Abstract: Multi-constraint constrains and reduces the test success rate, and it is difficult to make the knowledge points uniformly and automatically distributed in intelligent generating test paper. To solve these above problems, a multi-constraint hierarchical optimization strategy was proposed. It used hierarchical method to reduce the problem size, and used the tree structure to manage knowledge points and realize uniform distribution of knowledge points. With regard to the low success rate and efficiency of small test bank in generating test paper, a forecast calculation algorithm without backtracking was put forward based on hierarchical optimization algorithm to increase the test success rate. The experimental results indicate that the algorithm is suitable for large, medium and small question database, and all of them have good results.

Key words: intelligent test paper generating, multi-constraint, hierarchical optimization, algorithm without backtracking, knowledge uniform distribution

摘要： 针对智能组卷中多约束制约降低组卷成功率且难以实现知识点自动均匀分布的问题，提出一种多约束分级寻优的策略，通过分级降低问题规模，利用树形结构管理知识点实现知识点均匀分布;针对中小型题库组卷成功率低的问题，在分级寻优中针对章节约束和题型约束提出了一种基于预测计算的无回溯的智能组卷算法，提高组卷成功率。实验测试表明，算法适用于大、中、小型题库，均能得到较理想的组卷结果。

关键词: 智能组卷, 多约束, 分级寻优, 无回溯算法, 知识点均匀分布