The Research Of Intelligent Test Paper Auto-generation Based On An Improved Genetic Algorithm

Intelligent test paper is a multi-objective optimization problem.First, this thesis analyzes the evaluation indexes and their effects of test paper, and construction of a paper model with the evaluation indexes.Then,this thesis analysises and designs of genetic algorithm for intelligent test paper system's own characteristics. Through using the character encoding method, reduces the length of chromosome and simplifies the problem solving;for the change of fitness little,this use remainder stochastic sampling with replacement and adaptive crossover operator and adaptive mutation operator. This thesis selects of the fitness function with a penalty factor on the relative error weighting method,improves the accuracy of the fitness function. This references allele genetics knowledge,proposes a new method to duplicate titles and come off satisfactorily.lastly, this thesis studies for verifying the said course of actions. The results indicate that:this improved algorithm achieved a good test paper, and the efficiency and quality of the paper is obviously better than traditional genetic algorithm's.