A test paper generation algorithm based on diseased enhanced genetic algorithm

• Published in: Heliyon Vol. 9; no. 6; p. e17187
• Main Authors: Cui, JunChuan, Zhou, Ya, Huang, Guimin
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.06.2023; Elsevier
• Subjects: algorithms; Applications in subject areas; data collection; Distance education and online learning; Improving classroom teaching; mutation; probability; Applications in subject areas; Improving classroom teaching; Distance education and online learning
• ISSN: 2405-8440; 2405-8440
• DOI: 10.1016/j.heliyon.2023.e17187

With the continuous progress of society, tests, and exams appear more and more frequently in people's lives. Faced with the ever-increasing demand for test papers, efficient test paper generation algorithms have become more important. In this paper, we improved and proposed a Diseased Enhanced Genetic Algorithm (DEGA) based on the Genetic Algorithm (GA), and applied it to the test paper generation algorithm. I the crossover operator, the crossover probability that will change in different situations of the population is adopted. According to the characteristics of the test paper generation algorithm, we use the method based on the hamming distance to calculate the distance between individuals in the population. Aiming at the shortcoming that the mutation operator is too random, we designed and used a disease operator that includes three modules: natural disease, infection, and mutation. It effectively guarantees the distance between individuals in the population and improves the shortcoming that GA is easy to fall into a locally optimal solution. Finally, using the College English Test Band 4 (CET-4) questions from 2014 to 2021 as the data set, comparative experiments were carried out on the test paper generation algorithm based on Random Sampling Algorithm (RSA), GA, Enhanced Genetic Algorithm (EGA) and DEGA. The results show that when using the test paper generation algorithm based on DEGA, the generation of test papers is faster, the number of iterations is less, and the algorithm results are significantly better than other algorithms.