城市群视角下的主题发现与空间格局研究——以区域经济学文献为例  

Topic Detection and Spatial Patterns from the Perspective of Urban Agglomeration:A Case Study of Regional Economic Literature

作　　者：郭思成 李纲[1] 汪鸿沁泠 GUO Si-cheng;LI Gang;WANG Hong-qin-ling(Center for Studies of Information Resources,Wuhan University,Wuhan 430072,China)

机构地区：[1]武汉大学信息资源研究中心,湖北武汉430072

出　　处：《情报科学》2020年第10期125-132,共8页Information Science

基　　金：国家自然科学基金重大课题“国家安全大数据综合信息集成与分析方法”(71790612)。

摘　　要：【目的/意义】区域经济学领域文献是一种具有显著空间属性的科研数据。在城市群维度下对其进行主题发现可以反映出不同地理尺度下的主题分布的差异性以及不同城市群空间的研究态势。【方法/过程】以区域经济学文献为研究对象,根据著者机构映射到所属城市群空间,选取文献数量最多的5个城市群为例。以Word2Vec词嵌入思想将文献的题名、摘要信息表示成语义级别的词向量结构,并通过K-means聚类、t-sne降维结合tf-idf权重等分析方法,在更深的层面揭示不同城市群空间下的主题分布格局。【结果/结论】各城市群均有一定数量的主题分布在区域经济一体化及合作方向;各城市群中文献主题均涉及到了区域经济发展的框架性研究;各城市群内的研究对象多来自于该空间内部。【Purpose/significance】The literature in the field of regional economics is a kind of scientific research data with significant spatial attributes.Subject mining in the dimension of urban agglomeration can reflect the difference of subject distribution in different geographical scales and the research situation perception of different urban agglomeration spaces.【Method/process】Taking the literature of regional economics as the research object,five urban agglomerations with the largest amount of literature were selected as examples according to the mapping of literature institutions to the urban agglomeration space.Word embedding in Word2 Vec is used to express the title and abstract information of literature into the word vector structure at the semantic level.Through k-means clustering,t-sne dimension reduction combined with tf-idf weight and some other analysis methods,the theme distribution pattern in different urban agglomeration spaces is revealed at a deeper level.【Result/conclusion】Each city group has a certain number of themes distributed in the direction of regional economic integration and cooperation;all the literature topics in urban agglomerations involve the framework research of regional economic development.Most research objects in urban agglomerations come from the interior of the space.

关 键 词：城市群 区域经济学 机器学习 主题分布 

分 类 号：G250.2[文化科学—图书馆学]