| 使用潜语义分析的汉语作文自动评分研究 |
| |
| 引用本文： | 曹亦薇,杨晨.使用潜语义分析的汉语作文自动评分研究[J].考试研究,2007(1). |
| |
| 作者姓名： | 曹亦薇 杨晨 |
| |
| 作者单位： | 深圳大学师范学院，深圳大学师范学院 |
| |
| 摘 要： | 本研究是第一个使用潜语义分析技术对汉语作文进行计算机自动评分的研究。首先对202篇高中作文进行人工评分,两位评分员评分相关为0.62。然后使用潜语义分析评价作文,先得到内容分数,其和人工评价内容分的相关达到了0.47;再使用该内容分对总分进行回归,回归方程的决定系数为0.3,回归所得总分和人工评价总分的相关达到了0.55。本研究表明,潜语义分析技术在汉语作文自动评分中起着重要作用。进一步的研究需要寻找更多的指标,并辅以其他方法来提高评分效果。
|
| 关 键 词： | 潜语义分析 汉语作文自动评分 奇异值分解 |
Automated Chinese Essay Scoring with Latent Semantic Analysis |
| |
| Cao Yiwei Yang Chen Normal college of Shenzhen University,Guangdong,Shenzhen.Automated Chinese Essay Scoring with Latent Semantic Analysis[J].Examinations Research,2007(1). |
| |
| Authors: | Cao Yiwei Yang Chen Normal college of Shenzhen University Guangdong Shenzhen |
| |
| Institution: | Cao Yiwei Yang Chen Normal college of Shenzhen University,Guangdong,Shenzhen,518060 |
| |
| Abstract: | This is the first research of Automated Chinese Essay Scoring using the method Latent Semantic Analyze.First,get the human-graded-score of 202 essays written by middle school students,with the correlation 0.62 between two graders.Then get the computer-graded-content-socre by Latent Semantic Analyze, and the correlation of content score between two kinds of scores was 0.47.Last,get the holistic score by regression,with the R-square 0.3,and the correlation between computer and human is 0.55.The result shows Latent Semantic Analyze is sufficiently useful in Automated Chinese Essay Scoring.Getting the most exact score needs more features and other methods in coming researches. |
| |
| Keywords: | Latent Semantic Analysis Automated Chinese Essay Scoring Singular Value Decomposition |
| 本文献已被 CNKI 等数据库收录！ |
|
