【作者】 司红敏；

【导师】 张太红；

【作者基本信息】 新疆农业大学 ， 农业机械化工程， 2007， 硕士

【摘要】 试题类型是决定大规模教育考试信度和效度的因素之一,本文分析了现有传统笔试题型和现有网络考试平台题型之间的差异,介绍了布鲁姆知识测评原理,以及测量目标与题型的关系。针对现有eQuiz电子考试系统试题类型不足的实际,提出增加在线考试系统的新题型。客观性试题增加了匹配题和排序题;主观题性试题增加了名词解释、简答题、问答题和分析题。本文以eQuiz电子考试系统为模型,在不破坏原有试题结构的基础上研究、设计并实现上述题型,给出了新增加试题的底层数据存储结构、界面设计和实现方法。在判分实现上,客观题全部采用机器阅卷;主观题采用机器阅卷和人工辅助阅卷两种方式。对于程序设计和分析题,机器阅卷存在一定的局限性,只能采用人工辅助阅卷方式。机器评分是通过计算学生答案与标准答案两个文件之间的相似度来评定学生的主观题得分。其核心技术是潜在语义分析,通过SVD分解,消除同义词,多义词的影响,提高后续处理的精度。通过eQuiz电子考试系统扩充题型前后测试数据对比分析,表明扩充题型后的eQuiz电子考试系统更能客观,有效的测量出学生的知识掌握程度和技能水平。更多还原

【Abstract】 The question type of examination is the key factor that decides the reliability and validity in large scale test. In this paper, we analyzed question types of some existing commen tests and online examinations, introduced the taxonomy of educational objectives, and the relation between educational objectives and question types. We also analyzed and introduced the measurement goal which the question type could do. By doing what we mentioned above, we discovered that there was a large gap between the question types of online test and written test, which mainly based on the difficulty of question-design and the subjective questions score. In the article, we based on the question types insufficiency of the eQuiz Examination System, and then put forward some new question types which included matching question, ordering question, nominal explanation, simple answer question, question-and–answer question and analyzing question. At last, we not only designed the data storing structure, interface and the method adopt, but also added some new modules which included questions input and two kinds of subjective questions grading. In Subjective questions auto-grading, We adopt Latent Semantic Analysis. Through SVD decomposition, we eliminate synonyms, impact of the polysemous words, improve handling precision. In the end, we did three experiments to certificate the relativity between the lager scale written test and the eQuiz online test.更多还原