This paper puts forward a new approach for analyzing the problems of urban public management based on the data collected from “12345” citizen service hotline. The research sets up the geocoding by the extraction of the spatial information from the call logs of Sanya City, Hainan Province. Meanwhile, the characteristics of time, space and categories of inquiries are documented and depicted by combining the original information of hotline records with the geocodes. By using high frequency word distribution and correlation network, the research reveals the major problems of cities, and analyzes the featured spatial-temporal distribution patterns of corresponding problems. The factors which affect the occurrence of urban problems are discussed with the combination of other urban characteristics including various types of point of interest (POI) representing urban functions and road network representing urban structure. The results are as follows. 1) The frequency of calls fluctuates on a weekly basis, and the total number of records in midweeks is larger than that of the weekends. 2) The hottest five categories of urban problems are noise, construction, communication and network, urban water use and parking issues, among which the noise and construction problems are closely correlated. 3) Different urban problems undergo distinct temporal patterns, while all of them are distributed in urban activity-intensive areas. 4) The occurrence of urban problems is not significantly relative with the density of road network, but positively relative with the POI densities of public facilities, transportation facilities, and educational and academic facilities, and significantly negative with the POI densities of scenic spots, sports and entertaining utilities, and administrative offices. The results demonstrate the effectiveness of “12345” citizen service hotline in urban problem discovery and depiction. Furthermore, this paper proposes several pathways for the optimization of urban governance using big data. For instance, the accumulation and analysis of urban data, the integration of multi-source data, and the enhancement of the degree of data share should be seriously appreciated. It’s necessary to utilize these data to support the fine urban governance and promote the construction of smart city.

以海南省三亚市的“12345”市民服务热线数据为研究对象, 通过提取热线数据记录中的空间信息进行地理编码, 结合热线记录的原始信息, 刻画市民来电的时间、空间和类别特征。利用高频词分布及其相关性网络来归纳城市公共管理中的主要问题, 分析各类问题的时间变化特征和空间分布模式。用不同类型兴趣点(POI)反映城市功能, 用道路网反映城市结构, 探讨影响城市问题发生的因素。结果表明: 1) 热线记录以周为单位波动, 周中数量大于周末; 2) 城市问题发生数量前5类依次为噪音、建设施工、通信与网络、城市用水和停车, 且噪音与建设施工问题联系紧密; 3) 不同城市问题在时间上有不同的发生模式, 空间上集中分布在城市活动密集的区域; 4) 城市问题的发生与道路网密度关系不大, 与城市功能中的公共设施、交通设施和科教文化POI密度显著正相关, 与风景名胜、体育休闲和政府职能POI密度显著负相关。研究结果验证了“12345”市民服务热线在挖掘和刻画城市问题方面的有效性, 并提出利用大数据优化城市治理的途径(包括城市数据积累与分析、融合多源数据以及提高数据共享程度等), 对推动城市精细化治理和智慧城市建设具有迫切意义。