导　　师： 党小超

学科专业： 081203

授予学位： 硕士

作　　者： ;

机构地区： 西北师范大学

摘　　要： 本论文基于JAVA/XML的三级网络教学平台下网络考试系统的研究与实现,作为该网络教学平台中一个重要部分,从探讨网络考试的根本问题入手来展开研究,结合当前流行的网络开发技术和软件工程方法,构造网络考试系统模型;采用网络考试系统的相关测试理论,结合题库的设计原则,构建数据库;采用二项分布函数控制试卷难度比例;运用遗传算法解决组卷问题,给出了一种改进的遗传组卷算法,设计了基于题型块的编码方案和基于难度及考查点的适应度函数,结合自适应的思想,在算法的运行过程中对交叉概率PC和变异概率PM进行调整.依据该算法组出的试卷基本已经满足了题型、分数和答题时间的要求,试卷的实际平均难度和实际考查点覆盖率与用户的要求差距不到5％.该组卷算法对今后组卷系统的开发设计会有一定的借鉴作用.最后本文给出了对考试过程中可能出现的特殊情况的处理方案,提出了利用模糊理论来实现主观试题自动批改评分的构想,也对多媒体试题的可控性提出了自己的看法,并对下一阶段研究工作的目标做出了展望. With the rapid development of network technology , education techniques have got great promotion. This accelerates the emergence of network education, which offers a good opportunity for the long-distance training of high quality for those in - service teachers who are working in poor counties and rural communities in Gansu province. Considering the educational and economic conditions of such locations, we have built a network teaching platform, consisting of three resource centers - province, county and town.As we know, teaching evaluation is also an essential aspect of education besides gaining knowledge. In most schools, the examination is used as a chief means of deciding whether a student succeeds or fails. The normality and scientificity of examination, as well as organization and management, all directly exert influence on the accuracy and objectivity of examination results, thus will affect the teaching quality.This paper, Research and Implement of Network Examination System under Three-level Network Teaching Platform Based on JAVA//XML, which is an important part of the platform, begins with analyzing basic problems of network examination system. Combining the current popular network developing techniques and methods of software engineering, we construct the network examination system model. By adopting some related test theories and incorporating the design principles of database, we set up the item bank. We use binomial distribution function controlling paper's difficulty ratio. Since the auto-generating examination paper is a constrained multi-object optimization problem, this paper presents a new approach based on genetic algorithm to solve it. We define the crossover operator and mutation operator in the real coded and adjust the probabilities of crossover and mutation adaptively. The main parameters in genetic algorithm can then be adjusted automatically according to individual fitness. Experiments show that this kind of genetic algorithm could solve auto-generating paper problems more effectively than other methods with a better performance and practicability.The auto-generating algorithm based on adaptive GA will benefit those who are going to develop examination system.In the end of this paper, we put forward processing methods of special circumstance possibly emerging during the exam, and the conception of applying the misty theories to correcting subjective test question automatically. Meanwhile, it shows my own viewpoint about the multimedia test question , puts forward the problems that need betterment and extension, and points out the direction for further research.In addition, we have obtained the periodical achievements in the mathematics model , architectures, and auto-generating algorithm of examination system. This network teaching platform has already stepped into a stage of testing the overall function.

关 键 词： 网络教学平台 考试系统 组卷算法 遗传算法

分 类 号： [TP399 G434]

领　　域： [自动化与计算机技术] [自动化与计算机技术] [文化科学] [文化科学]