Study And Implementation Of Construction Strategies For Test Paper Based On Genetic Algorithm

With the rapid development of the computer technology and artificial intelligence, and the rapid improvement of Chinese educational information level, It becomes the hot focus in IT industry and education domain to achieve paperless examination and networking examination.With the development of the economy and society, and popularization of the lifelong educational concept, the item bank management system which is scientific, convenient and perfect is urgent needed. It is the powerful tool to accumulate items, to make test paper automatically, to save made-test paper, and to analyze test papers.The thesis, firstly, introduces the research background, the art of the state of the domestic and foreign research works, and main contents of this thesis.The theoretical basis of constructing item bank system, technique points of constructing item bank and the method of designing bank are discussed. The criterion of measuring tests and the parameters of test papers and items are analyzed based on the classical testing theory. The method of determining the difficulty coefficients of items is summarized. The thesis describes the idea of designing test paper automatically, the overall functions, and the realization theory. In order to realize the algorithms easily and to consider into the feasibility of educational measuring, an improved item bank is designed, which improves the successful ratio of making test paper and working efficiency.In the analysis of a large number of research results domestically and abroad, the thesis analyzes the algorithms for automatically constructing test papers, and establishes the mathematical model for constructing test papers automatically, whose control parameters are the types of items, the difficulty of items, the distinguishing degrees, the cognitive classification, and the time of arraying paper. Two kinds of test paper based on improved genetic algorithm strategy are proposed. Since the characteristics of finding global optimal solution adaptively and the high convergence speed, The two proposed algorithms solves the problem of the code length of the chromosome too long, and the difficulties to compute the fitness. The chromosome structure and fitness function are constructed and described based on the requirements of organizing test paper. The genetic operation of the problem is designed and the corresponding application software based on the genetic algorithm proposed in this thesis is programmed. It is laid a solid foundation to further improve the construction of the network item bank. Finally, the basic structure of paperless examination system is explored to determine the overall function of the paperless system and determine the direction of studying.