Weighted Aggregation Based Clustering Algorithm for Blog Tag Taxonomy Construction
Zhang Hong-jiang · Journal of Chinese Computer Systems · 2009
Hierarchical clustering is a widely used technique for taxonomy construction.However,ill conceived structure of taxonomy is a major factor threatening their use to further web applications.This paper presents WAC(Weighted Aggregation Based Clustering) algorithm,a novel hierarchical clustering algorithm,to overcome the shortcoming.WAC uses the numeric techniques of Algebraic Multi-Grid method to perform fine-to-coarse weighted aggregation recursively.Meanwhile every node in the taxonomy is described by a representative data to make users explore more easily.Evaluations on blog tag clustering demonstrate that WAC performs more consistently with human judgments than other alternative methods.