Design And Implementation Of Automatic Test Papergenerating System Based On Genetic Algorithm

To overcome the low speed of paper generating, the low success rate of generating paper, and the low quality of the paper in the test paper generating system on the traditional algorithm, the paper proposes that the heredity in the genetic algorithm is trained to learned, and improves the fitness of the chromosome; that the number of questions in the target test paper are coded by the real number coding instead of the traditional binary code; and that the chromosome length of is optimized. By these, the algorithm premature and the slow convergence in the traditional genetic algorithm can be solved. Based on the improved genetic algorithm, the automatic test paper generation system is designed. The paper includes the followings:(1)To solve the premature convergence to local optimal solution which appears in the traditional algorithm, chromosome encoding in the traditional algorithm is improved and the training contents of BP algorithm and changed by combining the chromosome coding with the weight adjustment training in BP algorithm so that the IGA which can quickly search the optimal solution is found.(2)The genetic algorithm leads to the speed of the generating test paper slow since it can fall into the local minima easily. The overall searching ability of BP algorithm can solve the problem because the correct instances of these classes can be gotten, the suitable chromosome can be selected, iteration times can be reduced, the chromosome gene number can be shorten, and the chromosome length can be optimized by the learning ability of BP algorithm. Therefore, by the overall searching ability of BP algorithm, the speed of test paper generating can increased. Meanwhile, on the teaching contents and the testing requirements of Fundamental Application of Computer Technology in Xian yang Technical & Vocational College, the improved genetic algorithm can be applied into the test paper generating system of the course.(3)On the requirements of the system, the models of all users are designed, the cases are described with UML and the system is modeled and designed. with the previous demands and data analysis, the functions of the system models and the data ER diagram are designed, and the core function models are Implemented, which includes information management, chapter management, knowledge point management, item management, paper management, test paper comparison.(4)the IGA the automatic generating paper system with the B/S structure is designed on Java EE Eclipse 4.4+JDK 1.7 and MySQL5.5 data system.The results show the system can effectively generate test paper.