Research On Automatic Testing Paper Generation Based On Hybrid Intelligent Algorithms

As the development of computer technologies and Artificial Intelligence, the research of exam paper auto-generating system has been paid more attentions gradually. The exam paper auto-generating system would be used to relieve teachers of repetitive works, and to improve the work efficiency. The intelligent function of forming exam paper would improve the quality of exam papers and increase the scientific and objectivity of test. The problem of exam paper generation is a multi-constraint and multi-objective optimization issue. It not only involves proposing the mathematics model of test paper auto-generating, but also includes studying on the application algorithms.Firstly, the basic theories and principles of testing paper generation are expounded and the constraint conditions are summarized. Anda mathematical model of testing paper generation is proposed. The multi-objectives is transformed to the simple target by using the weight coefficient method. It has solved the problem which many goals conflicts mutually, and has carried on the detailed discussion to constraint conditions of the mathematical model.Then, this thesis has studied the automatic-generating testing paper algorithm.We analyzed the shortcomings of current techniques. The tabu search algorithm of the genetic algorithm is introduced, a universal hybrid intelligent algorithm is designed to solve the problem of making testing paper. In the test papers model,the subject type, subject amount,and score value are taken as a basis and they are kept unchanged all the time during the process of initial species group formation and evolution, so that the optimization object was simplified.Finally, experimental result shows that this method is able to prevent successfully the excessive similar individuals in the species group from duplicating. It keep the variety of the group, and restrain the occurrence of premature phenomenon. The efficiency of genetic algorithm application in auto-generating test papers is improved. The operational time is shorter, and error is less after being improved.