Density Peak Clustering Algorithm Based On Three-Way Decision And Its Application

Density peak clustering algorithm is an algorithm that uses density peak points to realize clustering.Compared with other clustering algorithm,the traditional density peak clustering algorithm has already some benefit,such as high clustering efficiency,simple algorithm implementation,insensitive to noise,good robustness,etc.,and is widely used in many fields.However,the density peak clustering method still has some problems,such as the cluster centers need to be artificial selected on the decision graph,which has high subjectivity and uncertainty.If the distribution of data sets is relatively complex,the selection of clustering centers cannot be efficient and accurate.In order to overcome the defect that the density peak clustering method cannot automatic choice the clustering center,This thesis introduces the three-way decision theory to optimize the density peak clustering algorithm and proposes a new clustering algorithm.The feasibility and high efficiency of the proposed method are verified by the work of different data sets.Finally,it is applied to the news text data set to cluster the messy text data.The specific work of this thesis is as follows :(1)For the defects of traditional density peak clustering algorithm,the research idea of three-way decision algorithm is introduced,and a density peak clustering algorithm based on three-way decision is proposed.Firstly,the statistical characteristics of the two parameters of density and distance are used to put the qualified clustering center into the core domain,and the suspected clustering center point which is difficult to determine is put into the boundary domain.Then the concept of k-reachable domain is defined,and the suspected clustering center is clearly analyzed by the new criterion,so as to select the actual clustering center.(2)To solve practical problems,the density peak clustering algorithm based on threeway decision is used in news text data set to realize the application of text clustering.The content extraction,content cleaning,text segmentation,text feature extraction,weighting and dimension reduction of news text samples are carried out,and then the data are clustered by the method in this thesis.The results show that the density peak clustering method based on three-way decision can effectively solve practical problems.