改进遗传算法对医学题库组卷问题的应用研究

Research of generating test paper for medicine question bank by improved genetic algorithm

对含分支题的医学题库进行研究,并对遗传算法做了改进,提出了占位符编码方案、扩位交叉算子和重题优化策略。占位符编码方案能分段定长编码的同时累计各题型段的实际分支题量;扩位交叉算子能智能扩展落在分支题段的交叉点,避免因分支题段局部交叉而出现重题和实际分支题量与条件不符等情况;重题优化策略能快速替换重题,有效缩短组卷时间。仿真结果表明,改进的算法能适应不同题型,在不影响一般题型段抽取与进化的同时,精确控制分支题段的总分支题量和质量,是解决医学题库智能组卷问题的一种有效途径。

In this paper, an improved genetic algorithm is proposed for medicine question bank which includes problems with branches. The improved algorithm contains three parts; they are placeholder coding method, extended crossover operator and iterant problems optimization strategy. The placeholder coding method codes the individual into subsections according by the type and the required number of branches of questions. The extended crossover operator can extend the crossover points which locating in the subsections, thus avoiding bringing iterant questions and will not lead to the discrepancy between the actual number of branches and the required one. The iterant problems optimization strategy can replace iterant problems of test paper quickly. Simulation shows that this improved algorithm can applies to questions of all kinds of types, especially for questions with branches, for it can pre- cisely control the number of branches of subsection of this type of questions, and what＇s more important, it＇s au effective tech- nique of generating test paper for medicine question bank.