基于政务大数据的市民公共服务需求挖掘 ——以成都市为例
Mining Citizens’ Public Service Demands Based on Government Big Data: A Case Study of Chengdu
鱼 宁
成都市规划设计研究院 规划师，yn96328@163.com
张晓荣
成都市规划设计研究院 主创规划师
于儒海
成都市规划设计研究院 副总工程师
周 垠
成都市规划设计研究院 总工助理
肖竹韵
成都市规划设计研究院 主任规划师
摘要: 公共服务需求挖掘是提高公共服务水平的基础支撑，为社会精细化治理提供行动依据。当前，以政务热线为代表的政 务大数据富含高价值信息，为实时洞察市民诉求提供了全新渠道。聚焦政务热线数据，重构公共服务需求分类体系，利 用基于BERT的文本分类模型，实现公共服务需求的精细化识别。以成都市域为案例进行实证分析。结果表明，不同类 别、不同区域、不同时间的公共服务需求差异明显，表现出“医疗教育需求多”“中心区域聚集”“周期性波动”的特征。 研究结果对推动新型城市治理体系建设有重要意义。
Abstract: Mining public service demands serves as foundational support for improving public service levels and provides actionable insights for refined social governance. Currently, government big data represented by government hotlines contains highvalue information, providing a new channel for real-time insight into citizens’ demands. This study focuses on government hotline data, reconstructs the classification system for public service demands, and utilizes a BERT-based text classification model to achieve fine-grained identification of public service demands. Through empirical analysis in Chengdu, the results indicate that public service demands vary significantly across categories, regions, and time periods, exhibiting characteristics such as high demand for healthcare and education, centralized distribution in core urban areas, and periodic fluctuations. The results are of great significance for promoting the construction of a new urban governance system.
关键词：公共服务；政务热线；BERT；文本分类；成都市
Keyword: public services; government hotline; BERT; text classification; Chengdu
中图分类号：TU984
文献标识码: A
苗大雷，安雯晴. 城市社区精细化治理的多重内
涵与优化路径——基于“人本—技术—制度”
三维框架的分析[J]. 新疆社会科学，2023（5）：
163-173.
Miao Dalei, An Wenqing. The multiple connotations
and optimization paths of refined governance in
urban communities—an analysis based on the threedimensional framework of “people-centered—
technology—institution”[J]. Xinjiang Social
Sciences, 2023(5): 163-173.
刘跃，吴康，胡钰霆，等. 基于典型人群需求的公
共服务设施精细化治理——以北京“回天地
区”一刻钟社区服务圈为例[J]. 上海城市规划，
2023（6）：98-104.
Liu Yue, Wu Kang, Hu Yuting, et al. Refined
governance of public service facilities based on
the needs of typical groups—a case study of the
15-minute community service circle in Beijing’s
“Huitian Area”[J]. Shanghai Urban Planning
Review, 2023(6): 98-104.
申明锐，尹介琪，乔艺波. 基于时空大数据的县
域公共服务设施配置评估与调控方法[J]. 城市
发展研究，2024，31（7）：86-96.
Shen Mingrui, Yin Jieqi, Qiao Yibo. Evaluation and
regulation methods for county public service facility
allocation based on spatio-temporal big data[J].
Urban Development Studies, 2024, 31(7): 86-96.
王晓晖，李振锋. 市民需求侧智慧城市建设的现
状分析与对策研究[J]. 价值工程，2023，42（7）：
11-13.
Wang Xiaohui, Li Zhenfeng. Analysis of the current
situation and countermeasure research on smart
city construction from the citizen demand side[J].
Value Engineering, 2023, 42(7): 11-13.
孔祥夫，董波，徐可，等. 基于BERT的民生问题
文本分类模型——以浙江省政务热线数据为例
[J]. 北京大学学报（自然科学版），2023，59（3）：
456-466.
Kong Xiangfu, Dong Bo, Xu Ke, et al. A BERT-
based text classification model for people’s livelihood issues—a case study of government hotline
data from Zhejiang Province[J]. Journal of Peking
University (Natural Science Edition), 2023, 59(3):
456-466.
吴坚平，陈超超，金加和，等. 基于联邦学习的政
务大数据平台应用研究[J]. 大数据，2024，10
（3）：40-54.
Wu Jianping, Chen Chaochao, Jin Jiahe, et al.
Research on the application of federated learning
in government big data platforms[J]. Big Data,
2024, 10(3): 40-54.
孙宗锋，赵兴华. 网络情境下地方政府政民互动
研究——基于青岛市市长信箱的大数据分析[J].
电子政务，2019（5）：12-26.
Sun Zongfeng, Zhao Xinghua. Research on
government-citizen interaction of local governments
in the network context—big data analysis based
on the mayor’s mailbox in Qingdao City[J].
E-Government, 2019(5): 12-26.
赵娟，张小劲，袁振龙. 政务热线驱动城市治理
现代化的理论框架与体系构建[J]. 北京行政学
院学报，2025（2）：87-97.
Zhao Juan, Zhang Xiaojin, Yuan Zhenlong.
Theoretical framework and system construction
for urban governance modernization driven
by government hotlines[J]. Journal of Beijing
Administrative College, 2025(2): 87-97.
Peng X, Li Y, Si Y, et al. A social sensing approach
for everyday urban problem-handling with the
12345-complaint hotline data[J]. Computers,
Environment and Urban Systems, 2022, 94: 101790.
王昊，邓三鸿，朱立平，等. 大数据环境下政务数
据的情报价值及其利用研究——以海关报关商
品归类风险规避为例[J]. 科技情报研究，2020， 2（4）：74-89.
Wang Hao, Deng Sanhong, Zhu Liping, et al.
Research on the intelligence value and utilization
of government data in the big data environment—
a case study of risk avoidance in customs declaration commodity classification[J]. Scientific and
Technical Information Research, 2020, 2(4): 74-89.
官国宇，杨皓翔，王运豪，等. 基于误分类修正
的朴素贝叶斯分类器及其在政务热线行业分
类中的应用[J/OL]. 数理统计与管理，2004：
1-12. （2024-04-23）[2025-01-02]. https://doi.
org/10.13860/j.cnki.sltj.20231215-001.
Guan Guoyu, Yang Haoxiang, Wang Yunhao, et al.
A Naive Bayes Classifier based on misclassification
correction and its application in government
hotline industry classification[J/OL]. Journal of
Mathematics, Statistics and Management, 2004:
1-12. (2024-04-23) [2025-01-02]. https://doi.
org/10.13860/j.cnki.sltj.20231215-001.
胡文烨，郭文涛，李振业，等. 基于NLP的政企类
文本智能分类的实现与应用[J]. 电子技术与软
件工程，2022（16）：253-258.
Hu Wenye, Guo Wentao, Li Zhenye, et al. Implementation and application of intelligent classification
for government-enterprise texts based on NLP[J].
Electronic Technology & Software Engineering,
2022(16): 253-258.
刘睿珩，叶霞，岳增营. 面向自然语言处理任务
的预训练模型综述[J]. 计算机应用，2021，41
（5）：1236-1246.
Liu Ruiheng, Ye Xia, Yue Zengying. A survey of
pre-trained models for natural language processing
tasks[J]. Journal of Computer Applications, 2021,
41(5): 1236-1246.
陈钢. 融合RoBERTa和特征提取的政务热线工
单分类[J]. 计算机与现代化，2022（6）：21-26.
Chen Gang. Classification of government hotline
work orders by integrating RoBERTa and feature
extraction[J]. Computer and Modernization,
2022(6): 21-26.
He J, Zhang W, Yang M. The spatial and temporal
characteristics of urban public safety under the
residents’ complaints: evidence from 12345 data in
Beijing, China[J]. Journal of Urban Management,
2024, 13(2): 217-231.
杨辰，辛蕾，欧阳宏，等. 基于主客观评价的社区生
活圈公共服务设施优化策略研究——以南宁市
中心城区为例[J]. 上海城市规划，2024（1）：9-16.
Yang Chen, Xin Lei, Ouyang Hong, et al. Research
on optimization strategies for public service facilities
in community life circles based on subjective and
objective evaluations—a case study of the central
urban area of Nanning City[J]. Shanghai Urban
Planning Review, 2024(1): 9-16.
杜晓娟，王文静，金刚. 生活圈视角下的中小城
市公共服务规划对策研究[J]. 西部人居环境学
刊，2021，36（1）：66-73.
Du Xiaojuan, Wang Wenjing, Jin Gang. Research
on public service planning strategies for small and
medium-sized cities from the perspective of life
circle[J]. Journal of Human Settlements in West
China, 2021, 36(1): 66-73.
黄施洋，奚雪峰，崔志明. 大模型时代下的汉语
自然语言处理研究与探索[J]. 计算机工程与应
用，2025，61（1）：80-97.
Huang Shiyang, Xi Xuefeng, Cui Zhiming. Research
and exploration on Chinese natural language
processing in the era of large models[J]. Computer
Engineering and Applications, 2025, 61(1): 80-97.
张晓荣，周垠，张娜，等. 基于多源数据和机器学
习的人口空间化研究——以成都市为例[J]. 地
域研究与开发，2024，43（5）：173-180.
Zhang Xiaorong, Zhou Yin, Zhang Na, et al.
Research on population spatialization based on
multi-source data and machine learning—a case
study of Chengdu City[J]. Areal Research and
Development, 2024, 43(5): 173-180.
孙凯丽，罗旭东，罗有容. 预训练语言模型的应用
综述[J]. 计算机科学，2023，50（1）：176-184.
Sun Kaili, Luo Xudong, Luo Yourong. A survey
on applications of pre-trained language models[J].
Computer Science, 2023, 50(1): 176-184.
Kenton J D M W C, Toutanova L K. Bert: pretraining of deep bidirectional transformers for
language understanding[C]//Proceedings of naacLHLT. 2019.
LeCun Y, Bottou L, Bengio Y, et al. Gradientbased learning applied to document recognition[J].
Proceedings of the IEEE, 1998, 86(11): 2278-2324.
侯小培，高迎. 卷积神经网络CNN算法在文本分
类上的应用研究[J]. 科技与创新，2019（4）：
158-159.
Hou Xiaopei, Gao Ying. Research on the application
of convolutional neural network CNN algorithm
in text classification[J]. Science and Technology &
Innovation, 2019(4): 158-159.
孙刘成，黄润才. 融合LSTM和注意力机制的新
闻文本分类模型[J]. 传感器与微系统，2022，
41（9）：38-41.
Sun Liucheng, Huang Runcai. A news text classification model integrating LSTM and attention mechanism[J]. Transducer and Microsystem Technologies,
2022, 41(9): 38-41.
曾子明，万品玉. 基于双层注意力和Bi-LSTM
的公共安全事件微博情感分析[J]. 情报科学，
2019，37（6）：23-29.
Zeng Ziming, Wan Pinyu. Sentiment analysis of
public safety event microblogs based on doublelayer attention and Bi-LSTM[J]. Information
Science, 2019, 37(6): 23-29.
