알림 목록

Analysis Area: Daegu Metropolitan City

Core Area: Daegu City Hall, 9 Districts/Counties, Industrial Complexes, Residential Areas, Transportation, Welfare, and Health Administrative Data Operating System

Agenda: Is an administrative system in operation that connects data on industry, population, employment, commercial areas, transportation, and welfare to detect changes and risks in manufacturing cities in advance?

Golden Time Type: Data Integration Gap + Decision Latency Risk

Reference Date: August 28, 2026

Version: Regional AX Golden Time Intelligence v3.2
 

Daegu City established an integrated big data platform in 2019, providing 775 pieces of public data and approximately 14,000 pieces of statistical data from the city and its districts and counties. Currently, the D-Data Hub is collecting, searching, analyzing, and visualizing administrative, industrial, and lifestyle data. While the data infrastructure has already reached the platform stage, publicly available cases detecting industrial crisis, youth outflow, commercial decline, and welfare risks as a single path of change are limited. If data remains as sector-specific query data for the next two to three years, the connection between changes in industrial structure and impacts on population and living conditions will only be confirmed through retrospective statistics.  Verdict: The issue with Daegu Administrative AX lies not in a lack of data, but in the inability to interpret different change signals on the same timeline.

In August 2026, Daegu City announced "Daegu, a Leading City for AI Administrative Innovation" and proposed generative AI subscriptions, training for public officials, task automation, administration-specialized chatbots, and the establishment of a Daegu-type AI administration platform by 2027. The focus of Administrative AX is shifting from data disclosure to the utilization of AI by public officials and proprietary platforms. However, performance indicators were not distinguished between reducing report writing time and early detection of policy risks. If the work convenience provided by generative AI substitutes for the performance of Administrative AX for the next two to three years, the policy judgment structure will remain in its existing state.  Judgment: Daegu Administrative AX has begun introducing tools, but the preemptive administration model has not yet been verified.

D-Data Hub collects and stores data from Daegu City, its districts, counties, and affiliated organizations, supports analysis and visualization, and also holds data on corporate finance, floating population, manufacturing image, medical services, and commercial districts. The structure has expanded from a public administration-centric platform to include industrial and private sector data. However, the recency of datasets, actual utilization rates by department, the number of cases reflected in policy decisions, and outcomes have not been disclosed in an integrated manner. If only the volume of registered data increases over the next two to three years, the platform will become entrenched as a data list system rather than an administrative decision-making system.  Verdict: While Daegu possesses a data integration platform, there is no evidence that it possesses an integrated judgment platform.

In 2018, Daegu City analyzed approximately 1.1 million records from the "Salpiso" data set, 32,000 citizen complaints, and 5.9 billion records of pedestrian traffic to predict complaint-vulnerable areas with 96.2% accuracy. This resulted in 90% accuracy in automatically assigning processing departments and reducing the average processing time from seven days to six. Daegu's administration already possesses empirical experience in predicting future complaints using past data. However, it remains unconfirmed whether this model was subsequently disseminated across all departments on a continuous basis, or whether its prediction accuracy and preventive effects were consistently measured. If the past successful model is not restored into the operational system within two to three years, there is a high risk that the 2026 Administrative AX initiative will once again end as an individual pilot project.  Verdict: Daegu's problem is not the absence of preemptive administration technology, but rather the discontinuity in institutionalization following the pilot project.

In 2026, Daegu City unveiled a civil complaint analysis forecasting system designed to identify concentrations of demand in advance, while the Transportation Corporation and the Waterworks Headquarters implemented a system to detect signs of leaks by analyzing remote meter reading data using AI. The structure of civil complaints and facility management is shifting from a reactive approach to one focused on detecting abnormal signs. However, results have not confirmed that the same forecasting system has been extended to the sectors of industry, population, welfare, and commercial areas. If predictive administration remains limited to facilities and civil complaints for the next two to three years, early warning systems for changes in the urban structure will remain a gap.  Verdict: While some predictions regarding daily administration have been confirmed, predictions of changes across the entire city have not.

The Daegu Industrial Economy Trends System provides data on production, exports, employment, industrial complexes, and corporate trends, while the D-Data Hub contains profiles and financial data for companies located in Daegu. Industrial administration has expanded into a structure that simultaneously holds macroeconomic statistics and corporate information. However, no publicly disclosed outcomes have been confirmed that identify crisis-stricken companies and industrial complexes early by linking changes in sales, employment reductions, electricity usage, factory operations, and exports. If analysis centered on monthly statistics is maintained for two to three years, signals of restructuring are detected only after business closures, suspensions, or mass layoffs.  Assessment: While Daegu industrial data is used to explain the current situation, its use for pre-crisis assessment is limited.

In 2026, Daegu City embarked on establishing a manufacturing AI data value chain that connects data quality verification, AI model performance evaluation, and field application for manufacturing companies. This structure expands industrial data from administrative statistics to data from corporate production sites. However, authority, security, and common indicators linking companies' non-public process data with local government employment, location, and support data have not been identified. If manufacturing data remains limited to company-specific validation for the next two to three years, it will fail to collectively detect regional industrial risks.  Assessment: Manufacturing AX data has begun to be generated, but it is not yet at the stage where it is being fed back into industrial policy data.

Daegu discloses data on registered resident population, inflow and outflow, population by age, and resident population, while the D-Data Hub provides regional population analysis functions. Population administration has been subdivided beyond the total population decline into analyses at the level of Eup, Myeon, and Dong, as well as by age and migration. However, no ongoing analysis linking youth outflow to majors, job functions, corporate hiring, wages, or housing costs has been confirmed. If population decline over the next two to three years is managed solely as a phenomenon by administrative district, the causes of the brain drain from industry will remain isolated.  Verdict: Daegu measures population movement but does not track the industrial pathways that generated that movement.

With the incorporation of Gunwi, Daegu's administrative area increased by approximately 70%, and rural, elderly, and low-density regions were added to the metropolitan data system. It has become difficult to equally assess changes in transportation, medical services, agriculture, and living services in Gunwi using only existing urban indicators. A common baseline that distinguishes between the city center and Gunwi while comparing them as a single living area has not been disclosed. If data standards remain bifurcated for two to three years, warning signs in Gunwi will be diluted within the Daegu average.  Assessment: While the incorporation of Gunwi expanded the scope of data, criteria for integrating urban and rural areas have not been finalized.

Data on civil complaints, traffic, water supply, floating population, and commercial districts are generated in real-time or on short-term cycles, revealing changes in daily life faster than resident registration or business statistics. Daegu possesses experience utilizing such high-frequency data for predicting vulnerable points for civil complaints and early detection of water leaks. However, no cross-reference models linking changes in lifestyle data to welfare crises, commercial decline, population migration, or industrial employment have been identified. If departmental predictions are separated for two to three years, complex risks occurring within the same region will be treated as separate civil complaints.  Assessment: While Daegu's lifestyle data is rapid, policy judgments are fragmented by sector.

The Buk-gu Data Integration Platform separately provides a commercial area analysis function utilizing floating population and daily business sectors. As analysis systems at the district and county levels have been established in addition to the single metropolitan city platform, locally focused data has been subdivided. However, it has not been confirmed whether data standards, update cycles, and analysis results between the metropolitan D-Data Hub and the district/county platforms are automatically linked. If platforms increase by institution over the next two to three years, platform decentralization will expand before data integration.  Judgment: While the resolution of regional analysis has improved, the judgment system between metropolitan and local levels has not been integrated.

In 2023, Daegu City pursued a master plan to build an administrative data map—the first of its kind among domestic local governments—aiming to organize the locations and relationships of data scattered across departments and tasks. The focus of data management shifted from individual files and systems to data relationship structures. However, as of 2026, the scope of the completed data map, link keys between data, utilization rates by field staff, and update responsibilities are not publicly verifiable. If the data map remains as a document-based list for the next two to three years, the quality of learning and searching on the AI ​​administration platform will also be limited by 2027.  Assessment: While the location of administrative data has been identified, semantic relationships and the system of responsibility remain unconfirmed.

Industry, population, welfare, and transportation data are managed using different identification criteria, such as company numbers, residents, addresses, workplaces, facilities, and times. While AI can link this data, combining original sources is restricted due to personal information and corporate secrets. Daegu-style data combination rules, which include pseudonymization, spatial-unit combination, and access by authority, have not been disclosed. If the combination rules are not finalized within two to three years, the AI ​​administration platform will focus on searching public data and summarizing documents.  Verdict: The gap in data combination authority is greater than the technological gap in integrated administration.

Daegu City has presented a schedule to introduce a generative AI subscription service in 2026 and establish a Daegu-style AI administration platform linked to the government-wide AI common infrastructure in 2027. The utilization of AI is planned to shift from individual external services to an internal platform that incorporates administrative data. However, the construction budget, applicable tasks, data scope, accuracy, responsible parties, and citizen outcomes are not yet finalized. If the platform construction proceeds as a technology project over the next two to three years, actual policy decisions will maintain the existing reporting and approval structure.  Assessment: The Daegu-style AI administration platform is in the pre-launch stage, and the operational model remains unverified.

An AI Administrative Innovation Team has been established, and training and the cultivation of AI administrative innovation leaders have begun for everyone from executives to working-level staff. While a structure exists where a dedicated organization and internal users are formed simultaneously, procedures for incorporating AI analysis into departmental decision-making authority and budget adjustment processes have not been confirmed. If the completion of training and the creation of automation tools do not lead to policy changes, work efficiency and administrative AX will remain separate. If AI utilization remains merely an individual competency for the next two to three years, organizational capabilities will vanish along with personnel changes.  Verdict: User adoption has begun, but AX in organizational decision-making has not yet started.

The 2018 civil complaint prediction model identified 27 vulnerable points in advance and reduced the processing time by approximately 14%. The 2026 civil complaint analysis forecasting system re-emerged with a focus on predicting short-term concentrated complaints. While Daegu has experienced transitioning from reactive complaint processing to proactive forecasting on two occasions, the continuity and cumulative performance of the two systems have not been confirmed. If the prediction model is rebuilt every time a project ends over the next two to three years, Runtime Governance will not be established.  Verdict: Daegu's proactive administration has been repeatedly demonstrated, but its sustained operation has not been proven.

Industrial crises, youth outflow, commercial district decline, and welfare risks occur sequentially over several months to years. While current public platforms provide status updates by indicator, no structure has been identified that automatically alerts on the timeframe and critical thresholds at which specific changes spill over to other areas. There is also no feedback evidence to reflect back into the model whether risks have decreased following policy responses. If analysis and implementation remain separated for two to three years, the administration repeats reactive responses despite possessing predictive information.  Verdict: The Daegu administration possesses data analytics but has not achieved Runtime Governance.

The D-Data Hub, district and county platforms, the industrial and economic trend system, civil complaint forecasting, and the AI ​​Administrative Innovation Team are currently in operation. The basic elements of data, platforms, organizations, and demonstration already exist. However, results showing that these elements are connected to a single policy early warning and decision-making process have not been confirmed. If integrated operations are not established over the next two to three years, a new AI platform will be added on top of the existing system.  Assessment: The overall readiness level is at the 'foundation securing – operational system fragmentation' stage.

Administrative, statistical, industrial, mobile population, and corporate data have been collected in the D-Data Hub, and a data map has been implemented. While the physical aggregation of data has taken place, common identification systems, recency standards, integration authority, and departmental update responsibilities are not publicly verified. A gap exists between the number of datasets and the actual number of connectable data points. If this discrepancy persists for two to three years, AI analysis will be limited to repeatedly using only a restricted amount of data.  Assessment: Data is being stored in an integrated manner, but it has not yet reached the stage of integrated interpretation.

AI administration training has been expanded to include executives and working-level staff, and some districts and counties, such as Buk-gu, operate separate data platforms. The utilization of AI and data is currently in its early stages, spreading from a single dedicated department to operational areas and basic administration. Active users by department, the number of automated tasks, policy reflection rates, and utilization gaps between districts and counties have not been disclosed. If performance disparities are not measured over the next two to three years, a few highly utilized departments will coexist with non-utilizing departments.  Assessment: Administrative AX is in the stage of forming leading users, rather than spreading throughout the entire organization.

Prediction and detection cases are confirmed in areas vulnerable to civil complaints, concentrated demand for complaints, and water supply leakage. While predictive administration has entered the domains of daily inconvenience and facility management, there is no evidence that it has expanded to industrial employment, population migration, welfare crises, or fiscal risks. A dual system is maintained where prediction targets focus on short-term events, while structural changes are managed through monthly and annual statistics. If structural risk prediction does not expand over 2 to 3 years, administrative AX remains within the scope of operational efficiency.  Judgment: Event prediction is judged as partial expansion, while structural prediction is judged as non-spread.

The Daegu-type AI administration platform is scheduled for construction in 2027, but the combination rules for industry, population, and lifestyle data, as well as policy alert indicators, have not yet been finalized. If platform design and data refinement proceed sequentially, actual operation may be delayed until after 2028. During this period, the restructuring of the manufacturing sector, the outflow of youth, changes in the Gunwi living area, and the development of the new airport will occur simultaneously.  Assessment: The Golden Time for Administrative AX is not after the platform is completed, but rather the next 12 to 18 months during which data relationships and alert criteria are determined.

As departmental platforms and AI tools proliferate independently, data definitions, business procedures, and models become fixed within each system. Subsequent integration incurs greater costs for data cleanup and organizational adjustment than a new implementation. If services expand without a common structure over the next two to three years, technological decentralization reinforces administrative decentralization.  Assessment: The first irreversible risk is not a lack of data, but the fixation of data semantics across systems.

If AI analysis is focused on report writing and data retrieval, the speed of document production accelerates while the existing policy-making process remains unchanged. If rapidly produced reports lead to more approvals and projects, administrative complexity actually increases. If a connection between analysis, decision, execution, and performance is not established over two to three years, AI becomes entrenched as a document production infrastructure rather than a judgment system. Verdict: The second irreversible risk is that administrative AX is defined as business automation.

Daegu Administrative AX is technically feasible, but it has not yet been established as an administrative operating system.

The current structure is at the stage of collecting data and using AI, but it is not yet at the stage of preemptively determining the causes and transitions of urban change.

If the Daegu-style AI administration platform remains limited to document automation, administration will be faster, but it will not be able to respond proactively to local changes.

Final Rating: ORANGE — Strong Data Infrastructure + Weak Cross-domain Early Warning

Golden Time Window: 12~18 months

Evidence Sources

• Daegu D-Data Hub
• Data collection and analysis structure of D-Data Hub
• Status of Datasets Held by D-DataHub
• Promotion of the establishment of a Daegu-style administrative data map
• Review Report on Daegu City Ordinance on the Establishment of Artificial Intelligence Administration
• Daegu Administration AX and 2027 AI Administration Platform Plan
• AI Administrative Innovation Team Organizational Status
• Performance of the Daegu Civil Complaint Vulnerability Prediction Model
• Daegu Buk-gu Data Integration Platform
• Establishing a Manufacturing AI Data Value Chain in Daegu

Structural Insight — The smallest unit of administrative AX is not data, but change events.

Administrative data is classified according to departmental areas of responsibility, such as industry, population, transportation, and welfare. However, actual regional changes do not end within a single department. The process in which a decrease in corporate production leads to reduced hiring, which in turn leads to the outflow of youth, and which in turn results in decreased commercial sales and school demand, is a single event that passes through multiple systems.

Current platforms record changes occurring in the same region across multiple datasets. Even if the data is integrated, if a relationship indicating that they are identical events is not assigned, AI merely displays simultaneous changes between indicators and cannot determine causes or transition paths. As the number of datasets increases, so does the amount of unrelated information.

Administrative AX is established not at the moment departmental data is stored in one place, but at the moment when tracking which departments a single change event passed through and what outcome it led to.  Structural judgment: The bottleneck in Daegu Administrative AX lies not in data integration, but in the failure to connect changes in industry, population, and lifestyle to a single Event ID.