摘要
在计算机无纸化考试中,自动评分的设计与实现是最重要也是最难的一个环节.编辑距离算法通过统计替换、删除、插入、交换操作来计算两文本间相似程度,应用到文本匹配类的自动评分中具有较高的信度.在分析默写题评分需求的基础上,对编辑距离算法做出了相应的改进,提高了相似度计算的精度,并通过回溯的方法,实现了评分过程的反馈.实际应用的数据表明,该方法在汉字评分中具有普适性.
In the paperless exam （using computer）, the design and implement of computer-automated scoring （CAS） is the most important and most difficult part. The algorithm of Levenshtein Distance calculates the degree of similarity between two texts by statistics of replace, delete, insert and swap operation times, and has high reliability applying to CAS such as text matching. Based on the analysis of dictation exam needs, an improved algorithm of Levenshtein Distance proposed to get a more accurate similarity was proposed, and the feedback of scoring process was implemented by the method of Backtracking. The data shows that this method has a universal feature in Chinese CAS.
出处
《广州大学学报（自然科学版）》
CAS
2012年第4期79-83,共5页
Journal of Guangzhou University:Natural Science Edition
基金
国家语委"十二五"科研规划2011年度重大项目(ZDA125-3)资助
关键词
自动评分
编辑距离
文本匹配
相似度计算
回溯
computer-automated scoring
levenshtein distance
text matching
similarity computing
back tracking
