CAJ | 학술논문

本文提出了基于文本分类的主观题自动评分模型。模型采用文本词性相似度和文本浅层相似度作为分类器的条件属性，在一定程度上提高了文本的语义理解。通过对已有文本的学习，使用考生分数作为分类类别构建决策树分类器。将待测文本输入决策树分类器从而实现答案的分类，即完成自动评分。通过与人工阅卷过程对比，验证了系统是有效可行的，符合人工阅卷的过程。
본문제출료기우문본분류적주관제자동평분모형。모형채용문본사성상사도화문본천층상사도작위분류기적조건속성，재일정정도상제고료문본적어의리해。통과대이유문본적학습，사용고생분수작위분류유별구건결책수분류기。장대측문본수입결책수분류기종이실현답안적분류，즉완성자동평분。통과여인공열권과정대비，험증료계통시유효가행적，부합인공열권적과정。
The model adopts text speech and surface similarities as condition attributes for classiifer.In a certain extent,our model improves the semantic understanding of the text. Using the examinee scores as classiifcation categories to construct the decision tree classiifer through the study of existing text.Final y,the automatic scoring reduced to input the student answer to the decision tree and the tree decide what score should be marked.Compared with the manual scoring process,we ifnd our model is feasible and effective and in line with manual marking process.