An Integrated Software Quality Evaluation Framework for Agent-Based Systems

• Main Authors: Lin Chih-Te, 林致得
• Other Authors: 王素華
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/21214765747350290035

碩士 === 中華大學 === 資訊管理學系 === 93 === Agent technology has become a new direction for software development. It is expected to radically alter the way that software systems are conceptualized and implemented. However, an agent-based system might not have the expected benefits if the software design is not really conformed to the agent characteristics. Therefore, evaluating the quality of agent-based systems also plays an important role. Because of the lack of agent metrics and the fact that many agent-based systems are implemented by object-oriented technology, the object-oriented metrics are used in this research. Compared with the traditional methodology, the agent technology has the unique characteristics such as autonomy and social ability. Therefore, the quality of agent-based software systems should be evaluated from different aspects. In this research, the design quality of agent-based systems is evaluated in different phases: quality of individual agents, quality of the agent society, and quality of the system as a whole. A multi-phase evaluation framework is then defined to integrate the evaluation of all different phases. A prototype system is built based on the proposed framework. Sample quality models are presented to demonstrate the evaluation model of each phase. An example of agent-based systems is also analyzed and evaluated with the integrated evaluation framework. Various multi-phase quality models for evaluating agent-based software can be created or modified through this system. The agent application developers or software quality assurance members can utilize this system to ensure and improve the software quality of agent-based systems.