The Study Of Target-Oriented Migrating Workflow Based On BDI Architechture

Workflow management is a kind of technology to assemble business processes and make them accomplished automatically or semi-automatically. It is one of the hot issues on which has been focused and researched by many areas including Computer Science, Automation, Management and Advanced Manufacturing. As the popularization of Internet and the development of Agent technology, especially the Mobile Agent technology, Migrating workflow based on Mobil Agent becomes a new direction within the workflow management area.When applying Mobile Agent technology, most of the researchers of migrating workflow followed WfMC standards, which ask agents to have explicit logic about business processes and make process-oriented workflow specifications in advance to drive the agents. Although the process-oriented workflow methods can easily find solutions to problems for mobile agents, the inherent structure attributes of "process" weakens the agility of agents and requires designers to have experience and good knowledge about business processes and transfer logics of actions. To solve this problem, a new method called "Target-oriented Migrating Workflow" has been brought up. The main thought of this method is to create a target-driven mechanism, which builds belief, desire and intention for migrating instance (i.e. agent) based on initial state, target state and services context of the workflow. According to workflow target and knowledge it has, the instance can apperceive the workflow context build belief in a self-adaptive way. And the instance builds desire and intention to change the workflow state. By executing the actions as a result of the intention, the instance accomplishes the service detection, service selection, migrating decision and workflow execution.The concepts of belief, desire and intention come from BDI theories about building rational agents in Artificial Intelligence area. The BDI architecture is a representative for deliberative agents. It thinks an agent as a consciousness system to take action actively to achieve its goal. Belief is the agent's knowledge about environment and estimation about its own ability. Belief is the information component representing the basic knowledge and ability of an agent. Desire is the expectation an agent has about future. It can be a state the agent wants to become or hold, and can be created from outside or inside by reasoning. Intention is a commitment an agent makes to take actions and accomplish its goals. It drives the agent's actions directly.Base on BDI theories, this paper makes further research on target-oriented migrating workflow. According to traditional migrating workflow system, this paper proposes a target tree decomposing method to present workflow state. This method is to use ground sets of literals to represent the state of workflow, which as a result can be brought into the belief, desire and intention of the migrating instance. In this way, the instance measures the difference between target state and current state and chooses state transferring strategy. Then, a concept model for target-oriented migrating workflow system and definitions of its components are brought up. Also, the structure and framework of migrating instance based on BDI architecture is discussed. This paper proposes time tree with workflow state to be used to represent the instance's belief, desire and intention. The BDI model for migrating instance and functions in it are discussed then. As bringing in BDI components will surely increase the instance's scale, a separate and synergic architecture splitting the instance into decision-agent and execution-agent is proposed to solve the problem...