基于知识水平的改进智能遗传组卷算法设计

Improved Intelligent Genetic Generating Test Paper Algorithm Design Based on Level of Knowledge

针对信息化发展中在线试卷的组卷工作中存在的问题,诸如如何让考试的试题更好地检验学生的知识水平,怎样考察学生掌握和未掌握的知识等问题,探索提出了一种自适应的组卷方法,把学生个性化信息引入其中,采用期望的试卷难度、区分度作为约束条件,将从试题库选择的试题子集的难度和区分度值与期望的难度和区分度的差作为目标函数,从而提出一种个性化信息遗传组卷算法(Personalized Information Genetic Algorithm,PI-GA)。测验结果证明,在生成试卷的时候,PI-GA算法可以有效地为学生提供个性化试卷,对比几种常见的算法,执行时间最短,并且所组成的最终试卷中包含的学生未掌握试题数量具有灵活性。

For the existing problems in online generating test paper work of information technology development, such as how to make the examination questions test the students＇ knowledge level better, how to examine students＇ grasp of knowledge and other issues. This paper explores an adaptive test paper generation method, introduces the customers＇ personalized information, uses the expected paper difficulty and discrimination as constraints, uses the difference between the difficulty and discrimination value of the question subset which is selected from the item bank and the expected difficulty and discrimination as an objective function, so a personalized information genetic algorithmic（PI-GA） is proposed. Test results show that, when generating papers, the PI-GA algorithm could provide students with personalized papers, and has a shortest execution time compared to several common algorithms, and the number of questions that are not mastered in the formed final papers is flexible.