The Investigation And Realization Of Genetic Algorithm Based Examination Paper Auto-generation System Of Internet-based Examination

Examination paper auto-generation is to generate a examination paper with specific target for scores,time of the test,difficulties of questions, distinction, knowledge, types, cognition levels and so on in according to certain requirements by automatically selected questions from a database. Examination paper auto-generation is one of the most important components of Computer Managed Instruction (CMI). Several algorithms have been developed in the field, such as priority strategy, randomization, trace and trial strategy, and genetic algorithm (GA) etc. When encountering with a big solution space or multimodal problems, these algorithms are usually inclined to run into local extremes or difficult to solve. Since generated examination paper needs to fulfill variable demands while keeping randomness and rationality, a more effective algorithm is in great demand. The thesis is to develop an improved genetic algorithm, and apply the algorithm in the examination paper auto-generation problem. The main work in this paper is based on the fundamental principles of examination paper generation. Analyses of evaluating factor, i.e., their functions and relationships were made in details. A examination paper generation mode is developed with the distributions of the evaluating factor. Finally, a mathematic modal to generate examnation paper was established upon the examination paper's quality defined by the generation mode.A new adaptive genetic algorithm is developed and investigated. Adaptivity is introduced according to the variety of population fitness distribution to deal with the pre maturity and the low convergence speed of genetic algorithm. Several classical test functions for global optimization are adopted to validate the effectiveness. The test results show that the algorithm can improve global optimization and convergence speed, and the algorithm is quite robust.According to the characteristics of examination paper auto-generation problem, proper coding proposal and adaptive function are introduced. As a simulation instance, the examination paper auto generation problem in the class of computer literature was applied. The results show that adaptive GA can be successfully applied in examination paper auto-generation system; of which the speed is fast, the quality is high, and which is insensitive to its initial value.Analyses and designs of online examination system were made in the current work. A user friendly online examination system was developed with the use of JSP.