TensorFlow在大学英语语言训练中的应用

The Application of Tensor Flow in College English Language Training

北京邮电大学人文学院每学期都利用通用计算机化考试平台(大学英语语言技能训练系统)来进行英语测试,其中客观题系统能够对照答案直接给出分数,但是主观题只能依靠英语教师来逐个批改。本研究是利用tensorflow深度学习平台对英语口语表达题进行智能批改,取音素后验概率、语速ROS、关键词覆盖率、文本覆盖率、用词变化程度等维度作为特征,学生考试成绩为目标,用tensorflow深度学习平台进行模型训练,实现学生成绩的预测。

Beijing University of Posts and Telecommunications College of Humanities each year using the general computerized examination platform(College English language skills training system)for English testing,in which the objective system can directly give the answer to the score.But the subjective questions can only rely on English teachers to change one by one.In this study,we use the tensorflow,deep learning platform,to intelligently correct the oral expression of English and take the dimension of phoneme posteriori probability,speed rate ROS,keyword coverage,text coverage,degree of change of words,and so on.With tensorflow deep learning platform for model training to achieve student achievement prediction.