Studies in exam paper generating based on the genetic algorithm
The algorithmic workflow is presented and the new algorithm is applied to automatically generate tests with the question database of the undergraduate English course at Zibo Vocational College to alleviate the problem of premature convergence significantly.
Abstract
In order to overcome the problem of premature convergence in genetic algorithms for automatic test generation,a local hill climbing approach is employed to adaptively tune the dynamic parameters in genetic algorithms.We present the algorithmic workflow and apply the new algorithm to automatically generate tests with the question database of the undergraduate English course at Zibo Vocational College.Experimental results show that the new algorithm indeed alleviates the problem of premature convergence significantly.