Data governance method, apparatus, device, medium, and program product

Abstract

The application provides a data governance method, device, equipment, medium and program product, and relates to the field of data analysis. The method comprises the following steps: capturing and outputting a target event based on a double-path governance triggering mechanism, wherein the target event comprises a data quality abnormal event and a policy change event, and the double-path governance triggering mechanism is used for monitoring data quality abnormalities and policy change behaviors through two paths; constructing a directional blood relationship graph based on the target event; locating a risk root cause based on the directional blood relationship graph, and outputting a path emergency degree classification result corresponding to each path in the directional blood relationship graph; and performing data governance based on the risk root cause and the path emergency degree classification result. The method provided in the application embodiment can improve the efficiency of data governance.

Description

Technical Field

[0001] This application relates to the field of data analysis, and in particular to a data governance method, apparatus, equipment, medium, and program product. Background Technology

[0002] In related technologies, data governance platforms generally incorporate data lineage analysis, tag management, and strategy evaluation modules to support the basic framework of compliance governance. These platforms, employing a static, full-scale lineage graph construction method, while possessing global view capabilities, suffer from coarse-grained graphs and high computational redundancy in scenarios addressing specific issues and events. This makes it difficult to quickly pinpoint the root causes of risks, resulting in low efficiency in data governance. Summary of the Invention

[0003] This application provides a data governance method and apparatus that can solve the problem of low efficiency in data governance in scenarios facing specific problem events, where the granularity of the data map is coarse, the computational redundancy is high, and it is difficult to quickly locate the root cause of the risk.

[0004] To solve the above-mentioned technical problems, this application is implemented as follows:

[0005] In a first aspect, embodiments of this application provide a data governance method, the method comprising:

[0006] The dual-path governance triggering mechanism captures and outputs target events, including data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two separate paths.

[0007] Based on the target event, a targeted kinship map is constructed;

[0008] Based on the directional kinship map, the root causes of risk are located, and the path urgency classification results corresponding to each path in the directional kinship map are output.

[0009] Data governance is performed based on the aforementioned root causes of risk and the path urgency classification results.

[0010] Secondly, embodiments of this application provide a data governance apparatus, the apparatus comprising:

[0011] The capture module is used to capture and output target events based on a dual-path governance triggering mechanism. The target events include data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two separate paths.

[0012] A construction module is used to construct a directional kinship map based on the target event;

[0013] The localization module is used to locate the root causes of risk based on the directional kinship map and output the path urgency classification results corresponding to each path in the directional kinship map.

[0014] The data governance module is used to perform data governance based on the risk root causes and the path urgency classification results.

[0015] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores a program or instructions executable on the processor, and the program or instructions, when executed by the processor, implement the steps of the data governance method as described in the first aspect.

[0016] Fourthly, embodiments of this application provide a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the data governance method as described in the first aspect.

[0017] In this embodiment, a dual-path governance triggering mechanism is used to capture and output target events, including data quality anomaly events and policy change events. The dual-path governance triggering mechanism monitors data quality anomalies and policy change behaviors through two separate paths. Based on the target events, a directional lineage graph is constructed. The root causes of risks are located based on the directional lineage graph, and the path urgency classification results for each path in the directional lineage graph are output. Data governance is then performed based on the root causes of risks and the path urgency classification results. In this way, by simultaneously monitoring and capturing two types of target events—data quality anomalies and policy changes—through the dual-path governance triggering mechanism, and combining the target events to drive the construction of a directional lineage graph to narrow the analysis scope and reduce unnecessary computational power consumption, the directional lineage graph accurately locates the root causes of risks and classifies the path urgency, and the combined results of the root causes of risks and the path urgency classification are used to collaboratively carry out data governance. This achieves integrated perception, lightweight analysis, and precise tracing of multiple types of data risks, effectively improving the efficiency and accuracy of data risk governance. Attached Figure Description

[0018] Figure 1 One of the flowcharts illustrating the data governance method provided in this application embodiment;

[0019] Figure 2 A second schematic flowchart illustrating the data governance method provided in this application embodiment;

[0020] Figure 3 A schematic diagram of the structure of the data governance device provided in the embodiments of this application;

[0021] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] For ease of understanding, the following describes some aspects of the embodiments of this application:

[0024] Existing data governance platforms generally incorporate data lineage analysis, tag management, and policy evaluation modules to support the basic framework of compliance governance. They typically employ the following technical solutions:

[0025] 1. Static lineage graph modeling technology: Mainstream governance platforms often construct complete data lineage graphs by extracting, transforming, loading (ETL) logs and extracting metadata to support data flow tracing and risk analysis. However, such graphs are usually built statically with full data, resulting in delayed updates and coarse granularity, making it difficult to adapt to high-frequency data changes and policy iteration scenarios.

[0026] 2. Tag-Driven Policy Matching Mechanism: To achieve sensitive data identification and usage control, existing platforms widely use field-level tagging systems and construct tag-policy mapping rules. For example, "Protected Health Information (PHI) data" corresponds to a specific anonymization policy, and "Personally Identifiable Information (PII) data + advertising purpose" triggers an authorization process. However, such policy matching mostly relies on static rule tables or predefined templates, lacking dynamic generation and combination capabilities, and making it difficult to cope with the compliance complexities brought about by changes in tag combinations.

[0027] 3. Rule Engine-Based Compliance Assessment Method: In the policy execution phase, most platforms adopt a rule engine-based matching and judgment method to linearly resolve policy conflicts according to priority. However, this type of mechanism often cannot effectively handle execution ambiguities caused by multi-source policy conflicts, semantic overlap, or contextual differences.

[0028] While existing data governance platforms possess certain compliance governance capabilities in areas such as lineage analysis, tag management, and policy evaluation, they still exhibit the following technical shortcomings when facing complex governance scenarios involving high dynamism, multiple policies, and multi-tag fusion:

[0029] 1. The governance trigger mechanism is singular and has a delayed response, making it difficult to adapt to multi-source risk scenarios. Existing platforms mainly rely on rule engines, periodic inspections, or manual audits for risk identification and governance triggering, which generally suffers from passive response, coarse trigger granularity, and limited coverage. Dynamic events such as data quality fluctuations and policy configuration changes often cannot serve as governance entry points, resulting in potential risks not being identified and responded to in a timely manner.

[0030] 2. Static lineage graph construction results in high computational redundancy, making it difficult to support targeted issue tracing. Most platforms adopt a static, full-scale lineage graph construction method. Although it has global view capabilities, in scenarios facing specific issues and events, the graph granularity is coarse and updates are not timely, making it difficult to quickly trace the source and locate key transmission paths. It also consumes a lot of computational resources, affecting governance efficiency.

[0031] 3. The identification of risk propagation paths is crude, the ability to focus resources is weak, and the formulation of strategies lacks precision. The current governance process generally lacks the ability to analyze the risk intensity of the label propagation path, and cannot select key nodes and paths based on actual impact. High-priority risks are easily submerged in a large number of ineffective paths, resulting in unreasonable allocation of strategy resources and poor governance effects.

[0032] Please see Figure 1 , Figure 1 As shown in the figure, one of the flowcharts of a data governance method provided in this application embodiment includes:

[0033] Step 110: Capture and output target events based on the dual-path governance triggering mechanism. The target events include data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two paths respectively.

[0034] In this step, the aforementioned dual-path governance triggering mechanism refers to two independent monitoring paths operating in parallel. This mechanism can simultaneously monitor the status of different business dimensions to distinguish and identify different types of abnormal behavior. The paths can continuously collect the data operation status within the platform through real-time polling or passively receive abnormal information pushed by the platform through proactive event reporting. In the initial stage of method execution, the system relies on the dual-path governance triggering mechanism to capture and output target events. These target events refer to various data risk events generated during platform operation, which can be divided into two categories: data quality anomaly events and policy change events. Data quality anomaly events are those caused by anomalies in the data itself or the data processing process. These events can be triggered by changes to the data table structure or by core business indicator values ​​deviating from the normal range. Policy change events are those caused by adjustments to data compliance-related configurations. These events can be triggered by adjustments to the content of data sensitive tags or by changes to platform compliance control rules.

[0035] Step 120: Based on the target event, construct a directional kinship map;

[0036] In this step, the aforementioned targeted lineage graph is a link model built upon the natural flow dependencies of data. This model focuses only on data nodes and transmission paths directly related to the current target event, and does not cover all data assets on the platform. The targeted lineage graph can represent various data assets and their dependencies using nodes and connections, or it can use a network topology structure to outline the complete data flow links. Around the acquired target event, the system can define the scope of related data and build a corresponding targeted lineage graph. By using targeted analysis, the computational scope is reduced, lowering the resource consumption of a full traversal.

[0037] Step 130: Locate the root cause of risk based on the directional kinship map, and output the path urgency classification results corresponding to each path in the directional kinship map;

[0038] The aforementioned root causes of risk are the source data nodes or data processing operations that trigger the current target event, and also the starting point of the entire risk chain. The root cause can be a data table with an abnormal operating status, or a data calculation operation with an error in execution logic. The aforementioned path urgency classification results are the results of classifying each data flow path within the directional lineage graph according to the scope and severity of risk impact. Path urgency can be divided into three basic levels: high, medium, and low, or more tiered levels can be defined based on the platform's control granularity requirements. The system can trace back along the data links in the directional lineage graph to accurately locate the corresponding root causes of risk, and simultaneously conduct risk assessments on each data flow path within the graph, classifying the urgency level of each path and outputting the corresponding classification results.

[0039] Step 140: Perform data governance based on the risk root causes and the path urgency classification results.

[0040] When conducting data governance, the specific targets of governance work can be determined based on the root causes of risks, and appropriate handling methods can be matched with the urgency level classification results of the paths to achieve differentiated control. Data governance work can carry out emergency blocking operations to quickly curb the continued spread of existing risks, or it can carry out compliance investigation operations to uncover hidden risks in the chain and ensure the overall operational security and compliance of data assets.

[0041] For example, you can refer to Figure 2 In the data governance method of this application, a closed-loop intelligent governance process is formed from problem identification to path analysis, risk positioning, and response decision-making by constructing a multi-source perception triggering mechanism, an event-driven directional lineage map, an impact assessment path that integrates risk scoring, and an adaptive control compliance response strategy.

[0042] In the data governance method of this application, a dual-path governance triggering mechanism is used to simultaneously monitor and capture two types of target events: data quality anomalies and policy changes. A directional lineage graph is constructed in conjunction with the target events to narrow the analysis scope and reduce ineffective computing power consumption. Based on the directional lineage graph, the root cause of risks is accurately located and the path urgency is graded. Data governance is carried out in conjunction with the results of the root cause of risks and the path urgency. This achieves integrated perception, lightweight analysis and precise source tracing of multiple types of data risks, effectively improving the efficiency and accuracy of data risk governance.

[0043] Furthermore, addressing the common issues in existing data governance platforms such as a single governance trigger mechanism, low efficiency of graph analysis, and insufficient risk path identification capabilities, the data governance method implemented in this application constructs a dual-pathway governance trigger mechanism to include data quality fluctuations and policy change events in the scope of proactive monitoring, achieving real-time and comprehensive risk identification. It proposes an event-driven, targeted lineage graph construction method, focusing on key fields centered on events to dynamically generate the minimum problem propagation path, improving the accuracy and efficiency of graph construction. Furthermore, by combining graph structural features and node influence factors, a response path identification mechanism integrating risk scoring is designed to accurately identify high-risk paths and prioritize the allocation of policy resources, thereby improving the response speed, coverage depth, and risk control capabilities of data compliance governance.

[0044] Optionally, the dual-path governance triggering mechanism includes a data quality anomaly triggering path and a policy change sensitive metadata triggering path. The capture and output of target events based on the dual-path governance triggering mechanism includes:

[0045] Based on the data quality anomaly triggering path, and combined with at least one of the acquired data structure changes, business indicator fluctuations, and critical period anomalies, the data quality anomaly event is output.

[0046] Based on the sensitive metadata triggering path for policy changes, and combined with at least one of the obtained sensitive label adjustments and compliance policy configuration changes, the policy change event is output.

[0047] In the data governance process, quality anomalies are the most direct source of risk affecting business accuracy, while policy changes often imply potential compliance risks and accountability gaps. Both are high-frequency and high-risk triggering scenarios. Therefore, this implementation method can build a unified triggering mechanism around these two high-risk scenarios of quality anomalies and policy changes. By dynamically monitoring rule and metadata changes through the governance platform, it can achieve automatic identification and real-time dissemination of abnormal events.

[0048] The mechanism can consist of two parts: proactively capturing data-level errors by running quality rules; and identifying potential compliance risks by recognizing changes in sensitive labels or policy configurations.

[0049] In this implementation, the dual-path governance triggering mechanism within the overall monitoring system can be divided into two independently operating monitoring paths. Each path has its own function, identifying and outputting different types of risk behaviors. One of these paths is the data quality anomaly triggering path, which specifically monitors the operational status of the data itself. When operating based on this path, the system continuously collects the operational status of various data within the platform. When any one or more of the following conditions are detected—data structure changes, business indicator fluctuations, or anomalies during critical periods—a corresponding data quality anomaly event will be output. Data structure changes can include the addition or removal of fields within a data table, or modifications to the data table's index rules. Business indicator fluctuations can include core statistical indicators deviating from their normal range, or real-time computation indicators experiencing precipitous changes. Anomalies during critical periods can include data read response delays during peak business periods, or interruptions in data transmission during scheduled synchronization.

[0050] Another type is the sensitive metadata triggering path for policy changes, which is a channel for continuous monitoring of data compliance metadata. When working based on the sensitive metadata triggering path for policy changes, the system focuses on compliance-related configuration content. When either a sensitive label adjustment or a compliance policy configuration change is detected, a corresponding policy change event will be output. Sensitive label adjustments can be adding sensitive identifiers to ordinary business data or adjusting the risk level of existing sensitive data. Compliance policy configuration changes can be modifying cross-business domain access rules or updating the de-identification processing standards for sensitive data. The two paths run independently and in parallel, completing event capture and output according to their respective judgment logic, which can clearly distinguish risk events caused by different factors. At the same time, through the dual-path triggering mechanism for quality anomalies and policy changes, data quality detection results (such as fluctuation indicators) and policy configuration actions (such as indicator adjustments) can be used as proactive governance entry points, which can improve the real-time performance and coverage of governance triggering and break through the existing governance models that rely on static rules or manual driving.

[0051] For example, (1) Quality anomaly triggering based on rule execution:

[0052] The governance platform first supports the structured input of full-volume data quality rules, including applicable tables and fields, validation logic (such as SQL expressions or built-in functions), execution frequency, expected results, responsibility attribution, and applicable tags, forming a unified quality rule directory. After the rules are entered, the system periodically executes validation tasks according to the scheduling strategy and writes the records of failed validations to a failure details table or pushes them to a message queue. Each record contains information such as rule identifier, trigger time, partition key, primary key, and failure reason.

[0053] To overcome the limitations of traditional scheduling methods in terms of response time and business matching, this application proposes a multi-source triggering strategy that integrates metadata changes, business awareness, and key node identification, specifically including:

[0054] (1.1) Structure Change Driven. When a structural meta information change event is detected in the main table, special attention is paid to behaviors such as adding fields, modifying field types, and adjusting primary keys. Once a structural change is identified, the system can: automatically match the affected quality rules and generate default verification rule templates for the newly added fields; trace the downstream assets to see if they need to be synchronously verified and adjusted based on the changed table and field lineage, and immediately trigger the corresponding rule task to verify whether data anomalies have been introduced due to the structural adjustment.

[0055] (1.2) Business Fluctuation Awareness. The system establishes a trend monitoring model for core business indicators (such as daily order volume, number of active users, page views, number of API calls, etc.), calculates their normal fluctuation range through historical time windows, and uses the following formula to determine whether a trigger has been triggered:

[0056]

[0057] in, This represents the observed value of a core business indicator; This represents the historical average of the indicator; This indicates the historical standard deviation of the indicator; The sensitivity coefficient (default value 3) is set to define the range for "abnormal deviation" judgment.

[0058] In some optional implementations, a volatility scoring mechanism may also be included to quantify the severity of an event. The abnormal volatility scoring formula is as follows:

[0059]

[0060] in, To ensure that the weight of the indicators reflects their business importance, the system classifies risk levels based on the scores: a score of ≥4 is considered medium risk, and a score of ≥6 is considered high risk. High-level events will automatically enter the expedited verification and governance work order processing flow.

[0061] (1.3) Enhanced Triggering During Critical Periods. The system, in conjunction with the business calendar, automatically increases the verification frequency of quality rules during common high-risk periods (such as month-end settlements and major promotional launches). This frequency is determined by a function. Control, among which This is the default execution frequency. Risk level, This adds a risk weighting factor. This ensures the system can capture quality issues more densely during peak business periods.

[0062] After the verification task is completed, the system will aggregate the abnormal records according to the principle of "same source and same cause", encapsulate them into structured data quality abnormal events, including failure summary, sample data, rule identifier, trigger type (such as structural change, indicator fluctuation, key time point, etc.) and contextual business tag, and uniformly push them to the message bus and work order platform as a unified entry point for lineage analysis, impact assessment and compliance response.

[0063] In some other alternative implementations, it may also include:

[0064] (2) Sensitive metadata triggering based on policy changes:

[0065] In daily governance, the sensitivity labels of fields and their corresponding policy configurations (such as access control, retention periods, and de-identification requirements) change dynamically due to business development and policy adjustments. Traditional technical solutions often rely on manual registration or infrequent inspections, which carries risks such as delayed response and missed changes. Therefore, this application proposes a sensitive metadata change triggering scheme based on version management and a dual-channel awareness mechanism, building upon the existing capabilities of the metadata platform. The specific implementation process is as follows:

[0066] (2.1) All sensitive labels and compliance policies are version-based. Every time a field label is adjusted (e.g., upgraded from "normal" to "PII") or a policy clause is changed (e.g., the retention period is shortened or the access role is tightened), a version snapshot is automatically recorded and written to the audit table, including the differences before and after the change, the initiator, the time, and the operation channel.

[0067] (2.2) The system can identify changes through a dual-channel approach of active interception and passive comparison. When a user submits modifications through the platform configuration interface, the system immediately constructs a "compliant change event"; at the same time, the platform compares snapshot differences daily to capture policy changes outside the platform path and prevent omissions.

[0068] (2.3) For valid changes identified, the system can automatically calculate a compliance risk score, taking into account the following three dimensions: field sensitivity level. Impact on asset quantity Number of accessed roles The scoring model is as follows:

[0069]

[0070] in, Assign values ​​based on the sensitivity label level (e.g., PII=3, PHI=4, Confidential=2, ordinary fields=0). This refers to the number of tables or fields affected by this change. This indicates the number of roles that have access permissions. , , Adjustable parameter weights (initially set to 0.5, 0.3, 0.2) are used to balance the influence between dimensions.

[0071] Believe that when At times, it is considered high-risk and requires approval within the platform. Downstream processing can only be triggered after review and confirmation by the compliance officer. When the risk level is medium, directly construct and send a compliance change event; when... At this time, the risk is low, and only logs are recorded for audit purposes; no further triggering is required.

[0072] The system can uniformly encapsulate all high- and medium-level compliance change events into structured events and push them to the message bus. This drives the lineage analysis module to automatically identify affected data paths, perform sensitive data diffusion detection and violation risk scanning, forming a closed-loop linkage from event identification to impact analysis. Simultaneously, it proposes target field deconstruction methods for specific fields in governance events (such as indicator fields and lineage primary key fields), and combines multi-granularity lineage analysis at the table, field, and indicator levels to dynamically construct problem-centric targeted lineage graphs, avoiding computational redundancy and path interference caused by traditional full-scale lineage analysis.

[0073] Optionally, constructing a directional kinship map based on the target event includes:

[0074] Extract the event object and risk level from the target event;

[0075] The target for kinship analysis is determined based on the event object, and the depth of exploration is adjusted based on the risk level.

[0076] Based on the stated lineage analysis objectives and the stated exploration depth, the importance scores of each node in the upstream and downstream links associated with the event object are obtained;

[0077] Remove upstream and downstream links whose importance scores are lower than a preset threshold, inject event context into the remaining upstream and downstream links, and output a directional lineage graph.

[0078] In this implementation, data quality anomaly events and compliance policy change events captured in the previous stage can be used as input for path analysis, connecting to the event-driven lineage analysis engine built in this solution to achieve link identification and impact tracing. Unlike the traditional approach of full lineage and static query, this application introduces mechanisms such as event binding, path compression, node risk injection, and graph output, which reduces system resource consumption and enables more real-time and controllable risk path modeling at key nodes.

[0079] In this implementation, the event object and risk level can be extracted from the target event first. The event object is the core data subject that triggers this data governance process and serves as the benchmark for all link analysis. The event object can be a data form carrying core business or a scheduling task that performs data processing and flow. The risk level is the initial risk assessment standard inherent in the target event, which can characterize the initial degree of harm and scope of impact of the event, and can provide a core basis for adjusting the scope of subsequent graph exploration.

[0080] After extracting the core parameters, the lineage analysis target can be determined based on the event object, and the exploration depth can be adjusted based on the risk level. The aforementioned lineage analysis target can be understood as the core analytical direction of this data lineage tracing, used to define the overall logic of the link traversal. The lineage analysis target can be the original source of data generation upstream, or the subsequent flow and impact of data downstream. The exploration depth represents the maximum hierarchical range of the lineage graph traversal and parsing data links. The system can adaptively adjust the exploration depth according to the event risk level. High-risk events can be matched with a larger exploration depth to ensure complete coverage of all associated risk links, while low-risk events can be matched with a smaller exploration depth to reduce the traversal calculation of invalid links.

[0081] After determining the goals and depth of the lineage analysis, within the defined analytical scope, all upstream and downstream links that have data dependencies on the event object can be traversed to obtain the importance score of each data node in each link. The node importance score can be used to quantitatively characterize the core weight and influence of a single data node in the entire data flow link. The calculation dimension of the score can refer to the business importance of the node, or the frequency of daily data calls and accesses to the node.

[0082] After calculating the scores for all nodes, the system performs redundant link screening, removing upstream and downstream links whose node importance scores are below a preset threshold. The preset threshold is a standardized judgment value pre-configured by the system, used to distinguish between core effective nodes and invalid redundant nodes. The preset threshold can be set statistically based on the platform's daily operation and maintenance experience, or it can be trained and optimized based on link data from historical risk events.

[0083] After filtering out invalid links, the system injects event context into the remaining valid upstream and downstream links. Event context is specific auxiliary information tailored to the target event. It can record business scenario information about the event's occurrence, as well as the event's anomaly characteristics and occurrence time. By binding specific event contexts to valid links, the final constructed graph accurately matches the characteristics of the current risk event. After all processing is complete, the system outputs a targeted lineage graph adapted to the current risk analysis scenario.

[0084] The analysis scope can be dynamically narrowed based on the actual situation of the event, and invalid data links can be filtered out, effectively reducing the computing power consumption of map construction and subsequent risk analysis, and improving the accuracy and efficiency of data risk tracing.

[0085] For example, the following specific steps may be included:

[0086] (1) Definition of Event Analysis and Interpretation Objectives:

[0087] The system can begin by analyzing quality anomalies and compliance change events to extract their core metadata, including event type (e.g., rule failure, label change), object identifier (table, field, rule identification ID), triggering cause, and contextual tags (business domain, sensitivity level, source platform, etc.). For different event types, the system automatically matches appropriate path analysis targets. For example, quality events primarily focus on identifying "upstream causes and downstream impacts," aiming to trace the root cause of the problem and locate affected assets; while compliance events emphasize identifying "sensitive label propagation paths and potential violation nodes" for subsequent risk scanning and strategy assessment.

[0088] Based on the above event analysis results, targeted lineage path exploration tasks can be generated on demand, clearly defining the starting node of this task (such as a field, table, or job ID), the exploration direction (supporting upward, downward, or bidirectional searches), the initial risk score (which can be derived from volatility or compliance scores), and most importantly—the dynamically adjustable exploration depth. The path depth is determined by the risk perception path adjustment function:

[0089]

[0090] in, This indicates the risk score of the current event. It is the business activity level of the starting node (such as call frequency or access popularity). As a regulating factor, This is the maximum exploration depth allowed by the platform. The function logic ensures that high-risk, high-sensitivity events will automatically drive deeper path development, while low-activity nodes will receive more conservative exploration control, thereby optimizing resource consumption and task response time while ensuring analysis effectiveness.

[0091] (2) Targeted lineage exploration and path compression mechanism:

[0092] After the path parameters are set, the built-in event-driven lineage analysis engine will be triggered to construct the event propagation path based on metadata lineage. Unlike the traditional full traversal method, this application adopts a three-in-one mechanism combining targeted triggering, path compression, and strategy scheduling to balance computational efficiency and risk location accuracy.

[0093] In some optional implementations, the kinship mapping engine can be scheduled to initiate path expansion logic based on the event initiation node and exploration direction. To control path redundancy and avoid unnecessary resource waste, the exploration process automatically executes a path compression strategy, retaining only key link nodes that directly propagate to event fields, generate computational transformations, or depend on strategies. The system dynamically prunes based on the metadata dependencies between nodes, eliminating paths with no actual risk impact or low weight propagation.

[0094] Furthermore, during the path traversal, the system can calculate the propagation importance score for each node visited. The scoring function considers three core dimensions: access role density, whether it is a critical business node, and historical change frequency. Its calculation formula is as follows:

[0095]

[0096] in Indicates the density of visiting roles. To reference the total number of roles in this field or table, This represents the maximum number of roles that can access all assets in the system. Indicate whether it is a core business node; if it is a core asset. If it is a secondary core asset, then ,otherwise ; For historical frequency of change, This indicates the number of times a field, strategy, or lineage structure has changed within 30 days. Represents the maximum change frequency within 30 days across all nodes; weight , , It can be customized by administrators.

[0097] Subsequently, the system can automatically match the exploration strategy function based on the target analysis type of the event, including: The RootCause() function is scheduled to trace back to the source of the fault. Impact() identifies affected assets and The Propagation() function is a two-way sensitive label propagation function. The strategy selection is based on the event type and context, and is a dynamic decision: for quality rule failure events, RootCause analysis will be the primary method, supplemented by Impact analysis; for label adjustment events, the Propagation model will be called first; in situations where the strategy is uncertain or requires comprehensive evaluation, the three types of functions can be executed concurrently and cross-validated.

[0098] The final path exploration scope is controlled by the risk control perception depth function generated in the preceding steps, ensuring that the breadth and depth of the path are closely coupled with the level of event risk and business importance, achieving a balance between analysis effectiveness and resource consumption.

[0099] (3) Node risk injection and context fusion:

[0100] Furthermore, after the path is constructed, the system can inject the contextual information from the original events into each key node of the lineage, generating an enhanced path graph with "governance annotation capabilities." At this stage, the platform can abstract each node into a governance entity with multi-dimensional characteristics, accompanied by the following structured metadata: node type (field / table / job) and data asset ID; contextual tags such as platform, business domain, and data layer; source event annotations (e.g., abnormal indicator fluctuations, increased sensitivity levels, tightened access policies); node role positioning (e.g., key transit nodes, sensitive exit nodes, downstream highly coupled nodes); and node importance scoring (following the scoring model from the previous step). ).

[0101] Based on this, labels can be automatically generated for key propagation segments in the path, such as abnormal propagation segments, strategy diffusion segments, and high-risk compliance segments, to support rapid identification and coordinated response of downstream modules in a graph-level manner.

[0102] (4) Map output and task docking:

[0103] Furthermore, the path exploration and annotation results can be encapsulated into a structured event lineage graph object, serving as input assets for the standardized governance platform. This graph object can contain original event binding information (such as event ID, risk score, event type, and contextual labels), path exploration parameters (including starting node, direction, depth, and the adopted strategy function), and output a complete path graph in a graph database structure. It also includes a list of high-risk nodes and suggested response strategies (such as data isolation, permission verification, and downstream notification). The generated graph object will be pushed to the impact analysis module, strategy engine, and work order system via a message bus, achieving a closed-loop linkage from event identification to response execution, ensuring that the path analysis results have good transmissibility, interpretability, and downstream usability.

[0104] In the data governance method of this application, by integrating factors such as risk weight, impact intensity, and node type to form a comprehensive scoring model, high-influence paths and key propagation nodes can be screened more accurately, thereby achieving more targeted risk identification and resource focus.

[0105] Optionally, the step of locating the root cause of risk based on the directional kinship map and outputting the path urgency classification results corresponding to each path in the directional kinship map includes:

[0106] The directional kinship graph is transformed into a directed weighted graph, and initial responsibility nodes are marked. The weighted graph contains nodes representing data assets and edges representing data dependencies.

[0107] The responsibility value of each node is calculated iteratively in reverse, and candidate root cause nodes are determined based on the node responsibility values.

[0108] Based on the first condition, candidate root cause nodes are verified, and those that pass the verification are identified as risk root causes.

[0109] Based on node importance, access frequency, and data asset level, obtain the path risk score corresponding to each path in the targeted lineage graph, and classify and output the path urgency classification result corresponding to each path in the targeted lineage graph according to the path risk score.

[0110] The first condition includes at least one of the following:

[0111] There have been structural or strategic changes; there are high-risk operational logics; historical failure modes have been hit; and it is involved in the calculation chain of business metrics.

[0112] In this implementation, the event-driven graph generated in the previous stage serves as input. Through a custom-built responsibility transmission network and dynamic path weighting model, the root cause nodes that cause data quality anomalies or compliance risks can be identified, and their impact scope and urgency can be further assessed. This drives the issuance of circuit breaker mechanisms and automated governance instructions, achieving an integrated closed loop from discovery to response.

[0113] In this implementation, the directional lineage graph is first transformed into a directed weighted graph, and the initial responsible node is marked. The directed weighted graph is a standardized data topology structure adapted for algorithmic operations. This graph contains nodes representing various data assets and edges representing data dependencies between nodes. The initial responsible node is the core node that directly triggers the target event and is the starting point of the entire risk chain. The initial responsible node can be a business data node with data quality anomalies or a configuration metadata node where compliance policy changes have occurred. Transforming the unstructured lineage graph into a directed weighted graph allows for the quantification and structuring of data flow relationships, facilitating subsequent automated iterative calculations.

[0114] After completing the graph structure transformation and node labeling, the responsibility value of each node can be calculated iteratively along the data dependency chain. The node responsibility value can be understood as the weight of each data node's impact on this risk event and its proportion of responsibility. The process can be traced back layer by layer from the initial responsible node, iteratively updating the responsibility value of each upstream node based on the data transmission relationship. After calculating the responsibility value of all nodes, nodes with higher responsibility weights are selected based on their respective responsibility values, thus forming a set of candidate root cause nodes. Candidate root cause nodes can be upstream multi-level data processing nodes or the top-level original data source node.

[0115] After obtaining candidate root cause nodes, compliance and fault characteristic verification can be performed on all candidate root cause nodes in conjunction with the first condition. The candidate root cause nodes that pass the verification are then selected as the final risk root cause. The first condition is a multi-criteria used to identify the true risk root cause. In practice, verification can be completed using a single condition or by combining multiple conditions for comprehensive verification. During the judgment process, if a node has undergone structural or policy changes, it means that there are records of modifications to the node's own data structure or the compliance control policies it is bound to. If a node has high-risk operational logic, it could be due to vulnerabilities in the node's configured data aggregation operations or compliance risks in the node's cross-domain data transfer logic. If a node matches a historical fault pattern, it could be because the node has previously caused similar data anomalies or because the node's operational characteristics match the historical fault characteristics recorded by the platform. If a node participates in the calculation chain of business indicators, it could be because the node's data participates in the statistical calculation of the platform's core operating indicators or because the node's data participates in the real-time calculation of daily business monitoring indicators.

[0116] After identifying the root causes of risks, a quantitative risk assessment can be performed on all flow paths within the targeted lineage graph, obtaining a path risk score for each path. The path risk score calculation integrates three core dimensions: node importance, node access frequency, and data asset level, quantifying the overall risk level of a single link from multiple dimensions. After obtaining the risk scores for each path, the system classifies the scores according to preset grading standards, outputting the path urgency grading result for each path. This distinguishes the degree of risk and urgency of handling different data links, providing a quantitative basis for subsequent tiered circuit breaker measures and compliance risk governance.

[0117] For example, firstly, the event path graph is transformed into a directed weighted graph structure. , where nodes Represents governance entities (such as tables, fields, and jobs), edges Representing data flow dependencies, edge weights The value is determined by the lineage type (e.g., 0.3 for direct transmission, 0.7 for aggregation, and 0.9 for conversion-type jobs). Initial responsibility points (quality anomaly nodes or compliance change points) are marked in the graph, with an initial responsibility value of 1.0. The reverse responsibility propagation calculation process is then initiated using this as the source node.

[0118] In backpropagation, the system iterates through multiple rounds using the following responsibility update formula, comprehensively considering both propagation strength and node risk labeling:

[0119]

[0120] in, Represents a node In the Responsibility value in round iteration; Represents a node In the Responsibility value in each iteration (initially 0 or 1, depending on the event type); Indicates a node All downstream nodes Perform a traversal and summation; Indicates downstream node In the The responsibility value in the wheel; Indicates from node To the node The propagation weight of the edge reflects its data flow type (e.g., transformation job = 0.9, aggregation operation = 0.7, direct transmission = 0.3). The values ​​are dynamically weighted based on event labels (e.g., recent structural change nodes = 1.5, high business volatility nodes = 1.3), and iteration continues until responsibility converges or the maximum depth threshold is reached. Finally, the top-N nodes by responsibility value are extracted as candidate root causes, and multi-evidence verification is performed based on the following four dimensions: whether structural or policy changes have occurred; whether high-risk operational logic exists; whether historical failure patterns are matched; and whether they participate in the calculation chain of key indicators. The system will output a root cause report including confidence level, node type, and supporting evidence for data governance personnel to refer to.

[0121] During the impact assessment phase, the system assigns an urgency score to each downstream path, using the following formula:

[0122]

[0123] in, The score indicates the urgency level of the path node, which determines whether a circuit breaker operation is triggered and its level. The higher the score, the more urgent the situation. The node importance score is represented by factors such as the density of access roles, whether it is a core node, and the frequency of historical changes. This indicates the number of API calls made to a specific node or path within a recent period (e.g., within 5 minutes). This represents the maximum number of API calls observed across all nodes. Indicates the level of the data asset to which the node belongs ( ).

[0124] In the data governance method of this application, by standardizing graph structure transformation, iteratively calculating node responsibility, verifying multi-dimensional conditions, and scoring and classifying path risks, the true source of data risks can be accurately identified. At the same time, the urgency of the risk of each data flow path can be quantified, providing a reliable algorithm basis and data support for subsequent refined and differentiated data governance work.

[0125] Optionally, the data governance based on the root causes of risk and the path urgency classification results includes:

[0126] Based on the path urgency classification results, a three-level circuit breaker response is executed for the root cause of the risk and the corresponding link. The circuit breaker response includes:

[0127] If the path urgency classification result is high urgency, risk blocking operations will be performed on the root cause of the risk and the corresponding link.

[0128] If the path urgency classification result is medium urgency, risk isolation operations are performed on the root cause of the risk and the corresponding link.

[0129] If the path urgency classification result is low urgency, a risk warning operation is performed on the root cause of the risk and the corresponding link.

[0130] In this implementation, based on the identified root causes of risks and the urgency classification of paths, tiered circuit breaker control can be implemented on the risk sources and related data links. Different levels of governance measures can be matched according to the urgency of the risks, so as to achieve precise policy implementation and avoid business losses or untimely risk handling caused by uniform handling.

[0131] The three-level circuit breaker response is a tiered emergency management mechanism set up for data risk links. The overall approach is based on the urgency level of the path, with the root cause of the risk and the corresponding flow link as the sole object of action, to address data risk issues of different levels in a targeted manner.

[0132] When the path urgency level is classified as high urgency, it indicates that the risk associated with the current root cause is highly dangerous and spreads rapidly, severely impacting overall data service operations. In this case, risk blocking operations need to be performed on the root cause and its corresponding path. Risk blocking is the highest level of risk management, used to completely cut off the propagation and operation of risky data. Risk blocking can involve disabling the data read / write permissions of the root cause node or forcibly terminating the data synchronization and computation tasks of the corresponding path, preventing the continued spread of risk from the source.

[0133] When the path urgency level is classified as medium urgency, it indicates a potential risk to the link, with a certain risk of diffusion, but it will not cause a global business failure in the short term. In this case, risk isolation operations are performed on the root cause of the risk and the corresponding link. Risk isolation is a medium-level risk management measure used to physically isolate risk data from normal business data, limiting the scope of risk while ensuring the normal operation of the main business. Risk isolation can be achieved by separately archiving and storing the abnormal data corresponding to the root cause of the risk, or by restricting the cross-node flow permissions of the data in that link, thus achieving localized risk control.

[0134] When the path urgency level is classified as low urgency, it indicates a minor risk level, with only potential compliance or quality issues and no immediate business risks. In this case, a risk warning is issued for the root cause of the risk and the corresponding path. Risk warnings are a mild risk management tool used to record risks and provide manual intervention alerts. Risk warnings can be generated in the data governance platform as a corresponding risk record ledger, or they can be proactively pushed to operations and maintenance personnel along with descriptions of risk points and potential hazards, facilitating subsequent investigation and rectification by staff for refined governance.

[0135] For example, the system can trigger a three-tiered response based on an urgency score:

[0136] L1 (Emergency ≥ 8): Immediately freeze write permissions, disable scheduling tasks, and send an SMS notification to the management team;

[0137] L2 (5≤urgent<8): Lock change permissions, switch to backup data source, and notify the data manager via WeChat Work;

[0138] L3 (3≤urgent<5): Mark the asset as an alert and automatically generate a high-priority work order to push to the responsible person.

[0139] The circuit breaker status will be automatically lifted based on the results of rule reruns or policy scans, ensuring closed-loop execution of the governance process.

[0140] Furthermore, structured governance response instruction sets can be automatically generated based on event types, including atomic operations such as Data Definition Language (DDL) rollback, data source refresh, policy update, and permission revocation. These are then automatically combined into multi-step governance solutions by the decision engine and pushed to the scheduling system, policy platform, or notification system for automatic execution.

[0141] In the data governance method of this application, the above-mentioned layered circuit breaker approach can simultaneously take into account the stability of data business operations and the effectiveness of data risk control. It can accurately identify the scope of governance based on the root causes of risks and effectively improve the rationality and reliability of emergency data governance by matching the appropriate level of urgency.

[0142] Optionally, the data governance based on the root causes of risk and the path urgency classification results further includes:

[0143] The analysis scope is defined by taking the risk root cause as the starting point and combining it with the path urgency classification results;

[0144] Based on the directional lineage map and the high-risk nodes and corresponding sensitive tags within the analysis range, the propagation path of the sensitive tags in the link and the strength of each propagation path are obtained;

[0145] Based on the propagation path, the strength of each propagation path, and the analysis scope, pre-defined compliance risks and corresponding risk scenarios are identified.

[0146] When the risk scenario is determined to meet the preset violation pattern, the risk level corresponding to the target event is increased, and governance suggestions are generated based on the type of the risk root cause, the path urgency classification result, and the compliance risk.

[0147] In this implementation, after completing path analysis and root cause identification, further compliance assessment and dynamic control responses need to be performed on the data assets associated with the triggering event. This is especially important when the event involves complex compliance risks such as sensitive label propagation, permission overreach aggregation, and cross-domain calls, where traditional static policy libraries are insufficient to meet actual needs. This stage establishes a scalable response system through an intelligent label propagation mechanism, a dynamic policy group matching engine, and an adaptive control tower, covering the entire chain from risk propagation identification to response action implementation.

[0148] The aforementioned high-risk nodes can be understood as a set of nodes selected from the root cause of the risk and the defined scope of analysis. They can include not only the root cause node, but also other data nodes in the upstream and downstream links of the root cause that are affected and have compliance risks.

[0149] First, the analysis scope is defined by tracing the root cause of the risk and combining it with the urgency level of the path. The analysis scope represents the boundary interval of this sensitive data compliance review. Instead of using a full-domain scan, the system dynamically scales the review scope based on the risk source and the urgency of the path risk. High-urgency risk links can be adapted to a larger review scope, fully covering all potential risk propagation links, while low-urgency risk links can be adapted to a partial review scope. This reduces ineffective compliance analysis calculations and improves governance efficiency.

[0150] After defining the scope of analysis, based on the targeted lineage map and combining high-risk nodes and corresponding sensitive tags within the analysis scope, the propagation paths and strengths of sensitive tags in the data chain are statistically obtained. High-risk nodes are asset nodes in the chain that possess significant compliance risks and data risks. These nodes can be aggregation computing nodes carrying multiple types of sensitive data, or data transfer nodes with frequent cross-domain access. Sensitive tags are used to identify the confidentiality level and compliance attributes of data and are transmitted throughout the data flow. The propagation path can be understood as the complete trajectory of sensitive data spreading outward from the root cause of the risk. Propagation strength is used to quantify the level of compliance risk corresponding to a single trajectory. Propagation strength can be determined based on the frequency of direct transmission of sensitive data or the scale of aggregated sensitive data.

[0151] After obtaining complete data on the propagation path and intensity, and combining this with the aforementioned defined analytical scope, the system's pre-defined compliance risks and corresponding risk scenarios are identified. The pre-defined compliance risks are various types of data violation issues pre-defined by the platform, while the risk scenarios are the specific business forms in which compliance risks manifest in actual data flow. Compliance risks arising from the cross-business domain flow of sensitive data can be identified through trajectory analysis, and compliance risks arising from the convergence and superposition of multiple highly sensitive data types at the same node can be identified through intensity assessment.

[0152] After identifying risk scenarios, the identified scenarios can be matched and verified against the platform's preset violation patterns. These preset violation patterns are standardized violation judgment models accumulated over long-term operation and maintenance, used to accurately determine whether compliance risks constitute substantive violations. If a risk scenario match is successful, it indicates that the derivative compliance risks of this event have further escalated. The system can proactively upgrade the risk level corresponding to the target event, raising the overall handling priority and control standards for that event.

[0153] Simultaneously, it can intelligently generate suitable governance recommendations by integrating multiple core elements. The generation of governance recommendations can be based on the type of root cause of risk, the urgency level of the path, and the identified types of compliance risks. The final governance recommendations can be compliance remediation plans for optimizing strategy configurations, or long-term operation and maintenance optimization plans for risk nodes and flow links. This provides operation and maintenance personnel with accurate and implementable root cause governance basis, completely eliminating quality and compliance risks in the data link.

[0154] Emergency circuit breakers primarily aim to quickly contain risks. The data governance method implemented in this application focuses on deeply investigating potential compliance risks of sensitive data hidden in the data chain. Through source tracing, propagation analysis, risk identification, and intelligent suggestion generation, it achieves a closed-loop governance of data risks from emergency response to root cause compliance rectification.

[0155] For example, the specific process can be as follows:

[0156] (1) Detection of sensitive label propagation and identification of aggregation risks:

[0157] Label propagation paths can be simulated in the lineage graph based on high-risk nodes and their sensitive labels. In addition to edge connections, features such as job type and node call frequency can be considered during propagation to dynamically assign propagation strength to the path. Direct field transmission receives higher weight, while indirect transmission via aggregation or related jobs introduces attenuation.

[0158] Since cross-domain data flow and highly sensitive tag aggregation are the two most typical and impactful scenarios of data compliance risk, these two risk propagation patterns can be tracked. For cross-domain tag flow, such as medical data tags flowing into the marketing domain, it will be marked as "inter-domain propagation" and its compliance will be assessed. For highly sensitive tag aggregation scenarios, when multiple sensitive tags (such as PII, PHI) appear together at the same node, the system will determine whether it constitutes excessive permissions or illegal use based on the access role and purpose.

[0159] The system then compares the propagation results with preset high-risk compliance models, such as typical violation scenarios like "PII+PHI aggregated use in different business domains" or "PHI data flowing to uncertified platforms." Once a match is found, the event risk level will be automatically upgraded, and response suggestions will be generated.

[0160] (2) Dynamic aggregation strategy group matching and evaluation:

[0161] After identifying high-risk label propagation paths, the system will automatically match the corresponding compliance policy groups based on context such as label type, region, purpose of use, and access role. Each policy group defines policy items such as de-identification, usage boundaries, storage encryption, and access frequency control around specific regulatory scenarios.

[0162] The matching logic is dynamically generated based on the tag-region-purpose triplet. For example, a node containing PHI and used for data mining will be mapped to the "Health Data Processing" strategy group; a scenario involving PII and used for advertising will trigger the "User Authorization" strategy group.

[0163] (3) Adaptive control response and multi-machine handling strategy:

[0164] Based on the results of tag propagation and strategy evaluation, the system can activate the "Adaptive Control Tower" module to dynamically trigger control responses according to risk levels: for high-risk events, immediate mandatory measures are implemented, such as automatic desensitization, access freezing, and generating compliance reports; for medium-risk events, isolation mechanisms are triggered, including data splitting, differential privacy injection, or access warnings; for low-risk events, the highest priority fallback strategy is adopted by default, and the matter is submitted to compliance personnel for arbitration. All control actions are structured and encapsulated into instruction sets, including action type, target asset, execution method, and recovery conditions, and are then distributed to the corresponding system modules for automatic execution.

[0165] (4) Governance knowledge extraction and continuous strategy evolution:

[0166] After a compliance incident is resolved, the system can store the violation pattern, the path of tag propagation, and the results of strategy execution in the governance knowledge graph, and link it to the historical strategy library to gradually build risk characteristics and response templates for typical scenarios.

[0167] When similar events occur in the future, the system can match and recommend historically effective strategies based on label overlap (Jaccard similarity) and contextual semantic similarity, thereby improving strategy reuse and response efficiency, and dynamically adjusting recommendation priorities based on historical success rates.

[0168] Furthermore, the data governance method of this application also has the following advantages:

[0169] 1. Enhance the real-time performance and coverage of risk identification through a dual-path governance trigger mechanism. Compared to the passive triggering methods of existing governance platforms that mainly rely on rule engines, periodic audits, or manual inspections, a dual-path trigger mechanism with "data quality fluctuations" and "policy change events" as entry points can be constructed. This extends the governance triggering capability to both the data and policy layers, providing stronger real-time response capabilities and scenario adaptability, effectively filling the technical gap that dynamic risk events cannot be incorporated into the governance closed loop.

[0170] 2. Constructing an event-driven, directional lineage graph can effectively reduce analytical redundancy and improve problem localization efficiency. Addressing the issues of high resource consumption and slow response caused by the commonly used static, full-scale lineage graphs in related technologies, this application utilizes an event-centric, target-field-anchored directional graph generation mechanism. Combined with multi-granularity lineage analysis capabilities at the table, field, and indicator levels, it dynamically constructs a minimum necessary path graph. Compared to traditional static modeling methods, this application can generate graph nodes and edges on demand, reducing computational resource consumption and improving tracing accuracy and response efficiency for problem scenarios.

[0171] 3. A response path identification method integrating risk scoring enables precise identification of critical paths and focused governance resources. Unlike existing methods that rely solely on shallow traversal or linear rule-based determination of label propagation paths, this application's implementation method uses a comprehensive scoring model combining graph structure features and node influence factors to systematically evaluate the risk weight, influence intensity, and node role of propagation paths, identifying high-influence paths and key propagation nodes. This mechanism effectively improves the accuracy of strategy configuration and the efficiency of governance action deployment, overcoming the technical bottlenecks of extensive resource allocation and unfocused risk identification in traditional solutions.

[0172] It should be noted that in the data governance method of this application, by constructing a dual-path governance triggering mechanism, an event-driven targeted lineage graph generation method, and a response path identification framework that integrates risk scoring, dynamic identification of sensitive data tag propagation paths, intelligent assessment of policy conflicts, and precise scheduling of governance resources are achieved, thereby improving the real-time performance, focus, and closed-loop capability of compliance governance. Alternatively, this can be implemented through a data governance platform software system, with key modules including: a governance triggering engine (containing data quality and policy change perception mechanisms), a targeted lineage graph construction module (based on field deconstruction and event reverse tracing logic), and a path risk scoring model (integrating factors such as propagation intensity, node roles, and policy matching degree).

[0173] Please see Figure 3 , Figure 3 This is one of the structural schematic diagrams of a data governance device 300 provided in an embodiment of this application. As shown in the figure, the data governance device 300 includes:

[0174] The capture module 310 is used to capture and output target events based on a dual-path governance triggering mechanism. The target events include data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two paths respectively.

[0175] Construction module 320 is used to construct a directional kinship map based on the target event;

[0176] The positioning module 330 is used to locate the root cause of risk based on the directional kinship map and output the path urgency classification result corresponding to each path in the directional kinship map;

[0177] The data governance module 340 is used to perform data governance based on the risk root causes and the path urgency classification results.

[0178] Optionally, the dual-path governance triggering mechanism includes a data quality anomaly triggering path and a policy change sensitive metadata triggering path, and the capture module 310 is specifically used for:

[0179] Based on the data quality anomaly triggering path, and combined with at least one of the acquired data structure changes, business indicator fluctuations, and critical period anomalies, the data quality anomaly event is output.

[0180] Based on the sensitive metadata triggering path for policy changes, and combined with at least one of the obtained sensitive label adjustments and compliance policy configuration changes, the policy change event is output.

[0181] Optionally, the construction module 320 is specifically used for:

[0182] Extract the event object and risk level from the target event;

[0183] The target for kinship analysis is determined based on the event object, and the depth of exploration is adjusted based on the risk level.

[0184] Based on the stated lineage analysis objectives and the stated exploration depth, the importance scores of each node in the upstream and downstream links associated with the event object are obtained;

[0185] Remove upstream and downstream links whose importance scores are lower than a preset threshold, inject event context into the remaining upstream and downstream links, and output a directional lineage graph.

[0186] Optionally, the positioning module 330 is specifically used for:

[0187] The directional kinship graph is transformed into a directed weighted graph, and initial responsibility nodes are marked. The weighted graph contains nodes representing data assets and edges representing data dependencies.

[0188] The responsibility value of each node is calculated iteratively in reverse, and candidate root cause nodes are determined based on the node responsibility values.

[0189] Based on the first condition, candidate root cause nodes are verified, and those that pass the verification are identified as risk root causes.

[0190] Based on node importance, access frequency, and data asset level, obtain the path risk score corresponding to each path in the targeted lineage graph, and classify and output the path urgency classification result corresponding to each path in the targeted lineage graph according to the path risk score.

[0191] The first condition includes at least one of the following:

[0192] There have been structural or strategic changes; there are high-risk operational logics; historical failure modes have been hit; and it is involved in the calculation chain of business metrics.

[0193] Optionally, the data governance module 340 is specifically used for:

[0194] Based on the path urgency classification results, a three-level circuit breaker response is executed for the root cause of the risk and the corresponding link. The circuit breaker response includes:

[0195] If the path urgency classification result is high urgency, risk blocking operations will be performed on the root cause of the risk and the corresponding link.

[0196] If the path urgency classification result is medium urgency, risk isolation operations are performed on the root cause of the risk and the corresponding link.

[0197] If the path urgency classification result is low urgency, a risk warning operation is performed on the root cause of the risk and the corresponding link.

[0198] Optionally, the data governance module 340 is further configured to:

[0199] The analysis scope is defined by taking the risk root cause as the starting point and combining it with the path urgency classification results;

[0200] Based on the directional lineage map and the high-risk nodes and corresponding sensitive tags within the analysis range, the propagation path of the sensitive tags in the link and the strength of each propagation path are obtained;

[0201] Based on the propagation path, the strength of each propagation path, and the analysis scope, pre-defined compliance risks and corresponding risk scenarios are identified.

[0202] When the risk scenario is determined to meet the preset violation pattern, the risk level corresponding to the target event is increased, and governance suggestions are generated based on the type of the risk root cause, the path urgency classification result, and the compliance risk.

[0203] For details, see Figure 4As shown in the figure, this application embodiment also provides one of the electronic devices, including a bus 401, a transceiver 402, an antenna 403, a bus interface 404, a processor 405, and a memory 406.

[0204] Processor 405, used for:

[0205] The dual-path governance triggering mechanism captures and outputs target events, including data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two separate paths.

[0206] Based on the target event, a targeted kinship map is constructed;

[0207] Based on the directional kinship map, the root causes of risk are located, and the path urgency classification results corresponding to each path in the directional kinship map are output.

[0208] Data governance is performed based on the aforementioned root causes of risk and the path urgency classification results.

[0209] exist Figure 4 In this context, a bus architecture (represented by bus 401) is used. Bus 401 can include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 405 and memory represented by memory 406. Bus 401 can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 404 provides an interface between bus 401 and transceiver 402. Transceiver 402 can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 405 is transmitted over a wireless medium via antenna 403, which further receives data and transmits data to processor 405.

[0210] Processor 405 is responsible for managing bus 401 and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Memory 406 can be used to store data used by processor 405 during operation.

[0211] Alternatively, the processor 405 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD).

[0212] Optionally, the dual-path governance triggering mechanism includes a data quality anomaly triggering path and a policy change sensitive metadata triggering path. The processor 405 is specifically used for:

[0213] Based on the data quality anomaly triggering path, and combined with at least one of the acquired data structure changes, business indicator fluctuations, and critical period anomalies, the data quality anomaly event is output.

[0214] Based on the sensitive metadata triggering path for policy changes, and combined with at least one of the obtained sensitive label adjustments and compliance policy configuration changes, the policy change event is output.

[0215] Optionally, the processor 405 is specifically used for:

[0216] Extract the event object and risk level from the target event;

[0217] The target for kinship analysis is determined based on the event object, and the depth of exploration is adjusted based on the risk level.

[0218] Based on the stated lineage analysis objectives and the stated exploration depth, the importance scores of each node in the upstream and downstream links associated with the event object are obtained;

[0219] Remove upstream and downstream links whose importance scores are lower than a preset threshold, inject event context into the remaining upstream and downstream links, and output a directional lineage graph.

[0220] Optionally, the processor 405 is specifically used for:

[0221] The directional kinship graph is transformed into a directed weighted graph, and initial responsibility nodes are marked. The weighted graph contains nodes representing data assets and edges representing data dependencies.

[0222] The responsibility value of each node is calculated iteratively in reverse, and candidate root cause nodes are determined based on the node responsibility values.

[0223] Based on the first condition, candidate root cause nodes are verified, and those that pass the verification are identified as risk root causes.

[0224] Based on node importance, access frequency, and data asset level, obtain the path risk score corresponding to each path in the targeted lineage graph, and classify and output the path urgency classification result corresponding to each path in the targeted lineage graph according to the path risk score.

[0225] The first condition includes at least one of the following:

[0226] There have been structural or strategic changes; there are high-risk operational logics; historical failure modes have been hit; and it is involved in the calculation chain of business metrics.

[0227] Optionally, the processor 405 is specifically used for:

[0228] Based on the path urgency classification results, a three-level circuit breaker response is executed for the root cause of the risk and the corresponding link. The circuit breaker response includes:

[0229] If the path urgency classification result is high urgency, risk blocking operations will be performed on the root cause of the risk and the corresponding link.

[0230] If the path urgency classification result is medium urgency, risk isolation operations are performed on the root cause of the risk and the corresponding link.

[0231] If the path urgency classification result is low urgency, a risk warning operation is performed on the root cause of the risk and the corresponding link.

[0232] Optionally, the processor 405 is specifically used for:

[0233] The analysis scope is defined by taking the risk root cause as the starting point and combining it with the path urgency classification results;

[0234] Based on the directional lineage map and the high-risk nodes and corresponding sensitive tags within the analysis range, the propagation path of the sensitive tags in the link and the strength of each propagation path are obtained;

[0235] Based on the propagation path, the strength of each propagation path, and the analysis scope, pre-defined compliance risks and corresponding risk scenarios are identified.

[0236] When the risk scenario is determined to meet the preset violation pattern, the risk level corresponding to the target event is increased, and governance suggestions are generated based on the type of the risk root cause, the path urgency classification result, and the compliance risk.

[0237] It should be noted that the electronic device provided in this application embodiment is a device capable of executing the above-described data governance method. Therefore, all implementation methods in the above-described data governance method embodiments are applicable to this electronic device and can achieve the same or similar beneficial effects. To avoid repetition, this embodiment will not elaborate further.

[0238] This application also provides an electronic device, including: a processor, a memory, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the various processes of the above-described data governance method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0239] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the above-described data governance method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0240] This application also provides a computer program product, including computer instructions. When executed by a processor, these computer instructions implement the various processes of the above-described data governance method embodiments and achieve the same technical effects. To avoid repetition, they will not be described again here.

[0241] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0242] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0243] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A data governance method, characterized in that, The method includes: The dual-path governance triggering mechanism captures and outputs target events, including data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two separate paths. Based on the target event, a targeted kinship map is constructed; Based on the directional kinship map, the root causes of risk are located, and the path urgency classification results corresponding to each path in the directional kinship map are output. Data governance is performed based on the aforementioned root causes of risk and the path urgency classification results.

2. The method according to claim 1, characterized in that, The dual-path governance triggering mechanism includes a data quality anomaly triggering path and a policy change sensitive metadata triggering path. The capture and output of target events based on the dual-path governance triggering mechanism includes: Based on the data quality anomaly triggering path, and combined with at least one of the acquired data structure changes, business indicator fluctuations, and critical period anomalies, the data quality anomaly event is output. Based on the sensitive metadata triggering path for policy changes, and combined with at least one of the obtained sensitive label adjustments and compliance policy configuration changes, the policy change event is output.

3. The method according to claim 1, characterized in that, The construction of a directional kinship map based on the target event includes: Extract the event object and risk level from the target event; The target for kinship analysis is determined based on the event object, and the depth of exploration is adjusted based on the risk level. Based on the stated lineage analysis objectives and the stated exploration depth, the importance scores of each node in the upstream and downstream links associated with the event object are obtained; Remove upstream and downstream links whose importance scores are lower than a preset threshold, inject event context into the remaining upstream and downstream links, and output a directional lineage graph.

4. The method according to any one of claims 1 to 3, characterized in that, The process of locating root causes of risk based on the directional kinship map and outputting the path urgency classification results for each path in the directional kinship map includes: The directional kinship graph is transformed into a directed weighted graph, and initial responsibility nodes are marked. The weighted graph contains nodes representing data assets and edges representing data dependencies. The responsibility value of each node is calculated iteratively in reverse, and candidate root cause nodes are determined based on the node responsibility values. Based on the first condition, candidate root cause nodes are verified, and those that pass the verification are identified as risk root causes. Based on node importance, access frequency, and data asset level, obtain the path risk score corresponding to each path in the targeted lineage graph, and classify and output the path urgency classification result corresponding to each path in the targeted lineage graph according to the path risk score. The first condition includes at least one of the following: There have been structural or strategic changes; there are high-risk operational logics; historical failure modes have been hit; and it is involved in the calculation chain of business metrics.

5. The method according to any one of claims 1 to 3, characterized in that, The data governance based on the risk root causes and the path urgency classification results includes: Based on the path urgency classification results, a three-level circuit breaker response is executed for the root cause of the risk and the corresponding link. The circuit breaker response includes: If the path urgency classification result is high urgency, risk blocking operations will be performed on the root cause of the risk and the corresponding link. If the path urgency classification result is medium urgency, risk isolation operations are performed on the root cause of the risk and the corresponding link. If the path urgency classification result is low urgency, a risk warning operation is performed on the root cause of the risk and the corresponding link.

6. The method according to claim 5, characterized in that, The data governance based on the risk root causes and the path urgency classification results also includes: The analysis scope is defined by taking the risk root cause as the starting point and combining it with the path urgency classification results; Based on the directional lineage map and the high-risk nodes and corresponding sensitive tags within the analysis range, the propagation path of the sensitive tags in the link and the strength of each propagation path are obtained; Based on the propagation path, the strength of each propagation path, and the analysis scope, pre-defined compliance risks and corresponding risk scenarios are identified. When the risk scenario is determined to meet the preset violation pattern, the risk level corresponding to the target event is increased, and governance suggestions are generated based on the type of the risk root cause, the path urgency classification result, and the compliance risk.

7. A data governance device, characterized in that, include: The capture module is used to capture and output target events based on a dual-path governance triggering mechanism. The target events include data quality anomaly events and policy change events. The dual-path governance triggering mechanism is used to monitor data quality anomaly and policy change behavior through two separate paths. A construction module is used to construct a directional kinship map based on the target event; The localization module is used to locate the root causes of risk based on the directional kinship map and output the path urgency classification results corresponding to each path in the directional kinship map. The data governance module is used to perform data governance based on the risk root causes and the path urgency classification results.

8. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 6.

Images