University English Test Paper Composition Based on Multi-Strategy Improved Genetic Algorithm

As the computer technology continuously develops, the test paper composition using corresponding technology has been widely concerned. However, the existing test paper composition method has some problems such as slow test composition speed and poor quality. Therefore, this paper proposes a college English test composition method grounded on multi-strategy improved genetic algorithm. In this method, the fitness function of genetic algorithm is improved by using multi-constraint and multi-objective mathematical model, and the search speed of the population in the early iteration and the search precision in the later iteration are improved by using adaptive crossover and disease variation. Experiments denote that the maximum fitness value of the improved genetic algorithm is $11.5,16.3$, and 17.8 higher than other algorithms, and the success rate of test paper formation is $10.8 \%, 15.1 \%$, and $19.5 \%$ higher than other algorithms, respectively. The average test paper formation time is 15.2s. The question matching degree and test validity of the improved algorithm are $12.82 \%$ and $9.56 \%$ higher than that of the particle swarm optimization algorithm respectively, and the test paper differentiation is 8.30, which can effectively test the learning level of students. The difficulty coefficient of the test paper is 5.80, and the difficulty level is moderate. It can be concluded that the research can effectively raise the efficiency and quality of automatic test composition, reduce the workload of teachers, and meet the needs of different examinations.