| Summary: | 碩士 === 國立彰化師範大學 === 商業教育學系 === 89 === Researches of applying information technologies to educational assessment have made progress on developing computerized adaptive testing system. The supports for automated grading are mainly on objective test such as multiple-choice item. For open ended test such as essay- type questions, the support for online automated grading has yet to be developed. The traditional manual essay-grading is time-consuming and has long been criticized for not being objective. Therefore, to find more objective and effective supports for the essay-assessment task is worthy of exploration. Our research will focused on the exploration of grading methods and characteristics of essay questions. The difficulties lie in the processing Chinese words , how the objective criteria can be established based on the outcomes of subjective grading of human grader , and how to extract suitable characteristic values to represent students’ answer. To tackle the difficulties mentioned previously, we apply a Chinese natural language software tool with high precision to do the Chinese segmentation processing .We also employed the vector space model that is widely used in information retrieval ,in conjunction with the results of manual human grading to develop software tools for training and extracting objective grading rules and representation of students’ answer. Accordingly, we construct an automated essay question processing model and an software prototype. To see the effectiveness of the model we choose an undergraduate course to experiment the online essay question test. The results are collected and analyzed. The software prototype is modified based on the analysis of the result. Although some problems remain to be solved , the high grader reliability shown in the final results of our approach is encouraging.
|
|---|
