Modeling And Analysis For Agent-Oriented Requirement Engineering

In computer science research, Requirement Engineering and Agent-oriented Software Design are both hot topics. As a more abstract modeling technology, Agent poses a challenge to the traditional software requirement analyzing method, yet meanwhile it brings about improvement. In traditional requirement analyzing technology (object-oriented and structure-oriented), the requirement analyzing step and the program design step are often out of joint. The "requirement" of users seems to developers an abstract definition of product; the "requirement" of developers seems to users detail technology specification. As the communicative bridge between users and developers, the requirement document must describe "requirement" on a fairly abstract level. Such document is often roughly built and therefore cannot reflect the system requirement on the basic object level. Since the detail design step following the requirement analysis can not make use of its results, a great deal of detail design is needed. Due to the highly abstract nature of Agent itself, Agent-oriented program design agrees with the abstract description in the requirement analysis. Therefore the Agent-based requirement analyzing step and the detail design step can be combined seamlessly and supplement each other. The Agent-oriented requirement analysis aims to acquire the system basic component (the agent object), and analyze the system-service-required cooperation. The results of the requirement analysis will become the analysis object of the following Agent-oriented detail program design step.This thesis focuses on Agent-oriented requirement modeling and analysis technology. Considering Agent's automatic features, the model can hardly reflect Agent's social features. It's difficult to describe its social features from the technological angel. As a social object, Agent is different from other program design object. It should express complicated ideas and feelings, evaluate the surrounding environment and state, and then take actions. This means Agent should describe certain social concepts and relations in an uncertain way. As a matter of fact, the analysis and description of uncertain problems is a core content of the thesis. In this aspect a lot of research is done and the theory of analyzing and modeling uncertain problems is put forward. Used in the Agent-Oriented Requirement Engineering, the theory has become a dominant method to describe uncertain problems.Another important issue in Agent-oriented requirement analysis is the requirement conflict. Different from other analyzing methods, Agent-oriented requirement analysis must deal with a great deal of uncertain requirement conflicts. Some of the conflicts are non-functional requirement conflicts. Others are uncertain functional requirement conflicts. Appling the theory of analyzing and modeling uncertain problems, those new methods are successfully used to solve the conflict in requirement analysis. Compared with traditional precise way of analyzing conflicts, the method of analyzing uncertain requirement conflicts can figure out the root cause of requirement conflicts and work out feasible strategies to solve conflicts.Such new method takes into consideration both users' uncertain functional requirements conflicts and non-functional requirements conflicts.Most of today's Agent-oriented requirement analysis methods emphasize only the construction of Agent interaction model and overlook "how" to acquire these Agents and "why" these agents are needed. While incorporating the goal-oriented KAOS method, this thesis puts forward an Agent-centered requirement analysis method-FKAOS, in which most energy is spent to acquire and optimize the Agent object. Using goal-oriented method, we can acquire some rough Agent objects, which, however, cannot fully reflect Agent's features. Through asking "why" these agents are needed, this thesis analyzes these rough Agent objects and optimize Agent design. The purpose of optimizing Agent objects is to build a cooperative social system, in which Agent is moderate sized,...