| 基于标签语义关联的城市社群发现研究 |
| |
| 引用本文： | 毕崇武,叶光辉,彭泽,李明倩.基于标签语义关联的城市社群发现研究[J].现代情报,2009,40(3):14-23. |
| |
| 作者姓名： | 毕崇武 叶光辉 彭泽 李明倩 |
| |
| 作者单位： | 华中师范大学信息管理学院, 湖北 武汉 430079 |
| |
| 基金项目： | 国家自然科学基金项目"基于标签语义挖掘的城市画像计算与应用模型研究"（项目编号：71804055）；湖北省自然科学基金"基于社会化标签挖掘的智慧城市‘印象云’构建模式研究"（项目编号：2018CFB387）；中央高校基本科研业务费项目"基于社会化标签挖掘的城市画像研究"（项目编号：CCNU18QN040）。 |
| |
| 摘 要： | 目的] 目前社会认知在城市管理决策中的支撑权重还未充分体现，本文试图建立基于社会认知驱动的城市社群发现模式。方法] 首先利用LDA主题模型提取不同城市的城市特征属性，并利用情感强度计算方法对城市特征属性赋值；然后基于城市特征之间的共现关系构建城市网络，以此刻画城市之间的关联关系；最后依据网络拓扑关系和节点语义信息，计算城市之间的亲疏程度，挖掘具有重叠特性的城市社群。结果] 抓取知乎平台中描述我国省会城市主要特征的用户问答内容，按照上述方法依次抽取城市特征，构建城市网络，测算城市亲疏，最终挖掘出9个具有较强独立性的城市社群。局限] 未能引入分面组织与分析方法将城市特征多粒度化，进而探索不同特征粒度下城市之间的关系类型及强度。结论] 该方法能够依据社会认知发掘城市之间的潜在关联及其社群结构，有利于辅助现有城市社群划分方法并强化城市社群内涵。
|
| 关 键 词： | 标签语义关联 城市社群 城市网络 用户标签 城市画像 LDA主题模型 省会城市 用户问答 知乎 数据挖掘 |
Discovery of City Circle Based on Semantic Association of Social Tags |
| |
| Authors: | Bi Chongwu Ye Guanghui Peng Ze Li Mingqian |
| |
| Institution: | School of Information Management, Central China Normal University, Wuhan 430079, China |
| |
| Abstract: | Objective] This paper aimed to establish a public perception-driven division methods of city circles.Methods] Firstly,LDA was used to extract the city identities of different cities,and sentiment analysis was used to value city identities.Then,the city network was constructed to depict the correlation between cities,based on the coupling relationship between city identities.Finally,the relationship between cities was measured by network structures and analyzed by content semantics to discover city circles.Results] Firstly,we collected user's question-and-answer contents describing city identities of provincial capitals in China on Zhihu,an online Q&A platform.Then,we extracted city identities,constructed city network,and calculated relationship between cities.Finally,we discovered 9 city circles with strong independence in sample data.Limitations] The method of faceted organization and analysis with different granularities was not introduced to explore the relationship between cities by multi-granularity of city identities.Conclusions] This method could discover the potential relationship between cities and their circle structures according to public impression,redefine the criteria for the division of city circle,and strengthen the connotation of city circle. |
| |
| Keywords: | semantic association of social tags city circle city network social tags city profile LDA provincial capitals user's question-and-answer Zhihu data mining |
|
| 点击此处可从《现代情报》浏览原始摘要信息 |
| 点击此处可从《现代情报》下载免费的PDF全文 |
