|
|||||
|
|
| 基于NMF的用户模板构造方法 | |
| 引用本文： | 黄钢石,张亚非,陆建江,肖江.基于NMF的用户模板构造方法[J].情报学报,2004,23(4):394-398. |
| 作者姓名： | 黄钢石 张亚非 陆建江 肖江 |
| 作者单位： | 1. 解放军理工大学通信工程学院,南京,210007 2. 解放军理工大学,南京,210007 3. 东南大学计算机科学与工程系,南京,210096 |
| 基金项目： | 国家自然科学基金 (6 0 0 730 12 )，国家自然科学基金重点项目(6 99310 4 0 )支持 |
| 摘 要： | 文本过滤是网络安全领域的一个重要研究课题。基于内容的文本过滤关键在于建立语义层次上的用户模板。本文提出一种基于非负矩阵分解 (Non negativeMatrixFactorization ,NMF)的用户模板构造方法。该方法应用NMF算法分解项 文本矩阵来获取项之间的相关性。在此基础上 ,引入语义向量和权重向量的概念 ,并通过定义语义向量的类别区分度来提取用户模板。实际表明 ,与基于奇异值分解的潜在语义索引方法相比 ,该方法不仅较大地提高了过滤精度 ,而且具有计算速度快、占用存储空间较少的优点 |
| 关 键 词： | 文本过滤 用户模板 非负矩阵分解 潜在语义索引 Web挖掘 |
| 修稿时间： | 2003年10月27 |
A NMF-based Method for Constructing User Profile |
|
| Huang Gangshi ,Zhang Yafei ,Lu Jianjiang and Xiao Jiang.A NMF-based Method for Constructing User Profile[J].Journal of the China Society for Scientific andTechnical Information,2004,23(4):394-398. | |
| Authors: | Huang Gangshi Zhang Yafei Lu Jianjiang and Xiao Jiang |
| Institution: | Huang Gangshi 1,Zhang Yafei 2,Lu Jianjiang 3 and Xiao Jiang 1 |
| Abstract: | Text filtering is an important issue in network security field.The key to content based text filtering consists in constructing the semantic representation of users' profile.In this paper,a method based on non negative matrix factorization (NMF) for constructing user profile is presented.According to NMF,the term text matrix is decomposed to capture the relations between terms.Then,we introduce the concepts of semantic vectors and weight vectors,and define the class discriminative degree to extract the user profile.The result experimentally shows that,compared with latent semantic indexing based on singular value decomposition,this method not only improve filtering precision markedly,but also has the merits of fast computing speed and lesser memory occupancy. |
| Keywords: | text filtering user profile non-negative matrix factorization latent semantic indexing Web mining |
| 本文献已被 CNKI 万方数据 等数据库收录！ | |