In view of the difficulty and unreasonable proportion of test types in the current computer-assisted teaching system,a combined optimization model with multiple constraints is proposed to solve the problem of multiple index parameters in automatic test composition,which provides reference value for realizing scientific test composition strategy.The model comprehensively considers the total score,difficulty,exposure,proportion of question types,amount of chapter knowledge and training objectives of the test paper,and carries out quantitative weighting,establishes the total constraint equation of intelligent test paper composition,and constructs the com-bined optimization model with multi-parameter constraints.The question bank was established by computer sim-ulation,and the accuracy and efficiency of three heuristic search algorithms were analyzed.The results show that the overall accuracy and efficiency of the automatic paper composition using genetic algorithm are higher,and the average deviations in total score,difficulty,exposure rate,proportion of question types,amount of chapter knowl-edge and training objectives are 0.2％,0.074,0.1％,6.1％,8.2％and 9.6％,respectively.The optimal solution can be basically achieved within 230 iterations.
维普
万方数据