RAEDSS:یک سیستم پشتیبانی تصمیم گیری یکپارچه برای اقتصاد کشاورزی منطقه ای در چین

Recent advances in artificial intelligence, particularly in the field of multi-agent system theory and techniques, offer great promises in the development of decision support systems. This paper designs an agent-based regional agricultural economy decision support system (RAEDSS) to deal with complex decision problems. It introduces the architecture of the system, including interface agents, management agents, functional agents, model agents, information agents and knowledge agents and their interactions. Since dynamic analysis, evaluation, forecast, optimization and decision of regional agricultural economy are the central task of RAEDSS, this paper gives a detailed discussion on the decision processes and internal mechanisms in the system. Meanwhile, agent-based modeling is introduced to simulate and evaluate policy impact on rural development in different scenarios as an important part in RAEDSS. The simulation result shows that this agent-based agricultural development model is able to perform regeneration and is able to produce likely-to-occur projections of reality. The related issue such as building an agent based on the theory of classifier systems is also surveyed.

In recent years, more and more artificial intelligence technologies have been incorporated into decision support system (DSS) design frameworks in order to request intelligent problem-solving mechanisms and improve the decision-making processes and therefore to obtain a more powerful decision support. In addition, the development of distributed artificial intelligence frameworks provides a new methodology to solve problems for complex systems by dividing them into a number of agents which can cooperate to solve the problem using their own knowledge, goals, skills and plans. To improve the mechanism of problem solving more and more artificial intelligence technologies and, especially multi-agent systems (MASs) are being applied in the exploitation of DSS. Research into agents and the development of related technology [1], [2] and [3] has grown dramatically since the 1980s as domains suitable for their applications have emerged. Multi-agent systems (MASs) have been applied broadly to a variety of industrial problems [4], from electronic commerce [5] and [6], to supply chain management [7]. However, most multi-agent applications are still in the early phases of development and few standards have been widely accepted. Bentham [8] has developed an agent-based system to assist with decisions related to crop production, and [9] have created a multi-agent expert system to aid farmers with the selection of appropriate hybrids. Berger et al. proposed multi-agent systems as a modeling approach well suited for capturing the complexity of constraints as well as the diversity in which they appear at the farm household level, and illustrated how they may assist policymakers in prioritizing and targeting alternative policy interventions especially in less-favored areas [10]. Lobianco and Esposti [11] developed the Regional Multi-Agent Simulator (RegMAS) for long-term simulations of the effects of policies on agricultural systems. It is an open-source spatially explicit multi-agent model framework. In the dairy industry, a simple multi-agent system for heifer management was then developed as a proof of concept, in order to explore practical aspects related to an actual implementation of the technology [12]. This paper presents a number of perspectives on our ongoing work in developing agent-based regional agricultural economy decision-support system (RAEDSS), which is a decision-support framework that can be of use in supplying decision-enabling information in a number of economic domains for regional management. The paper is organized as follows. Section 2 discusses the basic concept of agents and multi-agent system. Section 3 presents the framework of the proposed agent-based decision support system. Section 4 introduces the developing system and mechanism. Section 5 gives a detailed discussion of agent-based modeling in policy simulation of rural and agricultural development in different scenarios. Section 6 concludes the paper.

Decision-making processes in regional systems usually very complex and are frequently partitioned into sub-problems. Very often, this decomposition involves several levels of decision and is by nature hierarchical. Solutions are proposed for each sub-problem, either by regional experts who work individually or by a group who analyze and solve a problem collectively. They deal with the problems based on multiple and complex interactions between the actors of the process. One of the main problems is to find a way to automate the process as much as possible, particularly so as to obtain automatically coherence and coordination among decision made locally by different actors at different levels. In this paper, we present an agent-based framework to design decision support system for management of regional agricultural economy. The flexibility and concurrent character of this approach is very promising as a way of designing and implementing a successful design support system. The decomposition of the problem into a number of modules enables the processing of more complex problems by the cooperative effort of agents using their own knowledge, goals, skills and plans. Multi-agent system provides an adequate structure to represent multiple and complex interactions between cognitive agents, which are justified by the participation of diverse knowledge sources and decision-centers in the process of regional economic decision. In fact, the agent technology has been identified as being complex and lacking maturity, since there are no firmly established methods by which to create a multi-agent system. Meanwhile, the agent-based model shows that we can simulate rural and agricultural behaviors of heterogeneous individual decision-makers to study overall characteristics of agricultural development. It means that through observing reciprocity of micro individuals, we can well research macro evolution of an entire region, and provide a solid platform and foundation for the future research on interaction process of agricultural development and urbanization. Our micro-simulation to rural problems in RAEDSS is just at the beginning, but the future is promising. A major challenge in developing such system is to build a success coordination mechanism for agents to solve a problem collectively.