security is one of the key requirements of mobile objects, and one of the most researched characteristics related to mobility. Resource management has been somewhat neglected in the past, but it is being increasingly addressed, in both the context of security and QoS. In this paper we place a few systems supporting mobile objects in perspective based upon how they address security and resource management. We start with the theoretical model of Actors that supports concurrent mobile objects in a programming environment. Then we describe task migration for the Mach microkernel, a case of mobile objects supported by an operating system. Using the OMG MASIF standard as an example, we then analyze middleware support for mobile objects. Mobile Objects and Agents (MOA) system, is an example of middleware level support based on Java. The active networks project, Conversant, supports object mobility at the communication protocol level. We summarize these projects, comparing their security and resource management, and conclude by deriving a few general observations on how security and resource management have been applied and how they might evolve in the future.