Automatic Scoring Method For Subjective Questions In Chinese Short Text Based On Sentence Vector

Automatic scoring of subjective questions is an indispensable part of large-scale evaluation,and its automatic scoring technology needs to be improved.However,due to the centralized answers and sparse data caused by short text subjective questions,it is difficult to classify text topics during automatic scoring.In view of this,this paper proposes a new automatic scoring model,the BTM-DOC model.This model combines the BTM topic model and the sentence vector model,fusing the text implicit topic information trained by the BTM topic model with the document vector trained by the sentence vector model in a high-dimensional vector representation to construct new features,not only making the model utilize the information of the entire corpus,We also use the local semantic spatial information of the Paragraph Vector to improve the implicit semantic information of BTM,effectively alleviating the problem of sparse text features.The experiment uses open text as a dataset,and selects SVM classifiers to compare the effects of summary texts from various models.The experimental results show that the clustering effect based on BTM-DOC model is superior to BTM,Doc2 vec,and BTM+Word2vec models.