The Application And Research Of The Paper-Composed Strategy Based On Genetic Particle Swarm Optimization Algorithm

With the development of the technology of computer and database, the network testing system gradually becomes a hotspot for research in recent years.Among this, the auto-composing test paper module is the core part of network testing system, its algorithm performance directly determines the quality of the whole examination system. However, according to the problems in the composing test paper system at present:the combination of questions is simple, the distribution of difficulty and discrimination index is not reasonable and no feedback functions and so on, searching for a suitable and efficient auto-composing algorithm is provided with important value and practical significance.This thesis respectively studies the current research status of the intelligent composing test paper, genetic algorithm, the particle swarm optimization algorithm and their application in the test paper, and analyzes the lack of the current system and the auto-composing test paper algorithms, and proposes combining GA and PSO as the algorithm which selects questions from database. Through analyzing the function of the item bank index and the relationship between several important indicators, constructing the model of generating papers and the objective function. Through the research of the characteristics of genetic algorithm and particle swarm optimization algorithm, combine the two algorithms together from the complementary perspective, the new algorithm mainly based on genetic operation, initialized groups using particle swarm optimization, proposed the individual maximum and global minimum of PSO for dynamically balancing the global optimization and local search in the process of group evolution. According to the features of problems for generating papers and hybrid algorithm, describing the specific steps that the genetic particle swarm optimization algorithm is applied in the test paper in detail, including that encoding, the determination of fitness function, the genetic operation etc.Analyzing paper from three aspects:results statistical analysis, test questions quality analysis and test quality analysis provide feedback effect for auto-composing paper.At last, the thesis makes a needs analysis, designs the structure of whole system, main function modules and databases in detail, and complete the development of the auto-composing test paper system. Examination paper analysis and experiments show that the hybrid algorithm applied to auto-composing paper to solve the problem that the distribution is unreasonable about the indicator of knowledge, difficulty and so on, to satisfy the need of stage test for teachers, has a great efficiency and better practicality.