基于改进WPCA的OSM贡献者特征降维与分类方法  

Improved WPCA-based Feature Dimension Reduction and Classification for OSM Contributors

作　　者：韦兴财 赵肄江[1,2] 刘毅志 WEI Xingcai;ZHAO Yijiang;LIU Yizhi(School of Computer Science and Engineering,Hunan University of Science and Technology,Xiangtan 411201,China;Key Lab of Knowledge Processing and Networked Manufacturing,Hunan University of Science and Technology,Xiangtan 411201,China)

机构地区：[1]湖南科技大学计算机科学与工程学院,湖南湘潭411201 [2]湖南科技大学知识处理与网络化制造湖南省普通高校重点实验室,湖南湘潭411201

出　　处：《地理信息世界》2021年第6期22-27,共6页Geomatics World

基　　金：国家自然科学基金(41871320);湖南省教育厅科学研究重点项目(19A172)。

摘　　要：由于开放街道地图(OpenStreetMap OSM)贡献者的非专业性,其贡献经验对数据质量有着非常重要的影响,因此聚类分析不同经验的贡献者具有实际意义。在将贡献者信息特征分为3类的基础上,提出一种改进的加权主成分分析方法(WPCA)对志愿者的贡献特征进行分组归一化、加权和降维,然后采用高斯混合模型(GMM)方法将贡献者聚类成4个不同的组,最后将结果与主成分分析法(PCA)、K-Means组合方法进行比较。通过比较分析,改进WPCA与GMM组合方法比PCA与K-Means组合方法的贡献者分类效果更好。Contributors have a significant impact on data quality of OpenStreetMap (OSM) because most of them are the non-professional.Therefore,it has practical significance to analyze contributors based on their different experiences.Firstly,the contributors’ information characteristics are divided into three categories.Then,an improved weighted principal component analysis (WPCA) method was proposed to group the contribution characteristics of volunteers into normalized,weighted and dimensionally reduction.The contributors were clustered into four different groups using Gaussian mixture model (GMM) method.Finally,the results were compared with the principal component analysis (PCA) combined with K-Means.Through comparative analysis,it is found that the method of improved WPCA combined with GMM is better than the PCA combined with K-Means in the classification of contributors.

关 键 词：开放街道图 志愿地理信息 特征降维 贡献者分类 

分 类 号：P208[天文地球—地图制图学与地理信息工程]