【作者】 李慧姝；

【导师】 张月琴；

【作者基本信息】 太原理工大学 ， 计算机软件与理论， 2007， 硕士

【摘要】 遗传算法是一种通过模拟生物界自然选择和遗传变异的机制来求解复杂问题的随机搜索和优化的方法；组卷问题是一个在一定约束条件下的多目标参数优化问题。自动组卷的效率和质量很大程度上取决于试题库的设计以及抽题算法的设计。传统的组卷算法存在“组卷速度慢、成功率低、组卷质量不高”等缺点。针对上述缺点，本文在充分分析和研究考试系统组卷算法的基础上，提出了遗传算法自动组卷策略，并将遗传算法自动组卷应用于在线考试系统中，最大程度地满足了用户的需求，具有科学性、合理性和较好的实用性。本文提出的基于遗传算法的自动组卷系统是在一定的约束条件下，通过分析试卷的各项评价指标、指标的作用及其相互关系，将其量化为描述目标试卷特征的一组指标分数分布列，即建立了采用各个评价指标的分布构建的成卷模式；在成卷模式的基础上建立最接近用户要求的多目标约束优化模型，并对此模型进行求解，利用权重系数法将多目标优化转化为单目标优化，建立了组卷数学模型。进而利用遗传算法对其进行求解，解集即为目标试卷中具有相同试题指标取值的试题数量。本文对传统遗传算法进行了适当的改进，利用C#编程将改进的遗传算法应用于自动组卷系统中，使组卷的成功率和收敛速度都得到明显提高，不仅适合于较大型题库系统，而且对于复杂的组卷条件也能满足，一次组卷还可以得到多份符合要求的试卷，并通过在线考试系统的实现，验证了所提理论的正确性。在线考试系统是随着计算机网络技术在教育领域的广泛应用而提出的，利用网络进行考试可以极大地提高考试的效率和质量。本文将遗传算法自动组卷应用其中，并从用户的实际需求出发，对在线考试系统应具备的主要功能进行了详细论述，分析了基于WEB的在线考试系统模型，运用面向对象的方法进行在线考试系统的设计和实现，具体运用UML模型图对在线考试系统进行了分析和设计。构建了一个基于B／S结构的，以ASP．NET技术和SQL Server 2000后台数据库为基础的，具备“用户管理、试题库管理、试卷管理、自动组卷、在线考试、试卷评阅、成绩统计”等多功能的在线考试系统。具有重要的理论意义和现实意义。更多还原

【Abstract】 Genetic algorithm(GA) is random search and optimizing method what is used to solve complex problems by simulates the natural selection in biosystem and the mechanism of aberrance in heredity. And the problem of composing examination papers is multi-objective parameter optimizing problem with some constraint conditions. The quality and efficiency of auto-generating examination papers are all determined by the test questions-database designs and get problems-terms algorithm. The conventional composing examination papers algorithms have some drawbacks such as slow speed, small success chance and poor quality. This paper, which aims at the drawbacks of auto-generating examination papers in examination system, completely analyzes and investigates the algorithm of auto-generating examination papers and puts forward the strategy of auto-generating examination papers based on GA, applies GA into the auto-generating examination papers in on-line examination system. The method meets the users’ extreme need because of scientificity, rationality and practicability.. This paper puts forward the auto-generating examination papers based on GA under some constraint conditions, analyses evaluating indicators, their functions and their relationship; evaluating indicators are quantified to a group of discrete indexes’s distributive columns, which are used to describe the characters of objective examination papers, namely a examination paper generation model is set up with the distributions of the evaluating indicators. Based on present examination paper model, a multi-objective optimization model, which seems much satisfying users’ demands, is established. In the process of solving the problem, weight coefficient method is introduced to turn multi-objective optimization problem to single-objective optimization, a mathematic model to generate examination papers is established. Then we find the solution by the use of genetic algorithm. The set of solution is just the quantity of test items that has the index value in objective examination papers.In this thesis, the traditional genetic algorithm has been improved appropriatly, and the author makes an improved genetic algorithm program of auto-generating examination papers using C#, the success efficiency and constringency speed have been boosted distinctly. This not only adapts to big database system, but also satisfies complex restricted conditions of generating examination papers. Many examination papers will been gained which accord with generating examination papers request by one-off generating. Through implementing the on-line examination system, the correctness of the theory has been validated. With the extensive application of network technique in the educating realm, the on-line examination system has been put forward. To test with the network can improve the efficiency and quality of the examination. This paper applies the auto-generating examination papers system based on GA to on-line examination system, and embarks from the users’ actual demands, gives a detailed elaboration on the main function of the on-line examination system and analyses on-line examination system model based on the WEB. This paper analyses how to design and implement the examination system by using the object-oriented method, specially the UML model diagram analyzing designing the on-line examination system. The on-line examination system bases on B/S structure, the ASP.NET technology and SQL-Server 2000 database, and the author developes a practical on-line examination system as to meet the requirements, such as the management of user’s information, the management of test questions database, the management of examination papers, the auto-generating examination papers, on-line examination, marking examination papers, analyzing test score etc. It has important theory and reality meanings.更多还原