Yanbin Peng

Method of two stages semantic service discovery

With the rapid increasing of Web services in the internet, it is a challenge to discover target service efficiently. Existing service discovery method is inefficient, which has a bad scalability. Aiming at this problem, a method of two stages semantic service discovery is proposed. In this method, service is divided into two classes: core service and assistant service. Service discovery process is divided into two stages: 1) in the first stage, the proposed method searches core service library to find target core service; 2) in the second stage, target core service is composed with related assistant service to form required target composite service. Experiment data shows that the new service discovery method can improve the efficiency of service discovery process to a great extent, thus has a good scalability.

A Model of Service Behavior Based on Petri Nets with Weights

Service composition technology promises a new solution to an age-old problem: how to implement complex, large scale, cross-enterprise business application. However, we must ensure the compatibility between component services in composite service. Current research works in this field only concern semantics compatibility of service interface, yet ignoring the compatibility of service behavior. Therefore, this paper proposes a model of service behavior based on Petri-nets with weight. Based on the novel model, a method is proposed to verify and compute service behavior compatibility.

Automatic Service Composition Verification Based on Pi-Calculus

With the number of services increasing in a rapid speed in the Internet, it becomes a research hotspot to discover target service according to the service requirement. When single service can not satisfy the service requirement, services can be composed to provide complex function. There are already a lot of service composition methods, so it is crucial to provide a method to verify the correctness of composite service. This paper proposes a pi-calculus based automatic method to verify it. Experiment data shows that with the importing of the new verification method, the precision of service discovery process has been improved.

PD-Agent: A Flexible Agent BDI Model for Autonomic Computing

Aiming at challenging problem in building autonomic computing system, policy based management technology and intelligent agent technology was combined to build autonomic elements and autonomic computing system. Through importing policy into agent BDI model, a flexible agent model (PD-Agent) is built to satisfy the requirement of autonomic element. To solve logic omniscience problem and related problems of policy operator, defeasible logic is imported. Working mechanism of our proposed system illustrates that the new-type agents can construct autonomic computing system in a seamless way.

Three Proposition Semantics for Motivation Operator of Agent

Motivation is an indispensable element in reasoning rational behavior of intelligent agent. Recent study on the semantics explanation of motivation operator has its shortcomings such as the impossibility to solve logic omniscience problem. This shortcoming fundamentally restrict agent from applying to real application domain. This paper proposes three proposition semantics to describe motivation operator and proves that the novel motivation operator can avoid logic omniscience problem and related problems.

Policy Enhanced Grid Computing

Grid computing technology is a promising method to implement large-scale distributed resources sharing and cooperative problem solving across virtual organization. Agent with its intelligent nature is fit for representing network entity to encapsulate, provide and invoke grid service in the internet. However, the challenging problem is how to reflect the changing business requirement in grid computing system. Policy, which can be revised at runtime, can solve this problem. Therefore, in this paper policy is integrated into agent BDI model, to reflect changing requirement of business application. Analysis shows that with the novel agent (P-Agent), grid computing system can be built in a seamless way.

Revising defeasible autonomic computing system

As the scale of virtual organization getting larger and larger, managing service cooperation within virtual organization becomes more and more complex. Autonomic computing is a natural selection to solve this management complexity crisis problem. Due to dynamic nature of virtual organization, policy reflecting administrators goal should be revised in time. As an extension of our former research works, this paper proposes a novel defeasible autonomic computing system where policy revision is considered through reasoning meta rules. Case study shows that with non-monotonic reasoning ability of defeasible logic, policy can be revised according to environment factors.

Goal Theory Based on Defeasible Logic

Traditional goal theory adopts Kripkean possible world semantics based normal modal logic to describe goal operator, which has monotonic trait. This paper argues that goal has non-monotonic nature; therefore, defeasible logic is imported to build goal theory. Case study shows that the novel goal theory can solve logic omniscience ideally.

Norm Formation Based on Defeasible Logic

Agents are prone to constitute virtual organization in the internet to support novel application such as grid computing. Due to autonomous nature of agent, norm must be provided to control agent’s behavior. This paper contends that norm formation process is an argumentation process, and import defeasible logic to formalize this process. Case study shows that norm can be formed in a seamless way in the framework of defeasible logic. Keywords-agent; virtual organization; grid computing; norm; defeasible logic

A model of flexible exception handing facilitated by polices and contracts

Due to the dynamic nature of virtual organization (VO), it is necessary that multi-agent system aimed to form and operate VO should be robust in cooperating process. We present a model of flexible exception handling facilitated by polices and contracts, in which the flexible evolvement process of exception handling is divided into three phases: exception analysis, plan generation, and plan deployment. On the one hand, contract signed by cooperative partners describes the cooperation process of the partners, and on the other hand, policies can be used to describe the way to handle exception happened during cooperating process. The three phrases and the import of polices makes the exception handling process more flexible, thus makes the MAS more robust.