A public data product compliance automatic detection system

Abstract

This invention relates to the field of data detection technology, and in particular to an automatic compliance detection system for public data products, comprising: a request parsing module for receiving data processing request information; a dynamic asset mapping module for real-time collection of data asset information associated with the data subject through data security monitoring probes; a feature extraction module for obtaining the historical authorization range sequence and asset sensitivity level tags of the data subject from the dynamic asset panorama; an analysis module including a built-in legal information database; a decision-making module for generating decision information; an execution module for executing corresponding access control based on the decision information; and a traceability recording module for collecting decision information throughout the entire process. This device, through the dynamic asset mapping module and various data security monitoring probes, constructs and continuously updates a dynamic asset panorama, which helps to solve the problem that traditional static ledgers cannot cover dynamically changing assets.

Description

Technical Field

[0001] This invention relates to the field of data inspection, and more particularly to an automatic inspection system for the compliance of public data products. Background Technology

[0002] The Public Data Product Compliance Automatic Detection System is an intelligent platform built on technologies such as rule engines and machine learning. It aims to conduct automated and standardized compliance reviews of the entire lifecycle of public data products. By incorporating national laws and regulations, industry standards, and local policy requirements, it can perform real-time scanning and intelligent judgment on the legality of data sources, the standardization of data processing, the sensitive information in data content, and the compliance of application scenarios. It is a core technical infrastructure to ensure the compliant circulation and development and utilization of public data.

[0003] Current public data compliance monitoring primarily relies on static ledgers and manual spot checks, which have systemic flaws: It cannot detect dynamically changing shadow assets in the network in real time, leading to unclear regulatory targets; it lacks comprehensive discretionary capabilities that integrate legal rules and historical behavior, only able to make rigid blacklist and whitelist judgments, resulting in a high rate of misjudgment; and it lacks tamper-proof evidence of the entire process of decision-making, making it difficult to provide a credible audit chain in case of disputes. This makes it difficult for traditional methods to achieve accurate, automated, and traceable compliance control in the face of dynamic and complex data circulation environments. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an automatic compliance detection system for public data products.

[0005] In a first aspect, the present invention provides an automatic detection system for the compliance of public data products, comprising:

[0006] The request parsing module is used to receive data processing request information and parse the data processing request information, which includes the data subject, processing purpose and timestamp;

[0007] The dynamic asset mapping module is used to collect data asset information associated with the data subject in real time through data security monitoring probes, so as to construct and continuously update the dynamic asset panorama, and automatically identify and mark shadow assets that are not included in the ledger.

[0008] The feature extraction module is used to obtain the historical authorization range sequence and asset sensitivity level label of the data subject from the dynamic asset panorama;

[0009] The analysis module includes a built-in legal information database, which is used to compare and match the processing purpose and the asset sensitivity level label with the legal information database to generate risk indicators;

[0010] The decision-making module is used to match and calculate the historical authorization range sequence and the risk indicators through a built-in weight model to generate decision information;

[0011] The execution module is used to execute the corresponding access control based on the decision information;

[0012] The traceability recording module is used to collect decision-making information throughout the entire process and to build an immutable decision traceability chain based on blockchain technology.

[0013] The request parsing module serves as the system's input interface, receiving data processing requests from external sources and performing structured parsing on these requests. Each data processing request includes at least three key fields: data body (to uniquely identify the requester), processing purpose (to clarify the requester's intent to use the data), and timestamp (to record the time the request was initiated). Through the request parsing module, the system can convert external requests into a unified, standardized data format for use in all subsequent processing stages.

[0014] The dynamic asset mapping module is connected to the request parsing module. It collects data information related to the data subject in real time through various data security probes deployed at key network nodes, application service interfaces, and database storage. Data security probes include, but are not limited to, full traffic probes, API interface probes, and database scanning tools. The dynamic asset mapping module uses the collected information to build and continuously update a dynamic asset panorama that includes structured data, unstructured data, and semi-structured data. The dynamic asset panorama can reflect the real-time status, distribution location, flow relationship, and attribute tags of data assets. Through automatic identification and marking functions, the dynamic asset mapping module can discover shadow assets that are not registered in the organization's formal asset ledger but actually exist in the network, and mark these shadow assets as assets not included in the management, thereby eliminating regulatory blind spots.

[0015] The feature extraction module is connected to the dynamic asset mapping module. It is used to extract the relevant historical authorization range sequence and the asset sensitivity level of the data subject from the dynamic asset panorama for the data subject of the current request. The historical authorization range sequence records the list of data assets that the data subject was authorized to access in the past specified time period and the authorization period. The asset sensitivity level is the sensitivity level of the data subject based on the data content recognition result, thereby transforming the raw stone information in the dynamic asset panorama into structured features that can be used for decision reference.

[0016] The analysis module stores relevant laws, regulations, departmental rules, industry standards, and other normative clauses related to the flow of public data products through a built-in, expandable legal information database. It takes the processing purpose and asset sensitivity level label as input information, compares and matches them with the clauses in the legal information database, and evaluates the legality of the request through preset matching logic or algorithms to generate a risk indicator that can reflect the degree of risk, such as a risk value expressed as a percentage or level.

[0017] The decision-making module is connected to the feature extraction and analysis modules. Through a built-in weighting model, it matches and calculates the historical authorization range sequence and risk indicators. The weighting model calculates the weight ratio of different input features in the final decision, and comprehensively evaluates information from different dimensions based on the weighting. Based on the calculation structure, it generates specific decision information, which includes at least three types: allow (allow data to be provided as requested); conditional restriction (allow data to be provided but requires specific processing); and complete block (prohibit the provision of the requested data).

[0018] The execution module connects to the decision module and is used to perform corresponding control operations based on the received decision information. For example, when the decision information is to allow access, the execution module allows the requester to directly access the data; when it is to completely block access, the execution module refuses access and returns an error message. When it is a conditional restriction, the execution module activates the built-in dynamic data masking scheme to perform desensitization, generalization, or noise addition on the original data to ensure that the final output data product meets the minimum necessary principle, that is, only the minimum data necessary to complete the processing purpose is provided, and the impact on personal privacy or trade secrets is minimized as much as possible.

[0019] The traceability recording module is connected to all the modules mentioned above and is used to collect decision-related information of the entire system process in real time, including but not limited to the original request content, input and output data of each module, intermediate calculation results, final decision information, and detailed logs of the execution operations. By storing this information and building an immutable decision traceability chain based on blockchain technology, the authenticity and integrity of the trajectory are ensured by storing each step of the operation, providing highly credible evidence for subsequent compliance review, accident accountability, and regulatory evidence collection.

[0020] Preferably, the dynamic asset mapping module further includes:

[0021] The change monitoring unit is used to monitor changes in data asset information associated with the data subject in real time, and automatically trigger the update of the dynamic asset panorama, while recording the change information of the update of the dynamic asset panorama.

[0022] Preferably, it further includes:

[0023] The asset and decision mapping module, connected to the change monitoring unit, is used to record the relationship between changes in the data asset information associated with the data subject and the decision information;

[0024] When the change monitoring unit detects a change in the data asset information associated with the data subject, it automatically triggers the decision module to restart, so as to review the decision information according to the association and generate a review report.

[0025] Preferably, the execution module further includes:

[0026] When the change monitoring unit detects a change in the data asset information associated with the data subject, the execution module is also used to switch to the execution of the new decision information contained in the review report, and to mark the change in the data asset information associated with the data subject corresponding to the review report as important.

[0027] Preferably, the analysis module further includes:

[0028] The text understanding unit has a built-in deep learning model trained based on the legal information database and judicial precedents. It is used to perform semantic analysis on the ambiguous expressions in the legal information database in combination with existing enforcement cases, so as to generate an executable rule containing specific data ratios to replace or supplement the ambiguous legal provisions in the legal information database.

[0029] Preferably, it further includes:

[0030] The collaboration module is used to establish encrypted communication links with compliance detection systems of multiple data platforms, and to collaboratively train a global risk assessment model without sharing the original data. The trained model parameters are then synchronized to the analysis module to update the clause understanding unit or the legal information database.

[0031] Preferably, the decision-making module specifically includes:

[0032] The weighting model is a dynamic weighting model, which introduces external regulatory policy change index and industry compliance event heat factor as weight adjustment variables;

[0033] When the external regulatory policy change index or the industry compliance event heat factor is detected to exceed the preset threshold, the weight allocation of various indicators in the adjustment decision is automatically triggered.

[0034] Preferably, it further includes:

[0035] The multi-scenario consistency maintenance module is used to maintain multiple copies of permission configuration information in multiple different business application scenarios and establish a copy change synchronization mechanism. When the permission configuration information corresponding to any business scenario changes, it automatically triggers the consistency update of the corresponding copies in other scenarios.

[0036] Preferably, the execution module further includes:

[0037] When the decision information is a conditional decision, the execution module initiates a dynamic data masking scheme under the conditional decision.

[0038] The dynamic data masking scheme adopts an isolated execution environment based on data sandbox technology. In this environment, the original data is desensitized, generalized, or noise-added, and the entire masking operation process is recorded for traceability and verification.

[0039] Preferably, the feature extraction module further includes:

[0040] The behavioral baseline analysis unit is used to establish a multidimensional behavioral baseline based on the historical access behavior of the data subject using a time series analysis algorithm, and to compare the real-time behavior of the data subject with the multidimensional behavioral baseline.

[0041] When the real-time behavior deviates from the multidimensional behavior baseline, it is identified as an abnormal access behavior that deviates from the baseline, and the abnormal access behavior is used as a feature for evaluating the risk indicator for the decision module to perform matching calculation;

[0042] The multidimensional behavioral baseline includes the access frequency baseline, the access time distribution baseline, and the access data volume baseline.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] Through the dynamic asset mapping module and its deployed data security monitoring probes, a dynamic asset panorama can be built and continuously updated. This helps to solve the problem that traditional static ledgers cannot cover dynamically changing assets, eliminates blind spots in data supervision, and provides a complete, accurate, and real-time data foundation for all subsequent compliance analyses. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0046] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0047] An automated system for detecting the compliance of public data products, comprising:

[0048] The request parsing module is used to receive data processing request information and parse the data processing request information, which includes the data body, processing purpose and timestamp.

[0049] The dynamic asset mapping module is used to collect data asset information associated with the data subject in real time through data security monitoring probes, so as to build and continuously update the dynamic asset panorama, and automatically identify and mark shadow assets that are not included in the ledger.

[0050] The feature extraction module is used to obtain the historical authorization range sequence and asset sensitivity level labels of the data subject from the dynamic asset panorama.

[0051] The analysis module includes a built-in legal information database, which is used to compare and match the processing purpose and asset sensitivity level tags with the legal information database to generate risk indicators;

[0052] The decision-making module is used to match and calculate historical authorization range sequences and risk indicators through a built-in weight model to generate decision information;

[0053] The execution module is used to execute corresponding access control based on the decision information;

[0054] The traceability recording module is used to collect decision-making information throughout the entire process and to build an immutable decision traceability chain based on blockchain technology.

[0055] The request parsing module serves as the system's input interface, receiving data processing requests from external sources and performing structured parsing on these requests. Each data processing request includes at least three key fields: data body (to uniquely identify the requester), processing purpose (to clarify the requester's intent to use the data), and timestamp (to record the time the request was initiated). Through the request parsing module, the system can convert external requests into a unified, standardized data format for use in all subsequent processing stages.

[0056] The dynamic asset mapping module is connected to the request parsing module. It collects data information related to the data subject in real time through various data security probes deployed at key network nodes, application service interfaces, and database storage. Data security probes include, but are not limited to, full traffic probes, API interface probes, and database scanning tools. The dynamic asset mapping module uses the collected information to build and continuously update a dynamic asset panorama that includes structured data, unstructured data, and semi-structured data. The dynamic asset panorama can reflect the real-time status, distribution location, flow relationship, and attribute tags of data assets. Through automatic identification and marking functions, the dynamic asset mapping module can discover shadow assets that are not registered in the organization's formal asset ledger but actually exist in the network, and mark these shadow assets as assets not included in the management, thereby eliminating regulatory blind spots.

[0057] The feature extraction module is connected to the dynamic asset mapping module. It is used to extract the relevant historical authorization range sequence and the asset sensitivity level of the data subject from the dynamic asset panorama for the data subject of the current request. The historical authorization range sequence records the list of data assets that the data subject was authorized to access in the past specified time period and the authorization period. The asset sensitivity level is the sensitivity level of the data subject based on the data content recognition result, thereby transforming the raw stone information in the dynamic asset panorama into structured features that can be used for decision reference.

[0058] The analysis module stores relevant laws, regulations, departmental rules, industry standards, and other normative clauses related to the flow of public data products through a built-in, expandable legal information database. It takes the processing purpose and asset sensitivity level label as input information, compares and matches them with the clauses in the legal information database, and evaluates the legality of the request through preset matching logic or algorithms to generate a risk indicator that can reflect the degree of risk, such as a risk value expressed as a percentage or level.

[0059] The decision-making module is connected to the feature extraction and analysis modules. Through a built-in weighting model, it matches and calculates the historical authorization range sequence and risk indicators. The weighting model calculates the weight ratio of different input features in the final decision, and comprehensively evaluates information from different dimensions based on the weighting. Based on the calculation structure, it generates specific decision information, which includes at least three types: allow (allow data to be provided as requested); conditional restriction (allow data to be provided but requires specific processing); and complete block (prohibit the provision of the requested data).

[0060] The execution module connects to the decision module and is used to perform corresponding control operations based on the received decision information. For example, when the decision information is to allow access, the execution module allows the requester to directly access the data; when it is to completely block access, the execution module refuses access and returns an error message. When it is a conditional restriction, the execution module activates the built-in dynamic data masking scheme to perform desensitization, generalization, or noise addition on the original data to ensure that the final output data product meets the minimum necessary principle, that is, only the minimum data necessary to complete the processing purpose is provided, and the impact on personal privacy or trade secrets is minimized as much as possible.

[0061] The traceability recording module is connected to all the modules mentioned above and is used to collect decision-related information of the entire system process in real time, including but not limited to the original request content, input and output data of each module, intermediate calculation results, final decision information, and detailed logs of the execution operations. By storing this information and building an immutable decision traceability chain based on blockchain technology, the authenticity and integrity of the trajectory are ensured by storing each step of the operation, providing highly credible evidence for subsequent compliance review, accident accountability, and regulatory evidence collection.

[0062] As an optional embodiment, the dynamic asset mapping module also includes:

[0063] The change monitoring unit is used to monitor changes in data asset information associated with the data subject in real time, and automatically trigger updates to the dynamic asset panorama. It also records change information for the updates to the dynamic asset panorama.

[0064] The change monitoring unit continuously monitors data asset information related to the data subject in real time. When a data asset is added, deleted, structurally changed, or its sensitivity level is changed, the change monitoring unit automatically triggers an update to the dynamic asset panorama. The update uses an incremental update matrix, modifying only the changed parts to improve update efficiency and reduce system load. At the same time, the change monitoring unit records the time, type, and state before and after the change, forming a change information log for subsequent auditing and tracking. This helps ensure that the dynamic asset panorama always remains consistent with the actual asset state, providing an accurate data foundation for other modules.

[0065] As an optional embodiment, it also includes:

[0066] The asset and decision mapping module, connected to the change monitoring unit, is used to record the relationship between changes in data asset information associated with the data subject and decision information.

[0067] When the change monitoring unit detects changes in data asset information associated with the data subject, it automatically triggers the decision module to restart, so as to review the decision information according to the relationship and generate a review report.

[0068] The asset and decision mapping module connects with the change monitoring unit to establish and maintain the correlation between changes in data asset information and historical compliance decision information. For example, it records the data assets involved in each asset change and links them with all compliance decisions previously made based on that asset. When the change monitoring unit detects a change in a specific data asset, the asset and decision mapping module automatically identifies all historical compliance decisions related to that asset based on the pre-established correlation and triggers the decision module to restart. This allows for the review of these historical decisions. Based on the new asset information after the change, the decision module recalculates risk indicators and makes a fusion decision, generating new decision information and outputting a review report. The review report details the decision adjustments caused by the asset change for system administrators or auditors to review.

[0069] As an optional embodiment, the execution module further includes:

[0070] When the change monitoring unit detects changes in the data asset information associated with the data subject, the execution module is also used to switch the execution review report to include new decision information, and to mark the changes in the data asset information associated with the data subject corresponding to the review report as important.

[0071] When the change monitoring unit detects a change in data asset information associated with the data subject, and the asset and decision mapping module triggers a decision review and generates new decision information, the execution module automatically switches to execute the new decision information contained in the review report. That is, it stops executing the old decision and instead executes the updated compliance control strategy. At the same time, the execution module marks the data asset information change event corresponding to the review report with an importance mark. This mark can be a metadata label or a log identifier, used to indicate that the change in the data asset information has caused a change in the corresponding decision information plus one, thereby increasing the correlation between the change in data asset information and the decision information, highlighting that the change is a key compliance adjustment triggered by a change in asset status, so that it can be given priority attention and traced in subsequent audits, reports or monitoring views.

[0072] As an optional embodiment, the analysis module further includes:

[0073] The text comprehension unit has a built-in deep learning model trained on a legal information database and judicial precedents. It is used to perform semantic analysis on ambiguous expressions in the legal information database in combination with existing enforcement cases, so as to generate an enforceable rule containing specific data ratios to replace or supplement the ambiguous legal provisions in the legal information database.

[0074] The text understanding unit incorporates a deep learning model trained on a legal information database and a large amount of judicial case data. This model uses natural language processing technology to learn the semantic features of legal provisions and their application in real cases. When certain provisions in the legal information database contain ambiguous expressions, such as descriptions of necessary scope or reasonable measures, the text understanding unit can perform in-depth semantic analysis of these ambiguous expressions by combining the specific context of the current request and existing enforcement cases. For example, it can quantify a large amount of data into more than 1,000 records or refer to existing judgments in real cases to implement reasonable measures. These generated enforceable rules can be used to replace or supplement ambiguous legal provisions in the original legal information database, thereby improving the accuracy and operability of compliance analysis.

[0075] As an optional embodiment, it also includes:

[0076] The collaboration module is used to establish encrypted communication links with compliance detection systems of multiple data platforms, and to collaboratively train a global risk assessment model without sharing the original data. The trained model parameters are then synchronized to the analysis module to update the clause understanding unit or legal information database.

[0077] The collaboration module establishes encrypted communication links with compliance detection systems deployed on multiple heterogeneous data platforms. Employing federated learning technology, it organizes all participants to collaboratively train a global risk assessment model without sharing the original data of each platform. Each platform trains its model using local data, uploading only encrypted model gradients or parameters to the collaboration module, which then aggregates and updates them to generate a global model that integrates knowledge from multiple parties. After training, the collaboration module synchronizes the updated model parameters to the analysis modules of each participating platform to update their local text understanding units or legal information databases. This achieves cross-platform knowledge sharing and model evolution while strictly protecting the data privacy of each platform.

[0078] As an optional embodiment, the decision module specifically includes:

[0079] The weighting model is a dynamic weighting model, which introduces external regulatory policy change index and industry compliance event heat factor as weight adjustment variables;

[0080] When the external regulatory policy change index or the industry compliance event heat factor is detected to exceed the preset threshold, the weight allocation of various indicators in the decision-making process will be automatically triggered.

[0081] The weighting model is configured as a dynamic weighting model, introducing an external regulatory policy change index and an industry compliance event heat factor as weight adjustment variables. The external regulatory policy change index can be generated by monitoring data such as the frequency of policy releases and the number of new regulations issued on the websites of relevant regulatory agencies. The industry compliance event heat factor can be generated by collecting reports on similar data compliance events from channels such as cybersecurity news, industry reports, and administrative penalty announcements. The dynamic weighting model obtains the values ​​of these two variables in real time or periodically and compares them with preset thresholds. When the monitored value of either variable exceeds the preset threshold, the model automatically triggers a reallocation of the weights of historical authorization range sequences and risk indicators in the decision-making process. For example, the weight of legal risk indicators is increased, and the weight of historical behavior is decreased, so that the system can automatically adopt a more stringent compliance review strategy when the regulatory environment becomes stricter or industry risks increase.

[0082] As an optional embodiment, it also includes:

[0083] The multi-scenario consistency maintenance module is used to maintain multiple copies of permission configuration information in multiple different business application scenarios and establish a copy change synchronization mechanism. When the permission configuration information corresponding to any business scenario changes, it automatically triggers the consistency update of the corresponding copies in other scenarios.

[0084] The multi-scenario consistency maintenance module is used to handle the consistency of permission configuration information of the same public data product in multiple different business application scenarios. By maintaining multiple copies of permission configuration information in each business scenario and establishing a copy change synchronization mechanism, when the permission configuration information corresponding to any business scenario changes, such as modifying the access frequency limit or adjusting the range of accessible data fields, the module can automatically detect the change and immediately trigger the synchronization of the changed content to the configuration copies of all other related business scenarios. This ensures that the permission policies in all scenarios are consistent, which helps to avoid security vulnerabilities or inconsistent user experience issues caused by inconsistent configurations.

[0085] As an optional embodiment, the execution module further includes:

[0086] When the decision information is a conditional decision, the execution module initiates a dynamic data masking scheme under the conditional decision.

[0087] The dynamic data masking scheme adopts an isolated execution environment based on data sandbox technology. In this environment, the original data is desensitized, generalized, or noise-added, and the entire masking operation process is recorded for traceability and verification.

[0088] When conditional decision-making is required, the system creates a temporary sandbox environment isolated from the external network for the requester. The original data is securely loaded into this sandbox, and all data processing operations are completed within the sandbox. After processing, only the result data that meets the requirements is output to the requester. The sandbox environment is then destroyed, and all temporary data is securely erased. The entire masking operation is recorded in detail by the traceability module for subsequent verification, including data input, processing steps, algorithm parameters, and output results. This achieves the highest security requirements of ensuring that the original data does not leave the domain and that the data is usable but not visible.

[0089] As an optional embodiment, the feature extraction module further includes:

[0090] The behavioral baseline analysis unit is used to establish a multidimensional behavioral baseline based on the historical access behavior of the data subject using time series analysis algorithms, and to compare the real-time behavior of the data subject with the multidimensional behavioral baseline.

[0091] When real-time behavior deviates from the multidimensional behavior baseline, it is identified as abnormal access behavior that deviates from the baseline, and the abnormal access behavior is used as a feature of the risk assessment indicator for the decision-making module to perform matching calculations.

[0092] Among them, the multidimensional behavioral baseline includes the access frequency baseline, the access time distribution baseline, and the access data volume baseline;

[0093] The behavioral baseline analysis unit, based on time series analysis algorithms, models the historical access behavior of each data subject, establishing multi-dimensional behavioral baselines, including: access frequency baseline (the distribution of the number of requests per unit time); access time distribution baseline (the pattern of request occurrences over time); and access data volume baseline (the distribution of the number of data records or the size of data involved in each request). When actually processing data requests, this unit compares the real-time behavioral characteristics of the current request with the pre-established individualized behavioral baselines. When a real-time behavior is detected to deviate significantly from the baseline in multiple dimensions, it is identified as an abnormal access behavior. This abnormal access behavior is used as a new feature input to assess the risk indicators of the current request and is then fed into the decision-making module for fusion calculation, thereby enhancing the perception and identification capabilities of dynamic security risks such as internal personnel violations, account theft, and data crawling.

[0094] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A public data product compliance automatic detection system, characterized in that, include: The request parsing module is used to receive data processing request information and parse the data processing request information, which includes the data subject, processing purpose and timestamp; The dynamic asset mapping module is used to collect data asset information associated with the data subject in real time through data security monitoring probes, so as to construct and continuously update the dynamic asset panorama, and automatically identify and mark shadow assets that are not included in the ledger. The feature extraction module is used to obtain the historical authorization range sequence and asset sensitivity level label of the data subject from the dynamic asset panorama; The analysis module includes a built-in legal information database, which is used to compare and match the processing purpose and the asset sensitivity level label with the legal information database to generate risk indicators; The decision-making module is used to match and calculate the historical authorization range sequence and the risk indicators through a built-in weight model to generate decision information; The execution module is used to execute the corresponding access control based on the decision information; The traceability recording module is used to collect decision-making information throughout the entire process and to build an immutable decision traceability chain based on blockchain technology.

2. The system of claim 1, wherein, The dynamic asset mapping module also includes: The change monitoring unit is used to monitor changes in data asset information associated with the data subject in real time, and automatically trigger the update of the dynamic asset panorama, while recording the change information of the update of the dynamic asset panorama.

3. The system of claim 2, wherein, Also includes: The asset and decision mapping module, connected to the change monitoring unit, is used to record the relationship between changes in the data asset information associated with the data subject and the decision information; When the change monitoring unit detects a change in the data asset information associated with the data subject, it automatically triggers the decision module to restart, so as to review the decision information according to the association and generate a review report.

4. The system of claim 3, wherein, The execution module further includes: When the change monitoring unit detects a change in the data asset information associated with the data subject, the execution module is also used to switch to the execution of the new decision information contained in the review report, and to mark the change in the data asset information associated with the data subject corresponding to the review report as important.

5. The system of claim 3, wherein, The analysis module also includes: The text understanding unit has a built-in deep learning model trained based on the legal information database and judicial precedents. It is used to perform semantic analysis on the ambiguous expressions in the legal information database in combination with existing enforcement cases, so as to generate an executable rule containing specific data ratios to replace or supplement the ambiguous legal provisions in the legal information database.

6. The system of claim 5, wherein, Also includes: The collaboration module is used to establish encrypted communication links with compliance detection systems of multiple data platforms, and to collaboratively train a global risk assessment model without sharing the original data. The trained model parameters are then synchronized to the analysis module to update the clause understanding unit or the legal information database.

7. The system of claim 1, wherein, The decision-making module specifically includes: The weighting model is a dynamic weighting model, which introduces external regulatory policy change index and industry compliance event heat factor as weight adjustment variables; When the external regulatory policy change index or the industry compliance event heat factor is detected to exceed the preset threshold, the weight allocation of various indicators in the adjustment decision is automatically triggered.

8. The system of claim 1, wherein, Also includes: The multi-scenario consistency maintenance module is used to maintain multiple copies of permission configuration information in multiple different business application scenarios and establish a copy change synchronization mechanism. When the permission configuration information corresponding to any business scenario changes, it automatically triggers the consistency update of the corresponding copies in other scenarios.

9. The system of claim 1, wherein, The execution module further includes: When the decision information is a conditional decision, the execution module initiates a dynamic data masking scheme under the conditional decision. The dynamic data masking scheme adopts an isolated execution environment based on data sandbox technology. In this environment, the original data is desensitized, generalized, or noise-added, and the entire masking operation process is recorded for traceability and verification.

10. The system of claim 1, wherein, The feature extraction module further includes: The behavioral baseline analysis unit is used to establish a multidimensional behavioral baseline based on the historical access behavior of the data subject using a time series analysis algorithm, and to compare the real-time behavior of the data subject with the multidimensional behavioral baseline. When the real-time behavior deviates from the multidimensional behavior baseline, it is identified as an abnormal access behavior that deviates from the baseline, and the abnormal access behavior is used as a feature for evaluating the risk indicator for the decision module to perform matching calculation; The multidimensional behavioral baseline includes the access frequency baseline, the access time distribution baseline, and the access data volume baseline.

Images