An OSM Contributors Classification Method Based on WPCA and GMM

Abstract

Contributors have a significant impact on data quality of OpenStreetMap (OSM) because most of them are the non-professional, so clustering analysis of contributors based on different experiences has practical significance. Firstly, this paper obtained 31 behavioural characteristics of contributors from OSM historical data. Then, a weighted principal component analysis (WPCA) method was used to reduce the dimensions of the contributors’ behaviour in the selected region. By using an unsupervised prototype-based Gaussian mixture model (GMM) clustering algorithm, contributors with similar contribution attributes in the London area were clustered into four groups. Finally, the characteristics of four different types of contributors are analysed, and two types of experienced and professional contributors are found, who contribute a large amount of high-quality data.