Enterprise data management system based on digital standard system

Abstract

The invention discloses an enterprise data management system based on a digital standard system. The enterprise data management system comprises the following modules: a data acquisition module, a digital standard granularity construction module, an automatic granularity matching and classification module, a dynamic quality control engine module and a data storage and feedback module. The data acquisition module is used for acquiring enterprise multi-source heterogeneous original data; and the digital standard granularity construction module is used for constructing a digital standard granularity hierarchical system comprising at least three core hierarchies. The invention relates to the technical field of enterprise data management. According to the enterprise data management system based on the digital standard system, a multi-level digital standard granularity system is constructed, internationalized data unified management is realized, and through standard construction of a plurality of core levels such as a strategic level, a service level and an operation level, data collaboration and governance of an enterprise full-service scene are supported; and the sharing performance and the expandability of the data are enhanced.

Description

Technical Field

[0001] This invention relates to the field of enterprise data management, and more specifically, to an enterprise data management system based on a digital standard system. Background Technology

[0002] With the development of information technology and digital transformation, enterprise data sources are becoming increasingly diversified, involving different business systems, device terminals, and organizational structures. Multi-source heterogeneous data differs significantly in format, semantics, and granularity, leading to the following problems: The lack of a unified data standard system means that traditional enterprise data standards often remain at the field level, without systematically building a standard granularity system covering multiple levels such as strategy, business and operations, making it difficult to support cross-departmental and cross-system data collaboration and management.

[0003] Manual data classification and mapping is costly and error-prone. In the current data governance process, data granularity identification and classification mapping rely heavily on human experience and judgment, which is inefficient and difficult to adapt to rapid business changes.

[0004] Fixed data quality verification rules lack adaptability. Different business scenarios have significantly different data quality requirements. Traditional verification mechanisms based on fixed rules are unable to respond to business changes and error trends in real time, affecting data credibility and usability.

[0005] Data quality feedback is lagging and there is a lack of closed-loop management capabilities. Most existing systems only perform verification at the data entry stage, lacking subsequent quality monitoring and historical error analysis, and thus failing to provide a basis for standard system iteration and quality improvement strategies.

[0006] In summary, existing technologies lack systematic support for standardization, multi-level granularity, automatic classification, and dynamic quality control in the data governance process, and there is an urgent need for a new system that can achieve intelligent governance of data quality throughout the entire process. Summary of the Invention

[0007] The purpose of this invention is to provide an enterprise data management system based on a digital standard system, which solves the problems of existing enterprise data management systems, such as the lack of a unified data standard system, high cost and error-prone manual data classification and mapping, fixed data quality verification rules, lack of adaptability, delayed data quality feedback, and lack of closed-loop management capabilities.

[0008] This invention achieves the above objectives through the following technical solution: an enterprise data management system based on a digital standard system, comprising the following modules: The module includes a data acquisition module, a digital standard granularity construction module, a granularity automatic matching and classification module, a dynamic quality control engine module, and a data storage and feedback module. The data acquisition module is used to collect multi-source heterogeneous raw data from enterprises; The digital standard granularity construction module is used to build a digital standard granularity hierarchical system containing at least three core levels. The automatic granularity matching and classification module has a built-in machine learning model, which is used to automatically match and classify the standard granularity level corresponding to each original data, and output the data-granularity matching result and the matching confidence level. The dynamic quality control engine module dynamically adjusts the strength and method of data verification rules in real time based on data-granularity matching results, matching confidence, historical data error types, and the current business context. The data entry and feedback module is used to enter data into the database and provide feedback on data that fails verification based on the verification results of the dynamic quality control engine module, and to record verification-related information to update the historical error database.

[0009] Furthermore, the core levels of the digital standard granularity hierarchy system include strategic level, business level, and operational level; The criteria for dividing each granularity level should at least cover data aggregation degree, time dimension, business impact scope and data user subject, and the standard attributes of each granularity level are stored in the standard granularity database, which supports dynamic updates of granular attributes according to the needs of enterprise business iteration.

[0010] Furthermore, the machine learning model in the granularity automatic matching and classification module adopts an architecture that combines gradient boosting trees with an attention mechanism; The model training process includes at least the steps of data preprocessing, sample labeling, model training, and model evaluation. The data preprocessing is used to clean historical enterprise data and extract multidimensional features to construct a feature matrix; The sample annotation is used to generate a training sample set that includes a training set, a validation set, and a test set; The model training is used to build a basic classification model and optimize the model hyperparameters; The model evaluation is used to determine whether the model meets the deployment standards based on preset evaluation indicators.

[0011] Furthermore, the dynamic quality control engine module implements the dynamic adjustment logic of the verification rules through the rule engine; The rule engine includes at least a rule base, an inference engine, and a dynamic adjustment module; The rule base stores at least three levels of validation rule templates; The inference engine is used to receive the output results of the granularity automatic matching and classification module, and, in combination with historical error database information and the current business context, determine the verification rule level that should be enabled. The dynamic adjustment module is used to calculate the rule adjustment coefficient based on the real-time verification results, thereby realizing the adjustment of the verification rule level.

[0012] Furthermore, the historical error database stores at least the following data: error data identification information, error occurrence time, error type, granularity level of the corresponding data, matching confidence level, enabled verification rules, and error handling results; The system regularly performs statistical analysis on historical error data and generates data quality analysis reports, providing a basis for the iteration of the digital standard granularity hierarchy and the updating of machine learning models.

[0013] Furthermore, it also includes: Data quality monitoring module; The data quality monitoring module is used to monitor the quality indicators of data after it is entered into the database in real time. The quality indicators include at least data accuracy, completeness, consistency and timeliness. When any quality indicator falls below a preset threshold for an extended period of time, the system automatically triggers an alarm mechanism and sends an alarm message to the specified object.

[0014] Furthermore, the data accuracy is calculated by comparing the consistency between the sampled data and the original data source. The data integrity is calculated by statistically analyzing the proportion of non-empty fields in the required fields of data at each granularity level. The data consistency is calculated by verifying the logical consistency between data across granularity levels. The timeliness of the data is calculated by taking time from data collection to data entry.

[0015] Furthermore, the system supports access from multiple terminals, including at least a PC management platform, a mobile application, and an API interface. The PC-based management platform provides functions for configuring a digital standard granularity system, managing machine learning models, viewing data quality monitoring reports, and performing manual review. The mobile application allows data collection personnel to receive feedback information, view personal data quality statistics, and receive system alarm information. The API interface supports integration with existing enterprise business systems, enabling automatic data collection and two-way synchronization, and employs an encrypted transmission protocol to ensure data security.

[0016] Furthermore, the preset evaluation indicators for the model evaluation include at least accuracy, F1 score, and variance of matching confidence. When the test set accuracy reaches the preset accuracy threshold and the F1 score reaches the preset F1 score threshold, the model is determined to meet the deployment standard; otherwise, iterative training is performed again.

[0017] Furthermore, the rule adjustment coefficient calculated by the dynamic adjustment module is the product of the ratio of the current error rate to the historical average error rate and the ratio of the current matching confidence to the baseline confidence. When the rule adjustment coefficient is greater than the first preset coefficient threshold, the verification rule level is raised by one. When the rule adjustment coefficient is less than the second preset coefficient threshold, the verification rule level is reduced by one level. When the rule adjustment coefficient is between the first preset coefficient threshold and the second preset coefficient threshold, the current verification rule level is maintained.

[0018] The beneficial effects of this invention are as follows: 1. Construct a multi-level digital standard granularity system to achieve unified international data management. Through the construction of standards at multiple core levels such as strategic, business, and operational levels, support data collaboration and governance across all business scenarios of the enterprise, and enhance the sharing and scalability of data.

[0019] 2. Machine learning is used to automatically identify data granularity, improving processing efficiency and accuracy. The granularity automatic matching and classification module integrates gradient boosting trees and attention mechanisms to improve the accuracy of data granularity identification, reduce manual intervention, and significantly reduce standard matching costs.

[0020] 3. The dynamic quality control engine enables intelligent rule adjustment, adaptively adjusting the verification intensity based on matching confidence and historical error behavior, achieving differentiated quality control, avoiding over-verification or missing verification, and improving data quality assurance capabilities.

[0021] 4. Establish a continuous closed-loop mechanism for data quality improvement. Through historical error databases and quality monitoring modules, continuously accumulate quality risk data to provide a basis for model training, rule upgrades, and standard iterations, thereby strengthening the evolution of data governance capabilities.

[0022] 5. Supports multi-terminal empowerment for rapid business response, with access via PC, mobile, and API to meet the needs of multiple roles such as data collection, monitoring, and manual processing, thereby improving the timeliness of data processing and business adaptability. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is the main flowchart of data governance in this invention; Figure 2 This is a diagram of the dynamic quality control mechanism of the present invention; Figure 3 This is a diagram of the data quality monitoring system of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1: Please see Figure 1-3 This invention provides a technical solution: an enterprise data management system based on a digital standard system, the system comprising: The data acquisition module is used to collect multi-source heterogeneous raw data from enterprises, including financial data, supply chain data, human resources data, and business operation data. Among them, multi-source heterogeneous raw data, multi-source means that the data comes from a wide range of sources, involving multiple different departments and business links within the enterprise, and heterogeneous means that the data structure and format are different. Data from different sources may use different database systems, file formats or data representation methods; raw data refers to unprocessed initial data, including financial data, supply chain data, human resources data and business operation data. The digital standard granularity construction module is used to build a "digital standard granularity hierarchical system". The system includes three core levels: strategic level, business level and operational level. The strategic level granularity corresponds to the aggregated data required for enterprise strategic decision-making, the business level granularity corresponds to the process data required for cross-departmental business collaboration, and the operational level granularity corresponds to the detailed data required for grassroots business execution. The digital standard granularity hierarchy system is a layered framework for standardizing enterprise data granularity, comprising three core levels: strategic, business, and operational. It aims to meet the data detail requirements of different management and business levels. Strategic granularity corresponds to the aggregated data needed for enterprise strategic decision-making. This highly aggregated and summarized data provides senior managers with a macro-level, holistic perspective, assisting in the formulation of long-term enterprise development strategies. Business granularity corresponds to the process-oriented data needed for cross-departmental business collaboration, focusing on the data flow and correlation at each stage of the business process, helping different departments achieve information sharing and collaborative work. Operational granularity corresponds to the detailed data needed for grassroots business execution; it is the most detailed and specific data used to guide the daily business operations of grassroots employees. The granularity automatic matching and classification module has a built-in machine learning model. The model takes the field attributes, data dimensions and business affiliation of the original data as input features, automatically matches and classifies the standard granularity level corresponding to each original data, and outputs the data-granularity matching results and matching confidence. Among them, a machine learning model is a computer program based on algorithms and statistics that can learn patterns and rules from large amounts of data and predict and classify new data based on the learned knowledge; input features are the data attributes used to train the machine learning model. In this module, input features include the field attributes, data dimensions, and business affiliation of the original data; data-granularity matching results are the results obtained by the machine learning model after determining the standard granularity level to which the original data belongs based on the input features; matching confidence represents the reliability or certainty of the model's data-granularity matching results, usually presented as a percentage. The dynamic quality control engine module dynamically adjusts the strength and method of data validation rules in real time based on data-granularity matching results, matching confidence levels, historical data error types, and the current business context. When the matching confidence is ≥90% and the historical error rate is <1%, the basic validation rules are adopted, and only field non-empty and data type consistency validations are performed. When 60% ≤ matching confidence < 90% or 1% ≤ historical error rate < 5%, the intermediate verification rule is enabled, adding logical correlation verification on the basis of basic verification. When the matching confidence is less than 60% or the historical error rate is greater than or equal to 5%, the advanced validation rules are triggered. In addition to the intermediate validation content, cross-granularity level cross-validation and manual review triggering mechanisms are performed. Data validation rules are a set of rules used to check the accuracy and completeness of data. Based on different conditions and application scenarios, they are divided into basic validation rules, intermediate validation rules, and advanced validation rules. Basic validation rules only perform field non-null and data type consistency verification. Intermediate validation rules add logical correlation verification to basic validation, checking whether the logical relationships between data are reasonable, such as the logical relationship between order amount and order quantity. Advanced validation rules, in addition to the intermediate validation content, additionally perform cross-granularity level cross-validation and manual review triggering mechanisms. Historical data error types are various data errors discovered during past data validation processes, such as missing data, data errors, and logical contradictions. The current business context refers to the environment and conditions of the current business activity, including business stage, business process, and business objectives. These factors affect the selection and application of data validation rules. The data entry and feedback module is used to enter data that passes the verification based on the verification results of the dynamic quality control engine module, and to feed back data that fails the verification and the reasons for the error to the data provider. At the same time, it records the rule parameters, error types and processing results during this verification process and updates them to the historical error database to provide data support for subsequent verification rule adjustments. The process includes: data entry, data storage, and historical error database. Data entry involves storing data validated by the dynamic quality control engine module into the enterprise's database for subsequent data analysis and use. Data failure feedback, including the reasons for the failure, is provided to the data provider for correction. Rule parameters are various parameters within the dynamic quality control engine module used to adjust the strength and method of data validation rules, such as thresholds and weights. The historical error database stores the rule parameters, error types, and processing results from the current validation process, providing data support and reference for adjusting subsequent validation rules.

[0026] It should be noted that during operation, a digital standard granularity hierarchy system is constructed, allowing different levels of data to correspond to specific needs, making data collection, storage, and use more organized and improving data management efficiency. The granularity automatic matching and classification module uses machine learning models to automatically classify data based on its characteristics, outputting matching results and confidence levels, and can quickly and accurately assign data to the appropriate granularity level. The dynamic quality control engine module adjusts the verification rules in real time based on various factors, adopting basic, intermediate, or advanced verification for different situations, improving verification efficiency while ensuring data quality. The data entry and feedback module not only handles data entry and feedback issues but also records verification information and updates it to the historical error database, providing a basis for subsequent rule adjustments and forming a closed loop of continuous optimization.

[0027] In one embodiment, the granularity construction module for digital standards is divided into different granularity levels based on factors including data aggregation degree, time dimension, scope of business impact, and data user entity. The specific granularity criteria are as follows: Strategic-level granularity: Data aggregation degree ≥90%, specifically, a single data point covers ≥90% of the summary information of the grassroots business units, the time dimension is quarterly or annual, the business impact covers the entire enterprise, and the main users of the data are the senior management of the enterprise; Business-level granularity: Data aggregation degree 30%-90%, time dimension in monthly or weekly units, business impact covers 2 or more departments, and the main users of the data are department heads and business collaborators; Operational granularity: Data aggregation degree <30%, time dimension in daily or hourly units, business impact limited to a single business process, and data users are grassroots business execution personnel; Furthermore, the standard attributes of each granularity level are stored in a standard granularity database, which supports dynamic updates of granular attributes according to the needs of enterprise business iteration.

[0028] This design categorizes data granularity levels based on aggregation degree, time dimension, business impact scope, and data user groups. Strategic-level data has high aggregation degree and impacts the entire enterprise, intended for senior management. Business-level data has moderate aggregation degree, covering multiple departments and used for business collaboration. Operational-level data has low aggregation degree, targeting individual business processes, and is used by frontline staff. Standard attributes are stored in the database and can be dynamically updated, precisely adapting to the data needs of different levels of personnel. This tightly integrates data management with the enterprise's business architecture. Senior management can quickly access macro-level aggregated data for strategic decision-making, department staff can collaborate based on cross-departmental process data, and frontline staff can access detailed data to efficiently execute tasks. The dynamic update mechanism allows the system to flexibly adjust with the enterprise's business development, always maintaining a match between data granularity and the enterprise's actual needs, thus improving the scientific nature and effectiveness of data management.

[0029] In one embodiment, the machine learning model in the granular automatic matching and classification module adopts an architecture combining gradient boosting trees and attention mechanisms. The model training process includes: Data preprocessing: Clean historical enterprise data, extract 128-dimensional features such as field type, data length, business tags, and aggregation frequency, and construct a feature matrix; Sample labeling: Business experts label historical data samples at the granular level according to the "digital standard granularity hierarchy system" to generate a training sample set, including a training set, a validation set, and a test set, in a ratio of 7:2:1; Model training: Using the granularity of labeled samples as the target variable, a basic classification model is built through gradient boosting trees. An attention mechanism is introduced to strengthen the weights of key features such as business labels and data aggregation degree. Cross-validation is used to optimize model hyperparameters, including learning rate, tree depth, and number of leaf nodes. Model evaluation: Accuracy, F1 score, and matching confidence variance are used as evaluation metrics. When the test set accuracy is ≥92% and the F1 score is ≥90%, the model meets the deployment standard; otherwise, iterative training is performed.

[0030] This design employs a gradient boosting tree combined with an attention mechanism in the machine learning model. It involves preprocessing historical data to construct a feature matrix, generating a training set by labeling samples with business experts, training the model using these labeled samples as target variables, and introducing an attention mechanism to strengthen the weights of key features. The model is evaluated using metrics such as accuracy. The gradient boosting tree builds a basic classification model, while the attention mechanism highlights key features, improving the model's accuracy in classifying data at a granular level. Data preprocessing and sample labeling ensure the reliability and effectiveness of the training data. Strict model evaluation standards ensure that the deployed model has high accuracy and stability, quickly and accurately matching raw data to the appropriate granularity level, providing a precise foundation for subsequent data processing and analysis, and improving data management efficiency.

[0031] In one embodiment, the dynamic adjustment logic of the verification rules in the dynamic quality control engine module is implemented through a rule engine, which includes a rule base, an inference engine, and a dynamic adjustment module. Rule base: Stores validation rule templates at three levels: basic, intermediate, and advanced. Each template contains a rule ID, validation fields, validation logic, triggering conditions, and processing methods. Inference engine: Receives the matching results output by the granular automatic matching and classification module, combines the error types in the historical error database with the current business context, and infers to determine the current verification rule level that should be enabled. Error types include missing fields, logical conflicts, and data redundancy. Dynamic adjustment module: Calculates the rule adjustment coefficient K based on real-time verification results. K = (Current error rate / Historical average error rate) × (Current match confidence / Baseline confidence); When K > 1.2, the validation rule level is increased by one. When K < 0.8, reduce the validation rule level by one. When 0.8≤K≤1.2, the current verification rule level is maintained, realizing flexible adjustment of the verification rules.

[0032] This design allows the rule engine to dynamically adjust validation rules. The rule base stores validation rule templates at different levels. The inference engine infers the appropriate rule level based on matching results, error types, and business context. The dynamic adjustment module calculates adjustment coefficients based on real-time validation results to achieve elastic adjustment. The rule base provides standardized validation rule templates for easy management and maintenance. The inference engine can comprehensively consider multiple factors and accurately infer the appropriate validation rule level, improving the targeting of validation. The dynamic adjustment module calculates adjustment coefficients and flexibly adjusts the validation rule level according to real-time validation conditions, achieving elastic adjustment of validation rules. This adapts to different data quality requirements, ensuring data quality while improving validation efficiency and reducing unnecessary validation costs.

[0033] In one embodiment, the historical error database stores data including error data ID, error occurrence time, error type, corresponding data granularity level, matching confidence, the verification rules used at that time, and error handling results. The system performs statistical analysis on the historical error data monthly and generates a data quality analysis report. The report includes the error rate change trend at each granularity level, high-frequency error types, and verification rule optimization suggestions, providing a basis for the iteration of the digital standard granularity level system and the updating of machine learning models.

[0034] This design allows the historical error database to store multifaceted information about error data, and monthly statistical analysis reports are generated. The detailed storage of error data provides a rich data source for subsequent analysis, helping to gain a deeper understanding of the overall picture of data errors. The monthly statistical analysis reports clearly present information such as the changing trends of error rates at each granularity level and high-frequency error types. This information provides a strong basis for the iteration of the digital standard granularity hierarchy system, allowing for targeted optimization of granularity division and standard attributes. It also provides data support for updating machine learning models, enabling them to learn more accurate data features and error patterns, improving the model's classification accuracy and data matching reliability, and ultimately enhancing the data quality of the entire enterprise data management system.

[0035] In one embodiment, the system further includes a data quality monitoring module for real-time monitoring of data quality indicators after data is entered into the database. These quality indicators include data accuracy, completeness, consistency, and timeliness. Data accuracy: The accuracy rate is calculated by comparing the sampled data with the original data source to determine the consistency. Accuracy = (Number of correct data entries in the sample / Total number of data entries in the sample) × 100%; Data integrity: Calculate the percentage of non-empty fields in required fields at each granularity level, and then calculate the integrity percentage as follows: Integrity = (Total number of non-empty required fields / Total number of required fields) × 100%. Data consistency: Verify the logical consistency between data across granular levels (such as the deviation rate between strategic-level data and business-level aggregated data), and calculate consistency = (number of unbiased data groups / total number of data groups) × 100%; Data timeliness: The time taken from data collection to data entry into the database. Timeliness is calculated as follows: (Number of data entries entered into the database within the specified time / Total number of data entries entered into the database) × 100%. When any quality indicator falls below the preset threshold (accuracy ≥ 98%, completeness ≥ 95%, consistency ≥ 97%, timeliness ≥ 96%) for three consecutive days, the system will automatically trigger an alarm mechanism and send an alarm message to the system administrator.

[0036] This design allows for real-time monitoring of the accuracy, completeness, consistency, and timeliness of data after it has been imported into the database. An alarm is triggered if any indicator falls below a threshold for three consecutive days. By monitoring multiple key quality indicators in real time, various problems that arise during the data import process can be identified promptly. Accuracy monitoring ensures data consistency with the original data source; completeness monitoring guarantees no missing fields; consistency monitoring maintains the logical relationships between data across different granular levels; and timeliness monitoring ensures timely data import. The automatic alarm mechanism when an indicator falls below a threshold quickly notifies the system administrator, enabling them to take timely measures to resolve the issue and prevent problematic data from impacting corporate decisions and business operations. This helps ensure the quality and availability of enterprise data and improves the timeliness and effectiveness of data management.

[0037] In one embodiment, the system supports multi-terminal access, including a PC-based management platform, a mobile app, and API interfaces, wherein: PC-based management platform: Provides functions such as configuring digital standard granularity system, training and updating machine learning models, viewing data quality monitoring reports, and manual review. Mobile App: Supports data collection personnel in receiving feedback on failed verification, viewing personal data quality statistics, and receiving system alarm information; API Interface: Supports integration with existing enterprise ERP, CRM, and OA systems to achieve automatic data collection and two-way synchronization. The interface uses HTTPS protocol for encrypted transmission to ensure data security.

[0038] This design allows the system to support multi-terminal access via a PC management platform, a mobile app, and API interfaces. The PC platform provides configuration, training, report viewing, and auditing functions; the mobile app allows data collection personnel to receive feedback, view statistics, and receive alarms; the API interface enables integration with enterprise systems, with encrypted transmission ensuring security, meeting the needs of different users in different scenarios. The PC management platform is comprehensive, facilitating system configuration and monitoring for administrators; the mobile app allows data collection personnel to access information promptly, improving work efficiency; and the API interface enables seamless integration with existing enterprise systems, promoting data flow and sharing, while encrypted transmission ensures data security during transmission, preventing data leakage. Multi-terminal access support enhances the system's flexibility and ease of use, helping enterprises manage data more efficiently.

[0039] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0040] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An enterprise data management system based on a digital standard system, characterized in that, Includes the following modules: The module includes a data acquisition module, a digital standard granularity construction module, a granularity automatic matching and classification module, a dynamic quality control engine module, and a data storage and feedback module. The data acquisition module is used to collect multi-source heterogeneous raw data from enterprises; The digital standard granularity construction module is used to build a digital standard granularity hierarchical system containing at least three core levels. The automatic granularity matching and classification module has a built-in machine learning model, which is used to automatically match and classify the standard granularity level corresponding to each original data, and output the data-granularity matching result and the matching confidence level. The dynamic quality control engine module dynamically adjusts the strength and method of data verification rules in real time based on data-granularity matching results, matching confidence, historical data error types, and the current business context. The data entry and feedback module is used to enter data into the database and provide feedback on data that fails verification based on the verification results of the dynamic quality control engine module, and to record verification-related information to update the historical error database.

2. The enterprise data management system according to claim 1, characterized in that: The core levels of the digital standard granularity hierarchy system include strategic level, business level and operational level; The criteria for dividing each granularity level should at least cover data aggregation degree, time dimension, business impact scope and data user subject, and the standard attributes of each granularity level are stored in the standard granularity database, which supports dynamic updates of granular attributes according to the needs of enterprise business iteration.

3. The enterprise data management system according to claim 1, characterized in that: The machine learning model in the granularity automatic matching and classification module adopts an architecture that combines gradient boosting trees and attention mechanisms; The model training process includes at least the steps of data preprocessing, sample labeling, model training, and model evaluation. The data preprocessing is used to clean historical enterprise data and extract multidimensional features to construct a feature matrix; The sample annotation is used to generate a training sample set that includes a training set, a validation set, and a test set; The model training is used to build a basic classification model and optimize the model hyperparameters; The model evaluation is used to determine whether the model meets the deployment standards based on preset evaluation indicators.

4. The enterprise data management system according to claim 1, characterized in that: The dynamic quality control engine module implements the dynamic adjustment logic of the verification rules through the rule engine; The rule engine includes at least a rule base, an inference engine, and a dynamic adjustment module; The rule base stores at least three levels of validation rule templates; The inference engine is used to receive the output results of the granularity automatic matching and classification module, and, in combination with historical error database information and the current business context, determine the verification rule level that should be enabled. The dynamic adjustment module is used to calculate the rule adjustment coefficient based on the real-time verification results, thereby realizing the adjustment of the verification rule level.

5. The enterprise data management system according to claim 1, characterized in that: The historical error database stores at least the following data: error data identification information, error occurrence time, error type, granularity level of the corresponding data, matching confidence level, enabled verification rules, and error handling results. The system regularly performs statistical analysis on historical error data and generates data quality analysis reports, providing a basis for the iteration of the digital standard granularity hierarchy and the updating of machine learning models.

6. The enterprise data management system according to claim 1, characterized in that, Also includes: Data quality monitoring module; The data quality monitoring module is used to monitor the quality indicators of data after it is entered into the database in real time. The quality indicators include at least data accuracy, completeness, consistency and timeliness. When any quality indicator falls below a preset threshold for an extended period of time, the system automatically triggers an alarm mechanism and sends an alarm message to the specified object.

7. The enterprise data management system according to claim 6, characterized in that: The data accuracy rate is calculated by comparing the consistency between the sampled data and the original data source. The data integrity is calculated by statistically analyzing the proportion of non-empty fields in the required fields of data at each granularity level. The data consistency is calculated by verifying the logical consistency between data across granularity levels. The timeliness of the data is calculated by taking time from data collection to data entry.

8. The enterprise data management system according to claim 1, characterized in that: The system supports access from multiple terminals, including at least a PC management platform, a mobile application, and an API interface. The PC-based management platform provides functions for configuring a digital standard granularity system, managing machine learning models, viewing data quality monitoring reports, and performing manual review. The mobile application allows data collection personnel to receive feedback information, view personal data quality statistics, and receive system alarm information. The API interface supports integration with existing enterprise business systems, enabling automatic data collection and two-way synchronization, and employs an encrypted transmission protocol to ensure data security.

9. The enterprise data management system according to claim 3, characterized in that: The preset evaluation indicators for the model evaluation include at least accuracy, F1 score, and variance of matching confidence. When the test set accuracy reaches the preset accuracy threshold and the F1 score reaches the preset F1 score threshold, the model is determined to meet the deployment standard; otherwise, iterative training is performed again.

10. The enterprise data management system according to claim 4, characterized in that: The rule adjustment coefficient calculated by the dynamic adjustment module is the product of the ratio of the current error rate to the historical average error rate and the ratio of the current matching confidence to the baseline confidence. When the rule adjustment coefficient is greater than the first preset coefficient threshold, the verification rule level is raised by one. When the rule adjustment coefficient is less than the second preset coefficient threshold, the verification rule level is reduced by one level. When the rule adjustment coefficient is between the first preset coefficient threshold and the second preset coefficient threshold, the current verification rule level is maintained.

Images