The geometric double-entity model method (GDEM) is proposed to gives far-near degrees between clusters (namely, GE distances), which integrate the absolute distance between nearest sample sets and the dense degrees of border regions of two clusters. GE distances are measured based on the clustering solutions of a clustering algorithm (e.g., Affinity Propagation), and have the advantage: GE distances can reveal Far-Near Relations of Clusters in the original data space, espeacially for the high-dimensional data.