Research On Intelligent Composing Test Paper Strategy Based On Genetic Algorithm

With the continuous progress of education reform and the rapid spread of computernetwork technology, online education with resource sharing, fast and convenient featureshas been developing rapidly. Among them, online exam has become a hot application in thefield of online education. It makes the exams break through the boundaries of time andspace, and exams change from previous papers exam into current online exam, so it ensuresthe fairness and impartiality of the examination. Strategy of composing test paper is the coretechnology of online exam. However, the present composing test paper strategy used mostlywith a slower speed, low efficiency is difficult to achieve the actual needs of users.Therefore, how to design a composing test paper strategy, which is a scientific, rational, andcan effectively meet the actual needs of users, is practical significance.Genetic algorithm that has the characteristics of simplicity of operation, fastconvergence rate, strong robustness and so on, is suitable to solve the problem ofcomposing test paper. Therefore, this paper researches the traditional genetic algorithmin-depth, summarizes the advantages and disadvantages of genetic algorithm, systematicallyanalyzes the causes of the early convergence phenomenon and concludes that the effectivesolution to the balance of the early convergence phenomenon lies to maintain poluationdiversity and convergence speed. Combining with the characteristics of composing testpaper problem, the paper puts forward the dynamic genetic algorithm based on speciesinvasion SIDGA, which creates the indicators of degree of premature convergence. Basedon the above, the algorithm improves the traditional genetic algorithm at both the micro andmacro level. At the micro level, it adopts the dynamic automatically adaptive crossover andmutation probability in order to maintain population diversity; At the macroscopic level,according to the principle of species invasion in biology, the algorithm randomly joins acertain number of individuals to replace the poorer individuals of the current new group atthe time of the algorithm tending to convergence in order to increases the diversity of thepopulation.The paper analyzes the basic attribute of test questions, evaluation criteria of testpaper’s quality and the proposition principle. Considering the test scores, time, the difficulty and the knowledge points, differentiate and other important factors, the paper confirmsimportant constraints on the test paper and proposes a mathematical model of intelligentcomposing test paper strategy and the objective function.Combining with target function of composing test paper problem, the paper appliesSIDGA to the composing test paper problem and concretely realizes the intelligentcomposing test paper strategy on the basis of SIDGA. Simulation experiments demonstratethat the intelligent composing test paper strategy can quickly and effectively generate testpapers, which can satisfy the practical needs of users.Finally, through analyzing the requirements, the paper develops the online examsystem under the environment of J2EE and applies the composing test paper strategy basedon SIDGA to online exam system. It shows that the composing test paper strategy based onSIDGA meets the practical needs of users efficiently.