中文摘要：

针对电影影评语句短小、特征矩阵稀疏问题,提出一种利用本体扩展特征矩阵的方法。首先通过传统与新型文本分类方法的比较和分析,发现适合中文短影评的分类方法,并通过试验证明决策树的短文本分类效果优于SVM、Bayes和KNN等文本分类方法,然后进一步利用决策树分类本体扩展后的特征向量。试验表明,基于本体扩展的中文短影评的分类效果比传统的分类效果提高3%,查准率达到90.1%。

英文摘要：

Aiming at the problems of film reviews that the sentences are short and characteristics matrix is sparse,a method using ontology to expand the matrix was proposed. Through comparison and analysis of traditional and developmental text classification methods,a suitable way for Chinese short film reviews classification was found. The experiment results proved that the decision tree is better than the SVM,Bayes and KNN in this essay,and the decision tree classifier was further used to classify the feature vectors of the ontology expanding. The results of experiment showed that the effect of Chinese short film reviews classification based on the ontology expanding was 3% higher than the traditional methods,and the classification accuracy reached 90. 1%.