It is very difficult to solve the problem about the intelligent test paper composition in the examination database with traditional mathematical method. With the quick development and wide applications of computer aided design, nowadays, the research on intelligent test paper composition in the examination database is mainly on the design of examination database and the design of the algorithm of composing test paper. The quality of the design of examination database and the efficiency of the design of composing test paper algorithm will determine the ultimate effect of automatic test paper.On the basis and research of analyzing many foreign relevant scientific documents, firstly, this paper studies the index system and the test paper model, and then analyses the shortcomings and deficiencies of current commonly used composing test paper algorithm. Secondly, according to the actual teaching requirements, the thesis designs and implements a new evaluation criterion restraint algorithm. It designs a algorithm of composing test paper based on improved self-adaption genetic algorithm. Both the theoretical analysis and the experimental comparison show that improved genetic algorithm can satisfy the needs for actual examinations. Improved genetic algorithm can obviously improve the ability of global research, the convergence speed and the quality than tradition genetic algorithm and self-adaption genetic algorithm. It proves that the improved algorithm is effective and superior. Finally, we detail the design and implementation processes of the item bank system based on the new evaluation criterion restraint algorithm, we test the main functions of the system and the composing test paper algorithm, test results show that the system has good practicability. |