Application Method Research Of Genetic Algorithms In Automatic Generating Test Paper

With the popularization of computer, and the development of computer technology, many tests uses the computer setting the questions, and with the arrival of the period of the network and developing rapidly, in-line test with the computer already become one irreversible tendency. The examination systems get the examination paper automatically through choosing questions of the examination database. Compares with the traditional hand work test paper, the computer-aided examination questions can reflecting the teaching the actual result more objective, really and comprehensively, it is advantageous to the promotion the enhancement of teaching quality. Therefore, the development of network test system has the vital practical significance.This article firstly introduced the computing group paper technology development process. In the domestic and foreign research present situation, it analyzed and contrasted the present four kind of automatic group paper strategies. And based on this, it proposed the reason that this topic must select the genetic algorithm to complete the automatic group paper. Secondly, through the summary to the topic actual system request and the automatic group paper's research background, the multi-analysis test question's group paper target, the examination paper target, the group paper pattern and the relations during each target, and it has established the group paper mathematical model according to each evaluating indicator's distribution, and it carried on simplified processing to the model, and mapped the group paper question space to the genetic algorithm space, used the genetic algorithm to solution actual group paper question. Importantly, in order to meet the actual design need of automatic group paper system, the topic improved the traditional genetic algorithm.The genetic algorithm has the auto-adapted overall situation optimization and the intelligence search technique, and the astringent good characteristic can satisfy the request of the automatic group paper well. In the present computer automatic group paper algorithm's research, many researchers have launched the massive research using the genetic algorithm in the computer automatic group paper method's application. After This article the multi-analysis genetic algorithm's basic concept and realized the technology, and according to the actual request of the project "Network Question bank Management System" to the automatic group paper, it has made the following improvement to the genetic algorithm: The code scheme used the partition real number code; it Used the condition to produce, and the total initial population which certain subspaces merge; The method of the index transformation of scale computed sufficiency function, and it may improve in the individual sufficiency distributed condition of the population; The improvement roulette selection algorithm has manifested the random number function fully; The union of microhabitat technology and the determination of iteration terminal condition, they enabled the algorithm to be able to retain the optimal solution during guarantee multiple, for obtaining the most superior combination of examination paper. Finally, it will the improvement of the genetic algorithm applied in the topic automatic group paper research.Finally, this article has carried on the system analysis and the design based on the improvement genetic algorithm's automatic group paper algorithm, and using the Java language, Eclipse as the developed platform, the MYSQL database to realize the system development and the algorithm test, and conformity to "Network Question bank Management System". The sixth chapter of the article has given the main algorithm of system.