摘要
通过分析文本的特征,提出了一种基于稀疏约束非负矩阵分解(NMFSC)的文本聚类新方法。该方法用NMFSC分解词-文本矩阵来降低特征空间的维度,并依照稀疏约束更好地控制稀疏度,然后利用簇中文本的相似性进一步细化簇。实验表明,与基于k-means的文本聚类方法和基于NMF的文本聚类方法相比,此方法具有较高的归一化互信息值(NMI),从而具有良好的聚类性能。
Through analyzing the characteristics of the text, a novel text clustering approach based on Non-negative Matrix Factorization with sparseness constraint （NMFSC） is presented. The method uses NMFSC decomposing word-text matrix to reduce the dimension of the feature space, and better controls sparsity with sparseness constraint, and then further refines clusters by using the similarity of documents in clusters. Compared with text clustering method based on k-means and text clustering method based on NMF, the results of experiment show that the method has high value of the normalized mutual information, thus it has good clustering performance.
出处
《计算机系统应用》
2011年第9期78-81,156,共5页
Computer Systems & Applications
关键词
文本聚类
细化簇
非负矩阵分解
稀疏表示
归一化互信息值
text clustering
refine clusters
non-negative matrix factorization
sparse representation
normalized mutual information
