At present,the text feature generation algorithm generally used vector space model(VSM) , the model used TF-IDF evaluation function to calculate individual features weight, the redundancy of text features generated by this algorithm were higher. In order to solve this problem, the paper put forward a text feature generation algorithm based on the clustering of weighted, first the algorithm weighted features with the initial weights, then used the semantics and the entropy to deal with the feature of the further weights, finally used features clustering to undertake eliminate redundant features. Experiments show that the algorithm is better than the traditional TF-IDF algorithm,classification accuracy is higher than the 5% average.%目前的文本特征生成算法一般采用加权的文本向量空间模型,该模型使用TF-IDF评价函数来计算单个特征的权值,这种算法生成的文本特征冗余度往往都比较高.针对这一问题,采用了一种基于聚类加权的文本特征生成算法,首先对特征候选集进行初始加权处理；然后通过语义和信息熵对特征进行进一步加权处理；最后使用特征聚类对冗余特征进行剔除.实验表明该算法比传统的TF-IDF算法的平均分类准确率高出5％左右.
展开▼
