中国省情综合地图集的内容图谱构建与主题表达可视分析

省情综合地图集使用专题地图对省域的资源禀赋和社会发展水平进行全面展示，是一项复杂的知识系统。通过对地图集中大量、复杂、非结构化的内容进行模式化重构，建立地图集的内容图谱，挖掘中国省情综合地图集的内容组织规律。首先，构建图集词汇向量和计算语义相似度，提取图集内容表达的标准主题词；然后，结合图集的“图组→图幅→指标”编排结构，对主题词进行语义层次聚类，构建图集内容表达的树型图谱；在此基础上，挖掘各省综合地图集内容图谱的频繁子图，形成图谱指纹，以此标识出中国省情综合地图集主题表达的共性特征。研究结果表明，中国省情综合地图集的主题内容具有层次化组织特点，可将指纹图谱作为框架，指导省情综合地图集的主题选择和内容组织。借助内容图谱和指纹图谱分析，还可进一步揭示出中国各省综合地图集在内容表达上具有明显的聚类特征和多样性特征。研究结果可为新编省情综合地图集设计提供依据。未来还可进一步探究不同类型地图集的内容图谱构成，丰富地图集设计与编制理论。

Construction of Content Tupu and Visual Analysis Subject Expression for Provincial Comprehensive Atlases in China

  Objectives  Provincial comprehensive atlas is a complex knowledge system. It uses thematic maps to comprehensively display the resource endowment and social development level of a province. We reconstruct the large, complex and unstructured contents of the atlas to establish the atlas content tupu, explore the basic content organization rules of provincial comprehensive atlas in China.  Methods  First, we construct the vocabulary vectors of thematic subjects presented in atlas, calculate their semantic similarities, and extract the standard expressions of subject words. Second, following the compile order of "map group, map sheet, and cartographic index", the semantic hierarchical clusters of subject words are figured out, and the content map of atlas are constructed in a tree-wised graph. Finally, the frequent subgraphs of content tupu among provincial atlases are examined to form an atlas fingerprint to identify the common features of subject expressions in the comprehensive atlas of China's provinces.  Results  It is shown that the thematic contents of provincial comprehensive atlas are organized hierarchically, and the fingerprint tupu can be used as a framework to guide the thematic selection and content organization during the comprehensive atlas compilation. With the benefit of content tupu and fingerprint tupu, it can be further revealed that the provincial comprehensive atlas in China has obvious clustering characteristics and diversity characteristics in content expression.  Conclusions  The research results provide a basis for the design of provincial comprehensive atlas. In the future, we will explore the content tupus of different types of atlas to enrich the theory of atlas design and compilation.