【摘要】 针对在线评测系统中缺少对学习者编程能力的客观反馈和评估指标权重难以确定的问题,建立遗传算法(genetic algorithm, GA)优化反向传播(back propagation, BP)(GA-BP)神经网络的编程能力评估模型。通过对在线评测系统的数据进行挖掘,提取编程能力评估指标,采用双参数平衡熵权法和离差最大化法的客观组合赋权方法确定指标权重以及编程能力评估值,将评估值作为GA-BP神经网络的期望输出,并与单一BP神经网络确定的编程能力评估结果进行比较。研究结果表明,利用该模型能够实现对学习者编程能力的评估且评估结果更准确。更多还原

【Abstract】 Aiming at the problem of lacking objective feedback on learners’ programming ability and the difficulty of determining evaluation index weights in online judge, an evaluation model of programming ability based on genetic algorithm and improved back propagationl(BP) neural network(GA-BP) was proposed. The evaluation indexes of programming ability were extracted by mining the data of the online judge system and using the objective combination weight method of the two-parameter balanced entropy weighting and the deviation maximization, in which the index weight and the programming ability evaluation value could be determined, and the evaluation value was used as the expected output of the GA-BP neural network. The evaluation results were compared with those of a single BP neural network. It is shown that this model can be used to evaluate learners’ programming ability and the results of evaluating are more accurate.更多还原