基于稀疏差异度的聚类方法在信息分类中的应用

针对文本信息聚类中的高属性维稀疏数据聚类问题,采用计算对象间稀疏特征差异度来度量文本对象之间的相关度,结合最小生成树的方法来进行聚类分析,提出一种基于稀疏特征差异度的聚类方法。通过实例表明,该算法对于多关键字匹配的文本信息分类十分有效,并可根据关键字的重要程度进行加权计算,使聚类更加符合实际情况。该算法将在高维稀疏数据挖掘中有着重要应用。

Research of Sparse Feature Difference-Based Clustering Method Applied to Information Classifying

The sparse feature difference degree and the minimum cost spanning tree are used to resolve the problem of high attribute dimensional data clustering which exists in information classifying,in order to support the central idea,an example is given in the paper.The algorithm is very efficient when a lot of key words need to be matched in information classifying,and it also can consider the different importance of the keywords,so the algorithm can get an objective result,and it will be very important in high attribute dimensional data mining.