【作者】 陈自郁；

【导师】 吴中福；

【作者基本信息】 重庆大学 ， 计算机系统结构， 2002， 硕士

【摘要】 基于Web的远程教育是一种利用网络进行教学的新型网络应用，但是目前大多数基于Web的远程教学系统缺乏智能性和自适应性。而在众多新技术中，Agent技术尤其适合改善远程教学的不足。Agent是一个具有自主能力、交互能力、反应能力和预动能力的软件实体，能代表用户或其它程序，以主动服务的方式完成一组工作。使用Agent技术可以增加教学内容的趣味性和人性化色彩，改善教学效果，增强系统的智能性和自适应性。本文通过分析传统基于Web的远程教学系统模型的不足，提出了一个基于Agent的远程教学系统模型。此模型共分6个模块，每个模块由多个Agent组成的。利用Agent间的协作机制，实现各模块间的知识交换与共享，形成了一个层次结构的多代理系统；利用Agent的智能推理能力，学习学生的个性特点，自适应的生成一个适合学生的教学策略，智能地组织教学环节，引导学生更好的学习，发挥更大潜力。模型的设计思想既发挥了教师的主导作用，又充分体现了学生的认知主体作用。本文所提出的教学系统模型，以学生模型模块最为重要，它是其它模块正常运行的基础。然而目前大多数关于学生模型的研究要么复杂难以实现，要么简单有很少的推理能力，针对这种情况结合作者所参与的远程教育项目，本文提出一个多Agent的学生模型。这个模型包括学生四个方面的特点，每个方面的特点由一个Agent对它进行推理。作者通过改进Sherlock II方法，运用概念图、模糊理论，设计出学生模型中各Agent的不确定推理算法。这些推理方法，不仅有较好的推理能力，而且实现简单。并且本文所提出的方法具有通用性和扩展性。作者把学生模型运用到学生自测系统中，通过测试，得出了一系列表结果，这些结果显示了加入学生模型的自测系统具有自适应性并且能够比较准确地推理出学生的认知水平，证明了学生模型中算法的可行性和正确性。本文所提出的设计方案具有很强的实用价值。更多还原

【Abstract】 Web-based Distance Education, employing the network as its channel of learning and teaching, is a kind of brand-new Internet application. However, a majority of web-based distance tutoring systems are devoid of intelligence and self-adaptation. Among most of new techniques, Agent technique is particularly suitable for web-based distance tutoring system to improve its shortcomings. Agent, bearing autonomy, sociability, reactivity, and pre-activeness, is a software entity, and able to stand in for other users or programs to complete work on its own initiative. In a word, using Agent technique can increase the interest of teaching content and the color of personalization, ameliorate teaching effect and enhance intelligence and self-adaptation of tutoring system. After analyzing the deficiencies of traditional web-based distance tutoring system models, this paper proposes an agent-based distance tutoring system model. The model is made up of six modules and each module includes a few agents. Utilizing the cooperative mechanism among agents, the model realizes knowledge exchange and share among the modules and shapes a lay-structure multi-agent system. Utilizing agent’s intelligent reasoning ability, the model can learn characteristics of student’s personality, adaptively build a sort of teaching strategy which adapts to student, intelligently organize teaching process and lead student to study better and to exert more potential. In addition, the designing ideas of the model not only bring a leading role of teacher into play but also fully reflect cognitive main body of student. The key module in the model mentioned above is student model, which is basis for other modules to well work. However, now most of related researches are too complex to realize or easily implement but have little reasoning ability. Aiming at this kind of condition and combining the projects that the author has made in the field of distance education, the paper puts forward a multi-agent student model. The model involves the traits from four aspects of student and each aspect has an agent to reason its traits. By improving on Sherlock II method and adopting concept map and fuzzy theory, the author raises a new kind of reasoning way for every agent in the student model. These ways not only hold relatively good reasoning ability but also are easily realized. Moreover, the ways discussed in the paper are in possession of expansibility and can be generally used. By introducing the student model into student self-testing system, the author gets a<WP=6>series of tables which show that the testing system with the student model takes on self-adaptation and can exactly infer student’s cognitive level. And the tables demonstrate that the ways in the student model are feasible and correct. Therefore, the scheme discussed in the paper has strongly practical worthiness.更多还原