A dual-mode big data platform for the cultural tourism industry

Abstract

This invention provides a dual-mode big data platform for the cultural, tourism, and sports industry, comprising a standardized platform, a lightweight platform, and a big data platform control system. The big data platform includes a mode control hub for switching between control modes, and employs a centralized control and distributed execution architecture with a metadata center as the core hub to ensure synchronization between the two platforms during operation and maintenance upgrades. The big data platform control system includes an intelligent operation and maintenance coordinator that schedules the standardized and lightweight platforms for operation and maintenance upgrades respectively; a data synchronization engine that uses bidirectional incremental replication technology to control data synchronization between the standardized and lightweight platforms; and a functional synchronization layer that ensures interface compatibility between the standardized and lightweight platforms through contract testing. By switching between the two types of big data platforms, the diverse needs of different cultural, tourism, and sports venues can be met, and the operation and maintenance upgrades and independent use of the two types of big data platforms do not affect each other.

Description

Technical Field

[0001] This invention belongs to the field of big data technology, and in particular relates to a dual-mode big data platform for the cultural, tourism and sports industries. Background Technology

[0002] A big data platform is a networked technology platform centered on distributed storage and computing, integrating multi-source heterogeneous data resources to provide one-stop services such as data collection, storage, processing, analysis, and visualization. Through efficient management and intelligent analysis of massive amounts of data, it supports data-driven decision optimization and business innovation across various industries. Currently, when cultural, tourism, and sports venues raise data governance and operation and maintenance needs, they typically only have one type of big data platform available for implementation. All requirements are configured and implemented on this platform, and all users use the same platform. Furthermore, if operational and maintenance issues arise, maintenance personnel need to handle the problems on a designated server within this type of big data platform. This process includes steps such as requirement definition, personnel allocation, requirement implementation, and system operation and maintenance, ultimately meeting the project's implementation and operation and maintenance requirements.

[0003] Patent CN120144676A discloses a real-time data processing method, system, device, and medium for a cultural tourism big data platform. Primarily related to the field of real-time data processing technology, it addresses the challenges of existing solutions in achieving a balance between real-time performance and accuracy, lacking deep learning and intelligent analysis capabilities, and failing to uncover the underlying patterns and potential value of data. The method includes: utilizing a pre-defined cleaning algorithm in the DWD layer of a split-layer data warehouse to remove meaningless data from original user behavior logs and ticketing data; using a pre-defined demand acquisition interface in the DIM layer of the split-layer data warehouse to obtain query conditions and associated operation interface names, extracting data matching the pre-defined query conditions from the user behavior logs and ticketing data in the DWD layer; and using the DM layer of the split-layer data warehouse to call the corresponding operation program of the associated operation interface, inputting the data matching the pre-defined query conditions into the operation program to obtain the result data. Summary of the Invention

[0004] When existing cultural, tourism, and sports venues raise data governance and operation and maintenance needs, they typically only have one type of big data platform available for implementation. However, the diverse needs of cultural, tourism, and sports venues mean that having only one type of big data platform cannot meet the optimal configuration for all requirements, impacting work efficiency. Furthermore, if only a single big data platform exists, its operation and maintenance can affect the overall project schedule. A single big data platform also creates a uniform standard for different users' operations, lacking flexibility.

[0005] To address the aforementioned technical problems, the present invention provides the following technical solution: a dual-mode big data platform for the cultural, tourism, and sports industry, comprising a standardized platform, a lightweight platform, and a big data platform control system; the big data platform includes a mode control hub for controlling mode switching, and adopts a centralized control and distributed execution architecture with a metadata center as the core hub to ensure synchronization between the two platforms during operation and maintenance upgrades; the big data platform control system includes an intelligent operation and maintenance coordinator that schedules the standardized platform and the lightweight platform to perform operation and maintenance upgrades respectively; it also includes a data synchronization engine that uses bidirectional incremental replication technology to control data synchronization between the standardized platform and the lightweight platform; and it also includes a functional synchronization layer that ensures interface compatibility between the standardized platform and the lightweight platform through contract testing.

[0006] The data synchronization engine adopts a multi-level pipeline architecture to solve the data consistency problem between the two platforms. The data synchronization engine includes a change capture module, a synchronization scheduling module, a conflict resolution module, and a data verification module. The change capture module captures change data from the standardized platform and the lightweight platform, merges them into a change set, converts the captured change set into a unified JSON Patch format, and then enters the distributed message queue Kafka. The synchronization scheduling module performs differentiated routing according to the data type of the cultural, tourism, and sports industry, and the conflict resolution module verifies the integrity and consistency of the data.

[0007] The change capture module uses the Debezium connector based on database logs to capture data changes in real time for standardized platforms; for lightweight platforms, it uses SQLite's write-ahead log listening mechanism to capture data changes.

[0008] The conflict resolution module uses a three-tier arbitration mechanism to verify the integrity and consistency of data. The basic layer processes data automatically according to preset rules, the business layer makes decisions based on timestamps and operation types, and the cultural, tourism and sports specialization layer makes decisions based on seasonal factors.

[0009] After each synchronization, the data synchronization engine performs end-to-end verification by the data verification module, ensuring consistency by comparing the hash values ​​of key business entities. When a difference is detected, an incremental repair process is automatically triggered, prioritizing the repair of real-time business data, while historical data enters the background repair queue. During periods when cultural, tourism, and sports venues are not open for business, the data verification module automatically performs full data verification to ensure that the data is completely consistent before operation.

[0010] The intelligent operations and maintenance coordinator drives the operations and maintenance upgrade process through a state machine. When an operations and maintenance upgrade is initiated, the intelligent operations and maintenance coordinator first freezes write operations on the target platform and marks the upgrade starting point with a global transaction ID. For standardized platform upgrades, user requests are automatically routed to the lightweight platform, while a shadow database is started on the standardized platform side to record changes. After the upgrade is completed, the data comparison tool calculates the differences between the primary database and the shadow database, generates incremental patches, and synchronizes them to the lightweight platform. Lightweight platform upgrades adopt a mirroring strategy, first creating a read-only copy of the standardized platform to provide services, and then performing configuration synchronization after the upgrade is completed.

[0011] The intelligent operations and maintenance coordinator implements a four-stage guarantee during the operation and maintenance upgrade process: the preparation stage performs metadata compatibility verification and resource pre-allocation; the execution stage adopts canary release, updating in batches according to the type of cultural, tourism and sports venues and user groups; the verification stage automatically runs contract tests and business scenario tests; the completion stage generates a data consistency report and unfreezes and writes it; during short-term high-traffic periods in cultural, tourism and sports venues, the intelligent operations and maintenance coordinator automatically delays non-critical upgrades; during long-term active tourist periods in cultural, tourism and sports venues, a lightweight platform priority mode is activated to ensure that the tourist experience is not affected.

[0012] The functional synchronization layer achieves interface compatibility through a contract testing and monitoring system; the metadata center stores the standardized definition metadata of all APIs. Before each upgrade, the contract testing service automatically generates test cases for typical cultural, tourism, and sports scenarios, and executes them in parallel on both platforms for verification; interface changes must pass backward compatibility checks, new parameters must define default values, and deleted parameters must be marked with a deprecation period; for major changes, a version coexistence strategy is adopted, and traffic is split between the old and new versions through the API gateway.

[0013] The big data platform's mode switching is controlled by the mode control center. When a switching command is triggered, the metadata repository of the mode control center is activated first. This repository stores the dependency graph of all modules. After parsing this graph, the metadata interpreter generates service composition commands and dynamically adjusts Istio traffic routing through the VirtualService rules of the Istio service mesh, directing requests for specific modules to the target cluster. At the same time, the resource scheduler creates elastic resource barriers, allocating namespace-level resource quotas for the lightweight mode and removing restrictions in the normalized mode, allowing access to the GPU acceleration pool. Data consistency during mode switching is ensured by the state synchronization engine, which uses the CRDT algorithm to resolve distributed state conflicts. When switching back from lightweight to normalized mode, the engine initiates an incremental checkpoint recovery mechanism: replaying key transactions during lightweight operation based on the operation logs and reconstructing the data view through a dynamic schema adapter. This adapter builds a virtualization layer on top of the data lake. In lightweight mode, columnar storage projection is automatically created, while in normalized mode, the full JSONB document is loaded. The resource allocation module dynamically adjusts resource weights based on the Prometheus metrics of the resource data during mode switching.

[0014] The standardized platform integrates a complete ETL process, data warehouse, machine learning algorithm library, and BI analysis tools. It saves the full dataset when storing data and supports post-event traceability and multi-dimensional drill-down analysis. The lightweight platform only carries the data processing and display modules. Service instances are set up on edge nodes or low-cost computing units close to the data source. It reduces network I / O and disk usage through highly compressed data transmission protocols and columnar storage projection. Resource consumption is isolated and limited based on resource quotas.

[0015] The beneficial effects of this invention are: By switching between two types of big data platforms within a single system, the diverse needs of different cultural, tourism, and sports venues can be met, and work efficiency can be improved, based on different demand scenarios. The system switching method allows maintenance and implementation personnel to work independently without affecting the stability and timeliness of venue data, or the overall project progress. Users can flexibly switch platforms according to their own circumstances and different scenarios, enhancing product flexibility and making it more personalized. Attached Figure Description

[0016] Figure 1 This is a diagram of the overall architecture of the present invention.

[0017] Figure 2 This is a flowchart of the data synchronization process of the present invention.

[0018] Figure 3 This is a flowchart of the operation and maintenance upgrade process of this invention.

[0019] Figure 4 This is a flowchart of the contract testing process for the functional synchronization process of this invention.

[0020] Figure 5 This is a diagram of the underlying switching logic architecture of the present invention.

[0021] Figure 6 This is a schematic diagram of the switching process of the present invention. Detailed Implementation

[0022] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] Example 1: This example provides a dual-mode big data platform for the cultural, tourism, and sports industries, such as... Figure 1 As shown, the system includes a standardized platform, a lightweight platform, and a big data platform control system. The big data platform includes a mode control hub for switching control modes, and adopts a centralized control and distributed execution architecture with a metadata center as the core hub to ensure synchronization between the two platforms during operation and maintenance upgrades. The big data platform control system includes an intelligent operation and maintenance coordinator, which schedules the standardized platform and the lightweight platform to perform operation and maintenance upgrades respectively. It also includes a data synchronization engine, which uses bidirectional incremental replication technology to control data synchronization between the standardized platform and the lightweight platform. The system also includes a functional synchronization layer, which ensures interface compatibility between the standardized platform and the lightweight platform through contract testing.

[0024] A big data platform is a networked technology platform centered on distributed storage and computing, integrating multi-source heterogeneous data resources to provide one-stop services such as data collection, storage, processing, analysis, and visualization. Through efficient management and intelligent analysis of massive amounts of data, it supports data-driven decision optimization and business innovation across various industries. Currently, when cultural, tourism, and sports venues raise data governance and operation and maintenance needs, they typically only have one type of big data platform available for implementation. All requirements are configured and implemented on this platform, and all users use the same platform. Furthermore, if operational and maintenance issues arise, maintenance personnel need to handle the problems on a designated server within this type of big data platform. This process includes steps such as requirement definition, personnel allocation, requirement implementation, and system operation and maintenance, ultimately meeting the project's implementation and operation and maintenance requirements.

[0025] When existing cultural, tourism, and sports venues raise data governance and operation and maintenance needs, they typically only have one type of big data platform available for implementation. However, the diverse needs of cultural, tourism, and sports venues mean that having only one type of big data platform cannot meet the optimal configuration for all requirements, impacting work efficiency. Furthermore, if only a single big data platform exists, its operation and maintenance can affect the overall project schedule. A single big data platform also creates a uniform standard for different users' operations, lacking flexibility.

[0026] Therefore, the dual-mode system of the big data platform for the cultural tourism and sports industry provided in this embodiment adopts a "centralized control + distributed execution" architecture to achieve synchronization assurance during operation and maintenance upgrades. This architecture uses a metadata center as the core hub, and a smart operation and maintenance coordinator to uniformly schedule the upgrade processes of the standardized platform and the lightweight platform. The data synchronization engine uses bidirectional incremental replication technology to establish real-time data channels between platforms, while the functional synchronization layer ensures interface compatibility through contract testing. The entire system is deployed on a containerized platform, supporting dynamic resource allocation and automatically adjusting synchronization strategies during peak tourist seasons and sporting events.

[0027] The specific implementation of data synchronization is to solve the data consistency problem between two platforms by using a multi-level pipeline architecture in the data synchronization engine. For example... Figure 2 As shown, in the change capture module, the standardized platform uses the Debezium connector based on database logs to capture data changes in real time, while the lightweight platform uses SQLite's Write-Ahead Log (WAL) monitoring mechanism. The captured change sets are converted into a unified JSON Patch format and then enter the distributed message queue Kafka. The synchronization scheduler implements differentiated routing based on the data type of the cultural, tourism, and sports sectors: real-time event data enters the high-speed channel, historical tourist records enter the batch channel, and configuration information enters the strongly consistent channel. The conflict resolution module employs a three-layer arbitration mechanism: the basic layer automatically handles conflicts according to preset rules, the business layer makes decisions based on timestamps and operation types, and the cultural, tourism, and sports-specific layer considers seasonal factors (e.g., prioritizing data from the lightweight platform during peak tourist seasons).

[0028] The data verification module performs end-to-end verification after each synchronization, ensuring consistency by comparing the hash values ​​of key business entities (such as scenic spot ticketing, sports event scoring, and cultural relic archives). When a discrepancy is detected, an incremental repair process is automatically triggered, prioritizing the repair of real-time business data, while historical data is placed in a background repair queue. During museum closures and non-operational hours at scenic spots, the system automatically performs full data verification to ensure complete data consistency before the next day's operation.

[0029] The operation and maintenance upgrade of the big data platform is driven by an intelligent operation and maintenance coordinator through a state machine. For example... Figure 3 As shown, when initiating an upgrade, the intelligent operations and maintenance coordinator first freezes write operations on the target platform and marks the upgrade starting point using a global transaction ID. For standardized platform upgrades, the system automatically routes user requests to the lightweight platform and simultaneously starts a shadow database to record changes on the standardized platform side. After the upgrade is complete, the data comparison tool calculates the differences between the primary and shadow databases, generates incremental patches, and synchronizes them to the lightweight platform. Lightweight platform upgrades employ a mirroring strategy, first creating a read-only copy of the standardized platform to provide services, and then performing configuration synchronization after the upgrade is complete.

[0030] The upgrade process is implemented in four phases: the preparation phase involves metadata compatibility verification and resource pre-allocation; the execution phase uses canary releases, updating in batches according to scenic area type and user group; the verification phase automatically runs contract tests and business scenario tests; and the completion phase generates a data consistency report and unfreezes and writes the data. During sporting events, non-critical upgrades are automatically delayed; during peak tourist seasons, a lightweight platform priority mode is activated to ensure that the tourist experience is not affected.

[0031] Functional synchronization is achieved through a contract testing and monitoring system, and the functional synchronization layer uses contract testing to ensure interface uniformity. For example... Figure 4 As shown, the metadata center stores standardized definitions for all APIs, including request / response formats, error codes, performance metrics, and other metadata. Before each upgrade, the contract testing service automatically generates test cases for typical cultural, tourism, and sports scenarios, and executes them in parallel on both platforms for verification. Interface changes must pass backward compatibility checks; newly added parameters must have default values ​​defined, and deleted parameters must be marked with a deprecation period. For major changes, the system adopts a version coexistence strategy, using an API gateway to separate traffic between the old and new versions.

[0032] In this embodiment, users can seamlessly switch modes when using the cultural tourism and sports big data platform through a three-dimensional switching channel. First, at the visual interaction layer, a mode switching matrix is ​​provided by a dynamically rendered console. When an administrator logs in, the system loads the platform configuration interface through the permission engine, presenting a selectable module tree for lightweight and standardized modes. After the user selects the required module, the front end pushes the configuration snapshot to the configuration center in real time via WebSocket, triggering the hot deployment engine to complete service reorganization within 500 milliseconds. For special roles such as mobile inspectors, the system integrates a context-aware engine that automatically switches to the appropriate mode by monitoring device type, network bandwidth, and business scenario in real time, combined with a machine learning model based on historical operation records. For example, when a sports venue is detected to be hosting a large-scale event, the congestion warning module in the standardized mode is automatically activated; while when scenic area staff use 4G networks for inspection, the mode is downgraded to lightweight mode. All user requests are dynamically routed at the API gateway layer. The gateway distributes requests to the corresponding clusters by parsing the "X-Platform-Mode" flag in the request header or through intelligent traffic analysis, achieving seamless business switching.

[0033] The underlying switching logic between lightweight and standardized platforms is built upon a four-layer collaborative architecture. For example... Figure 5As shown, when a switching command is triggered, the metadata repository of the control center is activated first. This repository stores the dependency graph of all modules. After parsing this graph, the metadata interpreter generates service composition commands and dynamically adjusts Istio traffic routing through the VirtualService rules of the service mesh, directing requests for specific modules to the target cluster. At the same time, the resource scheduler calls the Kubernetes API to create elastic resource barriers, allocating namespace-level resource quotas for lightweight mode, while removing restrictions and allowing access to the GPU acceleration pool in normalized mode.

[0034] Data consistency during the switch between the lightweight and standardized platforms is ensured by a state synchronization engine, employing the CRDT algorithm to resolve distributed state conflicts. When switching back from lightweight to standardized mode, the engine initiates an incremental checkpoint recovery mechanism: replaying critical transactions during the lightweight operation based on the operation logs and reconstructing the data view through a dynamic schema adapter. This adapter builds a virtualization layer on top of the data lake; in lightweight mode, it automatically creates a columnar storage projection, while in standardized mode, it loads the full JSONB documents. The resource allocation module dynamically adjusts resource weights during mode switching by monitoring Prometheus metrics from a third-party system in real time.

[0035] The lightweight big data platform represents the operational state of this system designed to address specific high-frequency, real-time, and mobile scenarios. From a technical perspective, it is not a standalone, simplified version of the system, but rather a "minimum functional set" dynamically extracted from the standardized platform through a sophisticated filtering and optimization mechanism. Its characteristics include high agility and economy: functionally, it only carries the core data processing and display modules, such as real-time crowd monitoring, ticket gate data statistics, and emergency alarm notifications, eliminating all time-consuming batch calculations and complex model prediction functions; architecturally, its service instances are scheduled to edge nodes or low-cost computing units closer to the data source, employing highly compressed data transmission protocols and columnar storage projection to significantly reduce network I / O and disk usage; in terms of resource consumption, it uses strict resource quotas for isolation and limitation, typically requiring only a fraction of the computing resources of the standardized platform for stable operation. From an application perspective, the lightweight mode is primarily geared towards frontline operators (such as scenic area inspectors and event site managers), prioritizing the timeliness of information and system availability, ensuring a smooth user experience on mobile devices such as smartphones or tablets.

[0036] A standardized big data platform represents the "complete" form of the system, a concentrated embodiment of its full data processing capabilities and intelligence level. Technically, it integrates a complete ETL process, data warehouse, machine learning algorithm library, and BI analysis tools, capable of performing a range of complex tasks from mining massive historical data and multi-dimensional deep analysis to AI model training and future trend prediction. Its architecture prioritizes throughput and computational depth, typically relying on powerful centralized cloud computing resources and potentially involving GPU accelerator cards to handle computer vision or deep learning workloads. Its data storage is a full dataset containing rich details and relationships to support arbitrary post-event traceability and multi-dimensional drill-down analysis. At the application level, the standardized model is primarily geared towards management decision-makers, data analysts, and operations planners. It is used to complete strategic tasks such as tourist consumption behavior analysis, long-term tourism resource planning, marketing campaign effectiveness evaluation, and potential risk simulation for large-scale events. These tasks do not require millisecond-level response times but demand comprehensive analysis, accuracy, and depth of insight to provide data support for long-term business growth and refined management.

[0037] The core difference between lightweight and standardized platforms lies in their design objectives, rather than simply adding or removing features. First, regarding core objectives, lightweight platforms prioritize "efficiency" and "economy," with their primary task being to ensure the continuity and timeliness of critical business operations in resource-constrained environments. Standardization, on the other hand, prioritizes "depth" and "comprehensiveness," its value lying in discovering hidden patterns and value within data to support long-term strategies. Second, in terms of technical implementation, this difference manifests in a comprehensive trade-off: the lightweight model trades speed and flexibility for feature reduction, data simplification, and resource constraints—a state of "targeted optimization"; the standardized model, however, unleashes the full potential of data by investing sufficient computing and storage resources—a state of "unconstrained exploration." Finally, in terms of application scenarios and user roles, the two form a perfect complement: lightweight platforms serve "present" on-site operations and temporary, unexpected needs, with users at the execution level; standardization serves "future" planning and "past" review, with users at the decision-making level.

[0038] The transition from a standardized to a lightweight big data platform is not a simple manual switch, but a near-real-time, multi-layered, systematic project planned by an intelligent control center. For example... Figure 6As shown, the entire process begins with a trigger signal. This signal can come from an explicit command from the administrator on the visual console, or it can be automatically generated by the embedded context-aware engine based on predefined policies. Once triggered, the control center first queries the metadata repository to determine the list of functional modules that need to be enabled or disabled for the currently requested mode, as well as their dependencies. Next, the resource scheduler intervenes, quickly creating or destroying the corresponding container instances on the Kubernetes cluster, or elastically scaling existing services, based on the resource quota for the target mode.

[0039] During the transition between standardized and lightweight big data platforms, the service mesh updates traffic routing rules in real time based on the new configuration. For example, all API requests from mobile devices are routed to lightweight API service instances, while requests from the data analytics backend are redirected to the standardized microservice cluster. The most critical step occurs at the data layer, where the dynamic data virtualization layer provides different data views to applications based on the switching target. When switching to lightweight mode, it rewrites complex SQL queries into simple queries for highly summarized tables and pre-computed materialized views, significantly improving response speed; while switching back to standardized mode, it restores full access to the original fine-grained data. All these operations are completed collaboratively within seconds, ultimately smoothly transitioning the entire big data platform's operational status, performance, and resource consumption from one form to another without the user's noticeable difference, thus achieving a dynamic switch of "one set of code, two forms."

[0040] The specific scenario in this embodiment is as follows: 1. Full-cycle management of large-scale cultural and sports events: During events such as city marathons, large concerts, or sporting events, the platform's operating mode dynamically adjusts according to the event's phases. In the event preparation and post-event debriefing stages, the system operates in standardized mode, utilizing historical data and predictive models for crowd simulation, facility planning, ticketing pricing strategy analysis, and comprehensive event effectiveness evaluation. On the day of the event, especially during peak entry and exit times, the system switches to lightweight mode, shutting down all time-consuming analysis tasks and concentrating all computing resources on core tasks such as real-time ticket verification, real-time generation of crowd heat maps, structured analysis of security monitoring video streams, and emergency channel scheduling. This ensures the on-site command center receives critical information with millisecond-level latency, guaranteeing the safe and smooth operation of the event.

[0041] 2. Seamless integration of routine maintenance and emergency response: In the daily operation of tourist attractions or museums, the platform can run in lightweight mode for extended periods, performing basic tasks such as visitor counting, ticket sales statistics, and equipment status monitoring with low power consumption, significantly reducing operating costs. Once the system detects anomalies through real-time data streams (such as a sudden exceedance of visitor density in a certain area, equipment malfunction, or a public health emergency), the context-aware engine immediately and automatically triggers a partial switch to standardized mode. At this time, advanced functions such as AI prediction models, visitor trajectory tracking, and evacuation simulation are instantly activated, providing managers with in-depth decision support and minimizing the impact of emergencies. After the incident is resolved, the system can automatically revert to lightweight mode.

[0042] 3. On-demand services for users with multiple roles: The same platform needs to serve users with different roles simultaneously, whose needs vary greatly. For frontline staff (such as security guards and ticket inspectors), requests from their mobile devices are automatically identified by the API gateway and routed to a lightweight backend, providing them with a simple, fast, and core user interface. For backend management and analysis personnel, their complex query and analysis requests are handled by a standardized service cluster, ensuring the depth and flexibility of the analysis. This intelligent switching based on user role allows employees in different functions to obtain the best user experience without having to deploy two separate systems for them, achieving an optimal balance between resources and user experience.

Claims

1. A dual-mode big data platform for the cultural, tourism, and sports industries, characterized in that: It includes a standardized platform, a lightweight platform, and a big data platform control system. The big data platform includes a mode control hub for switching control modes, and adopts a centralized control and distributed execution architecture with a metadata center as the core hub to ensure synchronization between the two platforms during operation and maintenance upgrades. The big data platform control system includes an intelligent operation and maintenance coordinator that schedules the standardized platform and the lightweight platform to perform operation and maintenance upgrades separately. It also includes a data synchronization engine that uses bidirectional incremental replication technology to control data synchronization between the standardized platform and the lightweight platform. It also includes a functional synchronization layer that ensures interface compatibility between the standardized platform and the lightweight platform through contract testing.

2. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 1, characterized in that, The data synchronization engine adopts a multi-level pipeline architecture to solve the data consistency problem between two platforms; the data synchronization engine includes a change capture module, a synchronization scheduling module, a conflict resolution module, and a data verification module; The change capture module captures change data from the standardized platform and the lightweight platform, merges them into a change set, converts the captured change set into a unified JSON Patch format, and then enters the distributed message queue Kafka. The synchronization scheduling module performs differentiated routing based on the data type of the cultural, tourism, and sports platform, and the conflict resolution module verifies the integrity and consistency of the data.

3. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 2, characterized in that, The change capture module uses the Debezium connector based on database logs to capture data changes in real time for standardized platforms; for lightweight platforms, it uses SQLite's write-ahead log listening mechanism to capture data changes.

4. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 2, characterized in that, The conflict resolution module uses a three-tier arbitration mechanism to verify the integrity and consistency of data. The basic layer processes data automatically according to preset rules, the business layer makes decisions based on timestamps and operation types, and the cultural, tourism and sports specialization layer makes decisions based on seasonal factors.

5. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 2, characterized in that, After each synchronization, the data synchronization engine performs end-to-end verification by the data verification module, ensuring consistency by comparing the hash values ​​of key business entities. When a difference is detected, an incremental repair process is automatically triggered, prioritizing the repair of real-time business data, while historical data enters the background repair queue. During periods when cultural, tourism, and sports venues are not open for business, the data verification module automatically performs full data verification to ensure that the data is completely consistent before operation.

6. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 1, characterized in that, The intelligent operation and maintenance coordinator drives the operation and maintenance upgrade process through a state machine. When the operation and maintenance upgrade is started, the intelligent operation and maintenance coordinator first freezes the write operations of the target platform and marks the upgrade start point through a global transaction ID. For upgrades to the standardized platform, user requests are automatically routed to the lightweight platform, while a shadow database is started on the standardized platform to record changes. After the upgrade is completed, the data comparison tool calculates the differences between the primary database and the shadow database, generates incremental patches, and synchronizes them to the lightweight platform. For upgrades to the lightweight platform, a mirroring strategy is adopted, first creating a read-only copy of the standardized platform to provide services, and then performing configuration synchronization after the upgrade is completed.

7. The dual-mode big data platform for the cultural, tourism, and sports industry according to claim 1 or 6, characterized in that, The intelligent operation and maintenance coordinator implements four-stage safeguards during the operation and maintenance upgrade process: the preparation stage performs metadata compatibility verification and resource pre-allocation; the execution stage adopts canary release, updating in batches according to the type of cultural, tourism and sports venues and user groups; and the verification stage automatically runs contract tests and business scenario tests. Upon completion of the phase, a data consistency report is generated and the data is unfrozen and written; during short-term high-traffic periods at cultural, tourism, and sports venues, the intelligent operation and maintenance coordinator automatically delays non-critical upgrades; during long-term active tourist periods at cultural, tourism, and sports venues, a lightweight platform priority mode is activated to ensure that the tourist experience is not affected.

8. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 1, characterized in that, The functional synchronization layer achieves interface compatibility through a contract testing and monitoring system; the metadata center stores the standardized definition metadata of all APIs, and before each upgrade, the contract testing service automatically generates test cases for typical cultural, tourism and sports scenarios, which are then executed and verified in parallel on both platforms. Interface changes must pass backward compatibility checks; newly added parameters must have default values ​​defined; and deleted parameters must be marked with a deprecation period. For major changes, a version coexistence strategy is adopted, and traffic is separated between the old and new versions through the API gateway.

9. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 1, characterized in that, The big data platform mode switching is controlled by the mode control center. When the switching command is triggered, the metadata repository of the mode control center is activated first. This repository stores the dependency relationship graph of all modules. After parsing this graph, the metadata interpreter generates service composition instructions and dynamically adjusts Istio traffic routing through the VirtualService rules of the Istio service mesh, directing requests for specific modules to the target cluster. At the same time, the resource scheduler creates elastic resource barriers, allocating namespace-level resource quotas for lightweight mode and removing restrictions in normalized mode, allowing the use of GPU acceleration pools. Data consistency during mode switching is ensured by the state synchronization engine, which uses the CRDT algorithm to resolve distributed state conflicts. When switching back to normalized mode from lightweight mode, the engine initiates an incremental checkpoint recovery mechanism: replaying key transactions during the lightweight operation based on the operation log and reconstructing the data view through a dynamic schema adapter; The adapter builds a virtualization layer on top of the data lake. In lightweight mode, it automatically creates columnar storage projections, while in normalized mode, it loads the full JSONB documents. The resource allocation module dynamically adjusts resource weights based on the Prometheus metrics of the resource data when switching modes.

10. The dual-mode big data platform for the cultural, tourism, and sports industries according to claim 1, characterized in that, The standardized platform integrates a complete ETL process, data warehouse, machine learning algorithm library, and BI analysis tools. It saves the full dataset when storing data and supports post-event traceability and multi-dimensional drill-down analysis. The lightweight platform only carries the data processing and display modules. Service instances are set up on edge nodes or low-cost computing units close to the data source. It reduces network I / O and disk usage through highly compressed data transmission protocols and columnar storage projection. Resource consumption is isolated and limited based on resource quotas.

Images