基于孪生神经网络的主观题自动阅卷评分技术

Automatic Scoring Technology of Subjective Questions Based on Twin Neural Network

智能阅卷能大大减轻老师人工阅卷的工作量,并能在一定程度上提高评卷的公平性。目前主观题自动阅卷评分精确度不高,应用不广泛。基于孪生神经网络的主观题自动阅卷评分技术,同时输入学生答案和参考答案进行相似度计算,从而估计学生答案的得分,改善单单基于句子表面特征的相似度计算方法,提高准确度。

Intelligent marking can greatly reduce the workload of manual marking and improve the fairness to a certain extent.At present,the accuracy of automatic scoring of subjective questions is not high,and it is not widely used.Based on the twin neural network,the automatic scoring technology of subjective questions can input students'answers and reference answers at the same time,and calculate the similarity,so as to estimate the scores of students'answers.It improves the similarity calculation method based on sentence surface features and improves the accuracy.