Abstract:

The generation of test papers is an optimized problem with multiobjective parameters under a certain restrictive condition. The optimization is implemented very difficultly by traditional methods. The quality and efficiency of autogeneration is determined by the design of test question databases and algorithms to extract questions. Based on the construction of the provincial excellent course Data Structure, an analysis of the features of traditional test paper algorithms and the parameters of strategy, this paper presents the design and realization of an automatic test paper generation system based on genetic algorithms. This algorithm searches for the best answer according to such restrictive conditions as test question types, quantity, difficulty level, difference level, score and answering time. In addition, the natural code is used in this algorithm in order to decrease the space of chromosomes. The crossover probability and the mutation probability are improved with the selfadaptation theory, so that the proper numbers of crossover probability and mutation probability can be found. After the accomplishment based on C#.NET, the system has been applied in practice and achieved good effect.

Key words: automatic test paper generation;genetic algorithms;item bank;data structure