Computation And Application Of Skyline Based On K-Medoids Clustering

In recent years,with the development of society,especially the Internet,the amount of application data has increased dramatically,resulting in the rapid development and widespread use of database query technology.Skyline query has been introduced into the database field because of its advantages in multi-objective decision making and has become a research hotspot in the fields of database query.However,most of the current Skyline queries are designed for single user scenario,which are limited to return single result set,or Skyline queries in multiple subspaces composed of different dimensions,and return multiple result sets.With the development of technology application and the emergence of new requirements,different needs of different users should be taken into account in practical applications.What they may be interested in is the different parts of the data in the global data,which have the characteristics of local inside similarity and local dissimilarity.On the basis of analyzing the existing Skyline query algorithm to solve this problem,a Skyline query algorithm based on K-Medoids clustering is proposed.The algorithm can be applied to many important fields,such as recommendation system,sensor network,data analysis,and so on.For example,we can deal with the problem of optimal selection of multiple commodities of different levels at the same kind when users purchase goods,and multi-objective optimization of sensor nodes satisfying different conditions in wireless sensor networks.This paper focuses on the following aspects:(l)When introducing the relevant algorithms of traditional Skyline query,the advantages and disadvantages of different algorithms are analyzed,as well as their computational efficiency.(2)In view of the fact that the existing algorithms do not take the similarity between data into account,a Skyline query algorithm based on K-Medoids clustering is proposed.The similar data is clustered into a cluster,and the different data is clustered into different clusters.The data in clusters are similar,and the data between clusters are different.(3)The dimension disaster problem is briefly introduced and analyzed,and reduces the dimension of data by positive correlation according to the correlation between different dimensions of data.Therefore,on the basis of the improvement of Skyline query algorithm based on K-Medoids clustering,a K-Medoids clustering Skyline query algorithm based on dimensionality reduction is proposed.The experimental part of this paper is divided into two parts:In the first part,six groups of experiments are designed to compare the average of Skyline points based on K-Medoids clustering with the average of Skyline points obtained by BNL(block nesting algorithm).The experimental results show that the average of Skyline points based on K-Medoids clustering is more selective than the average of Skyline points obtained by the BNL algorithm.In the second part,an experiment is designed.Through comparison,it is found that' the improved Skyline algorithm based on K-Medoids clustering is obviously better than the Skyline algorithm which is not improved before.At the same time,the proposed algorithm is validated by a specific case.