中文评教文本分类模型的研究

用文本分类的方法找出中文评教信息的情感倾向，使学生主观评价里蕴含的信息得到有效利用，是对现有评教系统的必要补充．采用基于潜在语义分析的方法对文本向量降维，并用支持向量机的分类方法对目标文本进行分类，得到每一条主观评价的情感倾向．分析了特征选择、特征抽取方法、降维维数、词性、训练集合与测试集合样本的比例等几方面对分类的影响，找到了较好的中文评教文本分类模型．

Research of Classification Model for Chinese Evaluation Text of Teaching

As a necessary complement,the emotional tendency of subjective evaluation information in the current teaching evaluation system is mined by text classification method in order to effectively utilize information of subjective evaluation.Latent semantic analysis(LSA) is adopted to reduce the dimensionality of text,and support vector machine(SVM) is used to classify the target text.Some factors of classification model,such as feature selection and extraction methods,the reduced dimension,the effect of part of speech on classification accuracy,and the ratio of training sample set to test set,are analyzed and a sound classification model is found.