| 基于聚类降维的改进KNN文本分类 |
| |
| 引用本文： | 刘海峰,姚泽清,刘守生,苏展.基于聚类降维的改进KNN文本分类[J].微计算机信息,2010(3). |
| |
| 作者姓名： | 刘海峰 姚泽清 刘守生 苏展 |
| |
| 作者单位： | 解放军理工大学理学院; |
| |
| 基金项目： | 项目名称:基于不完全语言信息的交互式群决策理论及应用资助;基金颁发部门:国家自然科学基金项目(70571087) |
| |
| 摘 要： | 特征向量的高维性以及训练样本分布不均影响文本分类器性能。提出了一种聚类模式下的KNN改进方法。首先使用一种改进的聚类方法对文本特征集进行初步筛选,随后使用一种基于类别的改进KNN分类器进行分类,减少了噪声样本对测试样本类别判定的干扰。试验结果表明本文提出的分类模型在分类效率上得到提高。
|
| 关 键 词： | 特征降维 聚类 文本分类 K平均 K近邻 |
An Improved KNN Text Categorization Method Based on Clustering |
| |
| LIU Hai-feng YAO Ze-qing LIU Shou-sheng SU Zhan.An Improved KNN Text Categorization Method Based on Clustering[J].Control & Automation,2010(3). |
| |
| Authors: | LIU Hai-feng YAO Ze-qing LIU Shou-sheng SU Zhan |
| |
| Affiliation: | LIU Hai-feng YAO Ze-qing LIU Shou-sheng SU Zhan(Institute of Sciences,PLA University of Science , Technology,Nanjing,210007,China) |
| |
| Abstract: | The problem of large feature dimension and data distribution odds reduces the KNN function. This paper brings forward an ameliorative KNN method basing on clustering. Firstly,use an improved k-mean clustering method to select feature. Secondly,recur to an improved KNN method to realize text categorization. This method decreases the disturbing that comes from the noise stylebooks. The experiment shows that this method has improved the precision. |
| |
| Keywords: | feature reduction clustering text categorization k-mean KNN |
| 本文献已被 CNKI 等数据库收录！ |
