Design And Realization Based Of Intelligent Test-paper Composing System On Genetic Ant Colony Algorithm

In computer-aided instruction (CAI), the design and development of intelligent test-paper composing system has become a hot research issue. It is a big challenge to guarantee that the test papers composed by computers can meet different demands of the users in a stochastic, scientific and reasonable way. This thesis mainly investigates the mathematical model of intelligent test-composing problems and the preprocessing technology of test bank, and proposes an intelligent test-composing algorithm based on genetic algorithm and ant colony algorithm used in test-composing system, and finally realizes a prototype system in test-composing based on the real situation of Hunan Institute of Science, Technology and Humanity. The contents of this paper are as follows:Firstly, it introduces related theory about intelligent test-composition and analyzes the advantages and disadvantages of some common intelligent test-composing algorithm. It then, based on the real situation of our college, constructs an evaluation index system meeting effective examining conditions, examining reliability and validity, and it has detailed analysis on each index, based on which a mathematical model of intelligent test-composition is constructed by adopting statistic methods.Secondly, to solve the problem of low-efficiency in recent test-composition by composing papers directly on test bank, we design a kind of ideal test questions which are taken as the training samples for the classification of test bank by Nearest Neighbor Classification Algorithm, so as to support the subsequent intelligent test-composing algorithm.Thirdly, it mainly analyzes the advantages and disadvantages of gene tic algorithm and ant colony algorithm, and proposes an intelligent test-composing algorithm based on genetic algorithm and ant colony algorithm used in test-composing system. A test experiment is carried out by adopting the preprocessing technology put forward in the forth Chapter, and the result shows that the success rate and the speed of test composition has been obviously improved.Finally, we make careful requirement analysis based on the current developmental situation of campuzs network construction and information program of teaching administration, and the need of examining and teaching of some major public courses in our college, based on which we carry out functional section classification of the intelligent test-composing system and design test bank system, and realize a user-friendly intelligent test-composing prototype system by taking SQL Server 2000 as back-end database platform and adopting Java, a recently popular device, for programming development.