结合分词技术与语句相似度的主观题自动判分算法  

Automated correcting algorithm for subjective questions based on segmentation technology and sentence similarity

作　　者：李玉红[1] 柴林燕[2] 张琪[1] 

机构地区：[1]承德医学院数学教研室,河北承德067000 [2]中国医学科学院北京协和医学院生物医学工程研究所,天津300192

出　　处：《计算机工程与设计》2010年第11期2663-2666,共4页Computer Engineering and Design

基　　金：河北省教育厅指令性课题基金项目(2007301)

摘　　要：针对网络考试系统中主观题自动评分面临的困难和问题,提出一种基于中文分词技术结合语句相似度的主观题自动判分算法。该算法利用字典与统计相结合技术将语句切分成具有独立意义的单词,通过计算词与词之间的相似度,并结合词形、词序和语句长度等多层次相似度计算得到整句的相似度。利用语句相似度计算结合改进关键字匹配算法得到最终分值。经课题中多媒体简答题的测试,结果表明了该算法的有效性。Aiming at the difficulties and problems faced by the automated correcting for subjective questions,an automated correcting algorithm for subjective questions based on Chinese segmentation and sentence similarity is proposed.Firstly,a sentence is segmented into some independent words based on the dictionary and statistics,and then the similarity of the whole sentence is obtained by computing the similarity of the words including the appearance,order and the length.Finally,the result is gotten by combining the sentence similarity and the improved keywords matching algorithm.The algorithm is perfect proved by experiment tests in the multimedia subjective questions.

关 键 词：分词技术 语句相似度 主观题 自动判分 关键字匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]