Showing papers by "Munindar P. Singh published in 2003"

Detecting deception in reputation management

Abstract: We previously developed a social mechanism for distributed reputation management, in which an agent combines testimonies from several witnesses to determine its ratings of another agent. However, that approach does not fully protect against spurious ratings generated by malicious agents. This paper focuses on the problem of deception in testimony propagation and aggregation. We introduce some models of deception and study how to efficiently detect deceptive agents following those models. Our approach involves a novel application of the well-known weighted majority technique to belief function and their aggregation. We describe simulation experiments to study the number of apparently accurate witnesses found in different settings, the number of witnesses on prediction accuracy, and the evolution of trust networks.

Searching social networks

Abstract: A referral system is a multiagent system whose member agents are capable of giving and following referrals. The specific cases of interest arise where each agent has a user. The agents cooperate by giving and taking referrals so each can better help its user locate relevant information. This use of referrals mimics human interactions and can potentially lead to greater effectiveness and efficiency than in single-agent systems.Existing approaches consider what referrals may be given and treat the referring process simply as path search in a static graph. By contrast, the present approach understands referrals as arising in and influencing dynamic social networks, where the agents act autonomously based on local knowledge. This paper studies strategies using which agents may search dynamic social networks. It evaluates the proposed approach empirically for a community of AI scientists (partially derived from bibliographic data). Further, it presents a prototype system that assists users in finding other users in practical social networks.

Agent communication languages: rethinking the principles

Abstract: Agent communication languages have been used for years in proprietary multiagent systems. Yet agents from different vendors--or even different research projects--cannot communicate with each other. The author looks at the underlying reasons and proposes a conceptual shift from individual agent representations to social interaction.

Trust, Reputation, and Security: Theories and Practice

Abstract: Discussions at the 5th Workshop on Deception, Fraud and Trust in Agent Societies held at the 1st International Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS 2002) centered around many important research issues. This paper attempts to challenge researchers in the community toward future work concerning three issues inspired by the workshop’s roundtable discussion: (1) distinguishing elements of an agent’s behavior that influence its trustworthiness, (2) building reputation-based trust models without relying on interaction, and (3) benchmarking trust modeling algorithms. Arguments justifying the validity of each problem are presented, and benefits from their solutions are enumerated.

Resolving Commitments among Autonomous Agents

Abstract: Commitments are a powerful representation for modeling multiagent interactions. Previous approaches have considered the semantics of commitments and how to check compliance with them. However, these approaches do not capture some of the subtleties that arise in real-life applications, e.g., e-commerce, where contracts and institutions have implicit temporal references. The present paper develops a rich representation for the temporal content of commitments. This enables us to capture realistic contracts and institutions rigorously, and avoid subtle ambiguities. Consequently, this approach enables us to reason about whether and when exactly a commitment is satisfied or breached and whether it is or ever becomes unenforceable.

Distributed enactment of multiagent workflows: temporal logic for web service composition

Abstract: We address the problem of constructing multiagent systems by coordinating heterogeneous, autonomous agents, whose internal designs may not be fully known. A major application area is Web service composition. We develop an approach that (a) takes declarative specifications of the desired interactions, and (b) automatically enacts them. Our approach is based on temporal logic, has a rigorous semantics, and yields a naturally distributed execution.

Emergent properties of referral systems

Abstract: Agents must decide with whom to interact, which is nontrivial when no central directories are available. A classical decentralized approach is referral systems, where agents adaptively give referrals to one another. We study the emergent properties of referral systems, especially those dealing with their quality, efficiency, and structure. Our key findings are (1) pathological graph structures can emerge due to some neighbor selection policies and (2) if these are avoided, quality and efficiency depend on referral policies. Further, authorities emerge automatically and the extent of their relative authoritativeness depends on the policies.

Nonmonotonic commitment machines

Abstract: Protocols for multiagent interaction need to be flexible because of the open and dynamic nature of multiagent systems. Such protocols cannot be modeled adequately via finite state machines (FSMs) as FSM representations lead to rigid protocols. We propose a commitment-based formalism called Nonmonotonic Commitment Machines (NCMs) for representing multiagent interaction protocols. In this approach, we give semantics to states and actions in a protocol in terms of commitments. Protocols represented as NCMs afford the agent flexibility in interactions with other agents. In particular, situations in protocols when nonmonotonic reasoning is required can be efficiently represented in NCMs.

An adaptive social network for information access: Theoretical and experimental results

Abstract: We consider a social network of software agents who assist each other in helping their users find information Unlike in most previous approaches, our architecture is fully distributed and includes agents who preserve the privacy and autonomy of their users These agents learn models of each other in terms of expertise (ability to produce correct domain answers) and sociability (ability to produce accurate referrals) We study our framework experimentally to study how the social network evolves Specifically, we find that under our multi-agent learning heuristic, the quality of the network improves with interactions: the quality is maximized when both expertise and sociability are considered; pivot agents further improve the quality of the network and have a catalytic effect on its quality even if they are ultimately removed Moreover, the quality of the network improves when clustering decreases, reflecting the intuition that you need to talk to people outside your close circle to get the best information

Trustworthy service composition: Challenges and research questions

Abstract: Web services have been gathering an increasing amount of attention lately. The raison d'etre of Web services is that we compose them to create new services. For Web services to be effectively composed, however, requires that they be trustworthy and in fact be trusted by their users and other collaborating services. In our conceptual scheme, principals interact as autonomous peers to provide services to one another. Trust is captured as a composite relationship between the trusted and the trusting principal. Principals help each other discover and locate trustworthy services and weed out untrustworthy players. The interactions of the principals combined with the needs of different applications induce interesting structures on the network. We apply multiagent systems techniques to model interactions among the principals. By varying the requirements of different applications, the needs of different principals, the existence of special principals such as trusted authorities, and the mechanisms underlying the interactions, we draw attention to a variety of important settings where Web services would be composed. One, leading to superior methods through which trust can be evolved and managed in realistic service-composition settings. Two, studying the relationships between aspects of trust for Web services and the evolution of Web structure.

Incentive mechanisms for peer-to-peer systems

Abstract: Most of the existing research in peer-to-peer systems focuses on protocol design and doesn’t consider the rationality of each peer. One phenomenon that should not be ignored is free riding. Some peers simply consume system resources but contribute nothing to the system. In this paper we present an agent-based peer-to-peer system, in which each peer is a software agent and the agents cooperate to search the whole system through referrals. We present a static and a dynamic pricing mechanism to motivate each agent to behave rationally while still achieving good overall system performance. We study the behavior of the agents under two pricing mechanisms and evaluate the impact of free riding using simulations.

Dynamic communities in referral networks

Abstract: Consider a decentralized agent-based approach for service location, where agents provide and consume services, and also cooperate with each other by giving referrals to other agents. That is, the agents form a referral network. Based on feedback from their users, the agents judge the quality of the services provided by others. Further, based on the judgments of service quality, the agents also judge the quality of the referrals given by others. The agents can thus adaptively select their neighbors in order to improve their local performance. The choices by the agents cause communities to emerge. According to our definition, an agent belongs to a community only if it has been useful to the other members of the community in prior interactions regarding a particular topic. Hence, the membership in different communities is determined based on relationships among the agents. This paper compares topic-sensitive communities of the above kind with communities as studied in traditional link analysis. It studies the correlation between the two kinds of communities as they emerge in referral networks and evaluates the two kinds of communities in terms of their effectiveness in locating service providers.

Engineering commitment-based multiagent systems: a temporal logic approach

Abstract: Commitments model important aspects of agent interactions, especially those arising in e-business. A small number of patterns of commitments accommodate a variety of realistic interactions among agents. We represent these patterns and agent behavior models formally and show how certain behavior models can be formally proved to be sound for certain patterns. Thus a designer may use a library of patterns and behaviors to engineer systems that are guaranteed to work correctly.

Commitments and causality for multiagent design

Abstract: This paper unifies two recent strands of research in multiagent system design. One, commitments are widely recognized as capturing important aspects of interactions among agents, but current approaches tend to emphasize individual commitments and typically restrict themselves to interactions between pairs of agents. Two, methodologies for multiagent system design consider protocols and coordination requirements, but do not seriously accommodate commitments. This paper proposes a methodology to infer commitments from an example conversation among several parties. Based on the conversation, we build a commitment causality diagram indicating the causal relations among the commitments. Using this diagram, we generate behavior models for each role. We show that the models produced successfully capture commitment-level protocols and allow flexible implementation of non-commitment communications provided the causal relations are preserved.

An Ontology for Web Service Ratings and Reputations

Abstract: Current Web services standards enable publishing service descriptions and finding services by matching requested and published descriptions based on syntactic criteria such as method signatures or service category. Emerging approaches such as DAML-S use DAML to formalize richer models for expressing capabilities of services. DAML-S would go beyond WSDL in terms of the richness of service descriptions. However, neither current Web services standards nor DAML-S provide a basis for selecting a good service or for comparing services that implement the same interface. In our view, service selection is the key problem to enable applicationto-application integration, which is the essential vision behind Web services. Existing approaches don’t address service selection, because service selection inherently involves trust and must consider criteria that are external to any published description of a service, whether in WSDL or DAML-S. Accordingly, this paper develops an ontology in which ratings of services (aggregated into reputations) can be organized and shared so as to facilitate service selection. This model is expressed in DAML and includes domainindependent as well as domain-specific attributes.

Self-organizing referral networks: a process view of trust and authority

Abstract: We are developing a decentralized approach to trust based on referral systems, where agents adaptively give referrals to one another to find other trustworthy agents. Interestingly, referral systems provide us with a useful and intuitive model of how links may be generated: a referral corresponds to a customized link generated on demand by one agent for another. This gives us a basis for studying the processes underlying trust and authority, especially as they affect the structure of the evolving social network of agents. We explore key relationships between the policies and representations of the individual agents on the one hand and the aggregate structure of their social network on the other.

Properties of referral networks: emergence of authority and trust

Abstract: Developing, maintaining, and disseminating trust in open environments is crucial. We develop a decentralized approach to trust in the context of service location. Service providers and consumers are modeled as autonomous agents participating in a multiagent system that functions as a referral network. When a service is requested, an agent may provide the requested service or give a referral to another agent. The agents can judge the quality of service obtained. Importantly the agents can adaptively select their neighbors, decide with whom to interact, and choose how to give referrals. The agents' actions lead to the evolution of the referral network. We study the emergent properties of referral networks, especially those dealing with their quality, efficiency, and structure. We first show how the exchange of referrals affect locating service providers, then identify undesirable network structures and show under which conditions these network structures emerge. A referral corresponds to a customized link generated on demand by one agent for another. Referrals thus yield a basis for studying the processes underlying trust and authority, especially as they affect the structure of the evolving social network of agents. Whereas existing work takes an after-the-fact look at Web structure, we can study the emergence of structure as it relates to the policies of the members. Further, we propose a graph-based representation of services that can be applied in conjunction with our referrals-based approach. This representation captures natural relationships among service domains and provides a simple means to accommodate the accrual of trust placed in a given party. Using this representation we study how number of interactions and ease at which agents try new providers affect locating trustworthy providers. The properties uncovered through this dissertation can serve as guidelines to develop robust referral systems that are both efficient and effective.

Trust, Reputation, and Security: Theories and Practice: AAMAS 2002 International Workshop, Bologna, Italy, July 15, 2002. Selected and Invited Papers

Abstract: How Human Trusters Assess Trustworthiness in Quasi-virtual Contexts- Challenges for Trust, Fraud and Deception Research in Multi-agent Systems- Designing for Trust- The Epistemic Role of Trust- Trustworthy Service Composition: Challenges and Research Questions- A Service-Oriented Trust Management Framework- A Fuzzy Approach to a Belief-Based Trust Computation- Annotating Cooperative Plans with Trusted Agents- Supervised Interaction - A Form of Contract Management to Create Trust between Agents- Evaluating Reputation in Multi-agents Systems- Towards Incentive-Compatible Reputation Management- Securing Agent-Based e-Banking Services- Specifying Standard Security Mechanisms in Multi-agent Systems- A Trusted Method for Self-profiling in e-Commerce- A Practical Study on Security of Agent-Based Ubiquitous Computing- Designing for Privacy in a Multi-agent World- Soft Security: Isolating Unreliable Agents from Society

Ladders of success: an empirical approach to trust

Abstract: Developing, maintaining, and disseminating trust in open environments is an important problem. We propose a lattice-based representation, which captures natural relationships among service domains and provides a natural means to accommodate the accrual of trust placed in a given party. In this manner, less important services (e.g., low-value transactions) can be used as gates to more important services (e.g., high-value transactions). Importantly, our lattice-based representation can be applied in conjunction with a social approach, such as one based on referrals. We first show that, where applicable, this approach yields superior efficiency (needs fewer messages) and effectiveness (finds more providers) than a vector representation. Next, we perform a sensitivity analysis of the lattice-based representation to study trade-offs between various factors that affect the performance of the approach.

The future of agent communication

Abstract: This short note discusses trends in computing environments where agent communication will have the most value. The main emerging trend in computing is toward openness, that is, toward large systems with dynamically changing, autonomous, heterogeneous components. This trend speaks the end of architecture in the traditional sense. More importantly, it emphasizes on declaratively specified interactions, that is, arms-length relationships, among independent components. Agents and multiagent systems provide key abstractions for computing in open settings. In particular, agent communications will be a key means for structuring distributed computations.

Trustworthy Service Caching: Cooperative Search in P2P Information Systems

Abstract: We are developing an approach for P2P information systems, where the peers are modelled as autonomous agents. Agents provide services or give referrals to one another to help find trustworthy services. We consider the important case of information services that can be cached. Agents request information services through high-level queries, not by describing specific objects as in caching in traditional distributed systems. Moreover, the agents autonomously decide whom to contact for a service, whom to provide a service or referral, whether to follow a referral and whether to cache a service. Thus the information system itself evolves as agents learn about each other and the contents of the caches of the agents change. We study here the effect of caching on service location and on the information system itself. Our main results are that, (1) even with a small cache, agents can locate services more easily; (2) since the agents that cache services can act like service providers, a small number of initial service providers are enough to serve the information needs of the consumers; and (3) agents benefit from being neighbours with others who have similar interests.

A Community-Based Approach for Selecting Service Providers.

Abstract: SREENATH, RAGHURAM MASTI. A Community-Based Rating System for Selecting Among Web Services. (Under the direction of Munindar P. Singh.) The current infrastructure for Web services has a static approach to discover a service. It is based on a common repository that has a simple search interface that lets the user query and find a provider for the desired service. More often than not, the repository produces a long list of service providers along with typical interfaces to talk to them. There is no support to evaluate service providers. Such evaluations are important to make a selection among competing service providers. This thesis develops a community-based approach for evaluating service providers. In this approach, agents cooperate with each other to evaluate different providers. Importantly, the agents rate each other, to decide how to weigh each other’s recommendations. The reasoning of each agent is enabled by a concept lattice, which supports a mechanism to rate the agents. A COMMUNITY-BASED RATING SYSTEM FOR SELECTING AMONG WEB SERVICES BY RAGHURAM SREENATH DEPARTMENT OF COMPUTER SCIENCE NORTH CAROLINA STATE UNIVERSITY RALEIGH, NC 27695 A THESIS SUBMITTED TO THE GRADUATE FACULTY OF NORTH CAROLINA STATE UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF COMPUTER SCIENCE RALEIGH DECEMBER 2002