【作者】 肖雪莲；

【导师】 沈霄凤；

【作者基本信息】 华东师范大学 ， 教育技术学， 2006， 硕士

【摘要】 在计算机自动测评(CAA)领域中，针对单项选择题、多项选择题和填空题等客观试题的自动批改技术已经很成熟，被应用于大型的考试系统中。但是，针对名词解释、简答题、论述题以及作文题等主观题型的自动批改技术，由于其受到人工智能、自然语言理解、模式识别等理论与技术发展的限制，所以至今还很不完善，没有实用的系统出现。 国内对主观题的批改依然是人工批改，但人工批改的结果会受很多主客观因素的影响，如学生卷面的整洁程度、学生字体的美丑、教师当时的情绪状态等，从而降低了学生之间竞争的公平性和公正性。 而计算机的运算速度快、效率高、精度高、不会疲劳，工作不受主观因素影响，尤其适用于主观题的自动批改。因此，研究如何利用计算机来实现主观题的自动批改具有很大的现实意义。 本文针对汉语理解的特点，将HNC(概念层次网络)理论引入到主观题的自动批改系统中来，利用HNC理论的知识表达方法以及句类分析技术来实现计算机对汉语句子的语义成分自动分析，设计出了基于HNC理论的主观题自动批改算法，并实现了一个主观题自动批改实验系统，达到了一定的批改效果。更多还原

【Abstract】 In the CAA (Computer Assisted Assessment) area, the techniques of the auto-assessment of the object questions, such as single choices, multiple choices and vacancy filling questions etc., have been very perfect and been used in large-scaled testing systems. But, when refers to the subject questions, such as the term explanation, short answer questions, essay questions and compositions etc., their auto-assess techniques remain unperfect, because the subject questions’ auto-assess techniques are restricted by the development of the techniques in Artificial Intelligence area, the Natural Language Understanding area and the Pattern Recognition area, so, there was no applied system in use.In China, the subject questions’ assessment is still in manual. What’ s more, there are many object or subject factors that will influence the result, such as the cleanesss of the student’ s papers, the beauty or ugliness of the student’ s handwriting, the teacher’ s mood etc. That brings down the justiness of the competition between the students.The computer, however, is prior to human in processing speed, efficiency, precision and never be tired. What’ s more, there is no object or subject factor that can influence its work. So, it is very fit for the auto-assessment of the subjective questions.In this paper, we introduced the HNC (Hierarchical Network of Concepts) theory into the Subjective-Question-Autoassess-System. Using its knowledge expression method and sentence category analyzing method to analyze the semantemes of Chinese sentences automatically. Then, we designed the according algorithms based on HNC theory. Finally, we developped the Subjective-Question-Autoassess system. The auto-assess result was pretty good.更多还原