Improvement and Design of Genetic Algorithm in Personalized Test Paper Composition System

Abstract

Based on the question of test papers in personalized learning, this paper makes special improvements and designs for the individual genotype, selection, crossover, and mutation processes in traditional genetic algorithms. At the same time, in the design of the fitness function, based on the disadvantages that the dimension cannot be unified when calculating the fitness function by linear weighting method in the traditional literature, a vector distance calculation method was selected to calculate the objective function, which solved the unification of different constraints Questions that differ between dimensions. In addition, based on the problem that duplicate questions may appear in one test paper, this paper designs a deduplication operator and adds it to the step of genetic algorithm.