摘要：该文运用多特征融合进行文本相似度的计算，并利用决策树算法C4.5进行文本分类，构建决策树分类器，完成对主观题的自动评阅。通过实验结果表明，该算法准确率高，与人工阅卷相接近，具有一定的应用前景。
关键词：多特征；相似度；决策树；文本分类；评阅
中图分类号：TP391.2文献标识码：A文章编号：1009-3044(2012)15-3579-04
Algorithm Design of Subjective Question Auto Assessment
MU Wei-wei1,2, WANG Guo-cai1
(1.College of Information Science and Engineering, Central South University, Changsha 410083, China; 2.Hunan Chemical Vocational Technology College, Zhuzhou 412004, China)
Abstract: This paper use the multi-features combinaion forr text similarity computing, and take use of the C4.5 decision tree algorithm for text classification to build a decision tree classifier. In this way, to complete the review on the subjective question automatically. Experi mental results shows that the algorithm accuracy rate close to the manual scoring, It has a certain degree of application prospect.
Key words: multi-features; similarity; decision tree classification; text classification; assessment
1）数据样本复杂，表达方式多样化，关键词的提取存在偏差；
2）多特征相似度匹配具有一定的优势，但基于特征的多样性，匹配程度还达不到100%；
3）文本分类算法还需进一步优化。
本文采用多特征相似度计算和C4.5决策树算法进行主观题自动评阅，通过实验结果表明，该算法性能优良，评阅准确率较高，具有一定的实用参考价值。