基于多源兴趣点的模糊地名空间范围划分方法

Delimitating the Vague Place Name Bound Using Multiple-Sources Points of Interest

由于人类对现实世界进行认知和概念化的过程存在模糊性,许多人们在日常生活中使用的地名往往是没有明确边界范围的模糊区域。大数据时代的开启,为模糊地名空间范围的确定与表达提供了新思路。本文提出由k最邻近(k NN)离群点检测算法结合高斯混合模型(GMM)的方法,基于多源兴趣点(POI)数据获取模糊地名空间范围边界。该方法具有能有效识别离群点数据、参数敏感度低的特点。最后,分析了多源POI数据的应用对结果的影响。

The process of cognizing and conceptualizing the world suffers from vagueness, therefore vernacular place names used in daily life may not correspond to an formal designated region or place. Nowadays, the applications of Big Data supply the new chances for delimitating vague place name. This paper presents a method for defining the boundaries of vague place name using multi-source points of interest, which combines kth nearest neighbor outlier detection algorithm and Gaussian mixture model. With low sensitivity of parameters, the method effectively eliminates POIs that may not lie in the range of the place names. Finally, it is discussed that the results are effected by multi-sources data.