中文主观题自动批改中相似句子检索算法  

Similar Sentence Search Algorithm in Automated Assessment System of Subjective Test Based on Chinese

作　　者：张小艳[1] 

机构地区：[1]西安科技大学计算机系,陕西西安710054

出　　处：《南京师范大学学报（工程技术版）》2007年第2期62-66,共5页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：陕西省教育厅专项课题基金(06JK248)资助项目

摘　　要：学生答案与标准答案语义匹配程度的计算是基于中文文字类主观题自动批改中的关键问题.提出了学生答案与标准答案匹配程度的计算分两步进行:候选相似语句的检索和基于语义依存的句子相似度计算.利用动态规划法实现候选语句检索,确定数量不多但有可能与标准答案相似的候选句子,然后对标准答案中的句子与少量的候选句子进行深层的句法分析,找出依存关系,并在依存分析结果的基础上进行语义相似度计算,得出最终的结果.该方法可以提高主观题自动批改的效率及准确性,具有一定的实用价值.The key problem in the automated assessment of the subjective test is the computation of the semantic marching degree between the student answer and the standard answer. The computation method of the matching degree between the student answer and the standard answer is proposed which includes two steps of searching for the similar candidate sentences and computing the similarity degrees between sentences based on semantic existence dependence. The dynamic programming is used to search for the candidate sentences, and then the candidate sentences are determined. The number of the determined sentences is not much, but they may have the similar meanings with the standard answer. Furthermore, the deep syntax analyses on the standard answer and the few candidate sentences are implemented to find the dependent relationship. The final result is acquired by the computation of the semantic similarity degree based on the result of the dependent analysis. This method has the practical value to a certain degree because it can improve the efficiency and the accuracy of automatic check about the subjective questions.

关 键 词：自动批改 动态规划法 语义相似度 依存树 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]