智能组卷中的试题难度评价建模

An Evaluation Model for the Item Difficulty Coefficient in Intelligent Test Paper Composition

主要研究了网络考试系统设计中利用遗传算法进行智能组卷的问题.首先根据用户对组卷的要求,利用遗传算法对组卷所需的约束条件设置各项指标,利用矩阵理论建立有利于遗传算子进行组合交叉和变异的数学模型;再针对组卷问题中的一个重要约束条件——试卷的难度指标,采用模糊数学方法和项目反应理论对试题库中每一小题进行综合评价试题难度的数学建模,以准确确定每道小题的试题难度系数,最终为实现遗传算法全局寻优和智能搜索奠定基础.

The present paper mainly studies the intelligent test composition by using genetic algorithms in network test system design. In the first place, the indicators are set according to user requirements on the test paper and the constraints by using genetic algorithm to carry out the test paper, using matrix theory to create the mathematical model that enabled genetic operator to crossover combination and mutation, and then one of the major constraints the difficulty of indicators for Test Paper is constructed, using fuzzy comprehen sive evaluation of the mathematical methods and item response theory to each question in the test database to evaluate difficulty of mathematical modeling in order to accurately decide the difficulty coefficient of each item, and ultimately lay the foundation for the realization of the genetic algorithm global optimization and in telligent search.