Multi-agent data governance method and system based on negative feedback optimization

Abstract

The embodiment of the application provides a multi-agent data governance method and system based on negative feedback optimization, which comprises the following steps: performing data governance on to-be-governed data based on a multi-agent system to obtain a data governance result, and creating a task execution path atlas in real time during data governance; performing evaluation processing on the data governance result based on a preset evaluation index set and a dynamic threshold set to obtain negative feedback data; querying the task execution path atlas according to the negative feedback data to obtain a task processing path; performing evaluation processing on the task processing path to obtain one or more to-be-optimized agents; performing matching processing based on the negative feedback data, a task type of the to-be-optimized agent and an optimization parameter matching library to obtain corresponding optimization data; performing optimization on the corresponding to-be-optimized agent according to the optimization data; and performing data governance on the to-be-governed data based on the optimized multi-agent system. The scheme can optimize and iterate the agent based on a negative feedback learning mechanism.

Description

Technical Field

[0001] This application relates to the field of electronic digital data processing, and in particular to a multi-agent data governance method and system based on negative feedback optimization. Background Technology

[0002] With the deepening development of the big data era, data governance has become a core foundation for enterprise digital transformation and intelligent applications. Traditional data governance technologies generally adopt workflow modes driven by single algorithms or manual rules, which are difficult to cope with complex quality defects (such as data redundancy, format errors, semantic conflicts, etc.) in massive heterogeneous data. Especially when facing dynamically changing governance needs, their adaptability and scalability have significant bottlenecks. In recent years, the combination of multi-agent systems and reinforcement learning has provided a new technical path for data governance. Through the collaborative decision-making of distributed agents, the system can process multiple types of data governance tasks in parallel (such as data cleaning, classification, standardization, and quality assessment), while the reinforcement learning mechanism endows the agents with the ability to dynamically optimize strategies in an interactive environment.

[0003] Most related technologies rely on positive reward signals to drive optimization, and lack systematic mining and utilization of negative feedback data generated during the data governance process (such as error handling results and conflict operation records), resulting in relatively low data governance processing efficiency. Summary of the Invention

[0004] This application provides a multi-agent data governance method and system based on negative feedback optimization, which solves the problem that the lack of systematic mining and utilization of negative feedback data generated during the data governance process in the prior art leads to relatively low data governance processing efficiency. It can optimize and iterate the agents based on the negative feedback learning mechanism to perform data governance in a more efficient and better way.

[0005] In a first aspect, embodiments of this application provide a multi-agent data governance method based on negative feedback optimization, including:

[0006] The multi-agent system performs data governance on the data to be governed to obtain the corresponding data governance results, and creates a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0007] The data governance results are evaluated and processed based on a preset set of evaluation indicators and a dynamic set of thresholds to obtain negative feedback data. The dynamic set of thresholds is updated every time the evaluation process is performed.

[0008] Based on the negative feedback data, the task execution path graph is queried to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library, the corresponding optimization data is obtained through matching.

[0009] Based on the optimized data, the corresponding intelligent agents to be optimized are optimized, and the data to be managed is re-managed based on the optimized multi-agent system.

[0010] Optionally, the real-time creation of the task execution path graph corresponding to the data governance process includes:

[0011] The data flow direction and data processing volume of each intelligent agent during the data governance process are recorded in real time, and a dedicated task node for each intelligent agent is generated. Directed edges between the dedicated task nodes are generated based on the data flow direction of each intelligent agent to obtain the execution path graph of the task to be adjusted.

[0012] The task execution path map is obtained by adjusting each dedicated task node in the task execution path map to be adjusted based on the data processing volume.

[0013] Optionally, adjusting each dedicated task node in the task execution path graph to be adjusted based on the data processing volume to obtain the task execution path graph includes:

[0014] Based on the data processing volume of each intelligent agent, an evaluation weight is assigned to the corresponding dedicated task node, and each evaluation weight is marked at a preset position on the corresponding dedicated task node to obtain a task execution path map.

[0015] Optionally, the preset evaluation index set includes completeness indicators, consistency indicators, and timeliness indicators. The step of evaluating the data governance results based on the preset evaluation index set and the dynamic threshold set to obtain negative feedback data includes:

[0016] The data in the data governance results are evaluated for completeness, consistency, and timeliness to obtain the corresponding completeness evaluation value, consistency evaluation value, and timeliness evaluation value.

[0017] The integrity assessment value, the consistency assessment value, and the timeliness assessment value are compared with the corresponding dynamic thresholds in the dynamic threshold set, and the result data with assessment values ​​less than the corresponding dynamic thresholds are determined as negative feedback data.

[0018] Optionally, the negative feedback data includes error data and error types, the task execution path graph includes each directed connection of task nodes and corresponding evaluation weights, and the step of querying the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, and evaluating the task processing path to obtain one or more agents to be optimized, includes:

[0019] Query the task node that outputs the error data in the task execution path graph, and obtain the task processing path by backtracking the task execution path graph based on the task node that outputs the error data.

[0020] Determine the first task node in the task processing path that is associated with the error type, and filter out the second task node in the task processing path whose evaluation weight is greater than a preset weight. Then, determine the first task node and the second task node as nodes to be evaluated.

[0021] One or more intelligent agents to be optimized are obtained by evaluating the task execution logs and resource consumption monitoring information associated with each of the nodes to be evaluated.

[0022] Optionally, the step of obtaining one or more intelligent agents to be optimized based on the task execution logs and resource consumption monitoring information associated with each of the nodes to be evaluated includes:

[0023] Anomaly detection is performed on each record in the task execution log to obtain the log detection results for each node to be evaluated;

[0024] Based on the node task type of each node to be evaluated, the corresponding resource consumption threshold information is determined, and the resource consumption monitoring information is compared with the corresponding resource consumption threshold information to obtain the resource consumption detection result of each node to be evaluated.

[0025] The agents corresponding to the nodes to be evaluated that have abnormal log detection results or resource consumption detection results are identified as agents to be optimized.

[0026] Optionally, the negative feedback data includes the error type of the erroneous data, and the matching process based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data includes:

[0027] The corresponding optimization parameter type in the optimization parameter matching library is determined based on the error type and the task type.

[0028] When there are multiple agents to be optimized, the task types of the multiple agents to be optimized are combined to obtain a task type combination. Based on the task type combination, the corresponding error type influence allocation information is queried to obtain the influence of each task type on the error type.

[0029] Based on the influence degree, the parameter values ​​of the corresponding optimization parameter types of the agent to be optimized are determined, and each of the optimization parameter types and the parameter values ​​are determined as the optimization data of the corresponding agent to be optimized.

[0030] Secondly, embodiments of this application also provide a multi-agent data governance system based on negative feedback optimization, comprising:

[0031] The data governance module is used to perform data governance on the data to be governed based on a multi-agent system to obtain the corresponding data governance results. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0032] The graph creation module is used to create task execution path graphs corresponding to the data governance process in real time.

[0033] An evaluation and processing module is used to evaluate the data governance results based on a preset set of evaluation indicators and a dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation and processing is performed.

[0034] The query module is used to query the task execution path map based on the negative feedback data to obtain the corresponding task processing path;

[0035] The evaluation processing module is also used to evaluate the task processing path to obtain one or more intelligent agents to be optimized.

[0036] The matching processing module is used to perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data.

[0037] The optimization module is used to optimize the corresponding intelligent agent to be optimized based on the optimization data;

[0038] The data governance module is also used to re-govern the data to be governed based on the optimized multi-agent system.

[0039] Thirdly, embodiments of this application also provide a multi-agent data governance device based on negative feedback optimization, the device comprising:

[0040] One or more processors;

[0041] Storage device for storing one or more programs.

[0042] When the one or more programs are executed by the one or more processors, the one or more processors implement the multi-agent data governance method based on negative feedback optimization as described in the embodiments of this application.

[0043] Fourthly, embodiments of this application also provide a storage medium for storing computer-executable instructions, which, when executed by a computer processor, are used to execute the multi-agent data governance method based on negative feedback optimization described in embodiments of this application.

[0044] In this embodiment, a multi-agent system is used to perform data governance on the data to be governed, resulting in corresponding data governance outcomes. A task execution path graph corresponding to the data governance process is created in real time. The multi-agent system includes multiple agents capable of independently executing data governance tasks. The data governance results are evaluated based on a preset set of evaluation metrics and a dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every evaluation. The task execution path graph is queried based on the negative feedback data to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Matching is performed based on the negative feedback data, the task type of the agent to be optimized, and an optimization parameter matching library to obtain corresponding optimized data. The corresponding agent to be optimized is then optimized based on the optimized multi-agent system, and the data governance of the data to be governed is performed again. This solution addresses the problem in existing technologies where the lack of systematic mining and utilization of negative feedback data generated during data governance leads to relatively low data governance efficiency. It enables iterative optimization of agents based on a negative feedback learning mechanism, achieving data governance in a more efficient and superior manner. Attached Figure Description

[0045] Figure 1 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization provided in this application embodiment;

[0046] Figure 2 A flowchart of a multi-agent data governance method based on negative feedback optimization, which includes creating a task execution path graph, is provided for embodiments of this application.

[0047] Figure 3 A schematic diagram of a task execution path map to be adjusted, provided for an embodiment of this application;

[0048] Figure 4 A schematic diagram of a task execution path map provided in an embodiment of this application;

[0049] Figure 5 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization for determining negative feedback data, provided in an embodiment of this application;

[0050] Figure 6 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization for determining the agent to be optimized, provided in an embodiment of this application;

[0051] Figure 7 A flowchart of a multi-agent data governance method based on negative feedback optimization for determining optimization data of an agent to be optimized, provided in an embodiment of this application;

[0052] Figure 8 A block diagram of a multi-agent data governance system based on negative feedback optimization is provided for embodiments of this application.

[0053] Figure 9 This is a schematic diagram of the structure of a multi-agent data governance device based on negative feedback optimization, provided in an embodiment of this application. Detailed Implementation

[0054] The embodiments of this application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of this application and are not intended to limit the scope of the embodiments. Furthermore, it should be noted that, for ease of description, only the parts relevant to the embodiments of this application are shown in the accompanying drawings, not the entire structure.

[0055] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0056] This application provides a multi-agent data governance method based on negative feedback optimization, which can be applied to multi-agent data governance scenarios such as bank credit assessment, intelligent traffic scheduling, and enterprise-level data governance. In this multi-agent data governance method based on negative feedback optimization, the server is the executing entity for each step.

[0057] Figure 1 A flowchart of a multi-agent data governance method based on negative feedback optimization provided in this application embodiment is shown below. Figure 1 As shown, it specifically includes:

[0058] Step S101: Based on the multi-agent system, perform data governance on the data to be governed to obtain the corresponding data governance results, and create a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0059] The multi-agent system (MAS) characterizes a system composed of multiple agents capable of independently performing data governance tasks. These agents can collaborate and coordinate to complete tasks. Using this MAS, data governance can be performed on data to be governed, yielding data governance results. Data to be governed refers to the data currently awaiting governance. During the data governance process, a corresponding task execution path graph is created in real-time. The task execution path graph can be a graph displaying the dependencies between agents during data governance. This graph can consist of nodes and edges. In one embodiment, creating the task execution path graph involves recording the data flow of each agent during the data governance process in real-time, generating dedicated task nodes for each agent, and generating directed edges between these dedicated task nodes based on the data flow of each agent to obtain the task execution path graph. By creating a task execution path graph corresponding to the data governance process, the data governance process can be displayed more intuitively.

[0060] Step S102: Evaluate the data governance results based on the preset evaluation index set and the dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation process is performed.

[0061] The preset evaluation index set can be a pre-set set of various indicators used to evaluate the data governance results. The dynamic threshold set is a set of standard dynamic thresholds corresponding to each evaluation index. This dynamic threshold set is updated every time an evaluation is performed. Negative feedback data can be obtained by evaluating the data governance results using this preset evaluation index set and the dynamic threshold set. Negative feedback data is used to characterize data that does not meet standard conditions, such as erroneous data or error types, identified through the evaluation of the data governance results. In one embodiment, the preset evaluation index set includes multiple evaluation indicators. One method for evaluating data governance results is to evaluate each result data in the data governance results according to the multiple evaluation indicators to obtain multiple corresponding evaluation values. Each evaluation value is then compared with the corresponding dynamic threshold in the dynamic threshold set. Result data with evaluation values ​​less than the corresponding dynamic threshold (a preset number or more) are identified as negative feedback data. By evaluating the data governance results using the evaluation index set and the dynamic threshold set, negative feedback data can be determined from multiple dimensions, improving its rationality and accuracy.

[0062] In one embodiment, a dynamic threshold set update method may involve, after identifying negative feedback data, calculating the average evaluation value of each evaluation indicator for the non-negative feedback data among multiple result data, comparing the average evaluation value of each evaluation indicator with the corresponding historical average evaluation value within a preset time period, and if the difference between the average evaluation value and the corresponding historical average evaluation value is greater than a preset value, calculating the average of the average evaluation value and the corresponding historical average evaluation value, querying the corresponding update threshold based on the average value, and updating the corresponding dynamic threshold. By continuously updating the dynamic threshold, the accuracy of the evaluation process can be improved.

[0063] Step S103: Query the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, evaluate the task processing path to obtain one or more agents to be optimized, and perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data.

[0064] Specifically, querying the task execution path graph through negative feedback data yields the corresponding task processing path. The task processing path refers to the path traversed when executing the sub-task corresponding to the negative feedback data. Evaluating this task processing path yields one or more agents to be optimized. Agents to be optimized can refer to those with defects or problems requiring optimization. The task type refers to the type of task assigned to the agent, such as data acquisition, data cleaning, or data integration. Matching the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library yields the corresponding optimization data. The optimization parameter matching library can be a database storing optimization parameters for different situations; the optimization parameters can be the types of parameters the agent optimizes, such as learning rate and weights. Optimization data refers to the data used by the agent to be optimized during optimization. In one embodiment, the task execution path graph includes various directed connected task nodes. One way to determine the agent to be optimized is to query the task nodes that output negative feedback data in the task execution path graph, perform path backtracking in the task execution path graph based on the task nodes that output negative feedback data to obtain the task processing path, perform error detection processing on the task execution logs of each task node in the task processing path to obtain the detection result, and determine the agent corresponding to the task node with the abnormal detection result as the agent to be optimized.

[0065] In one embodiment, the negative feedback data includes the error type and error severity of the error data. One matching process may be to determine the corresponding matching optimization parameter type in the optimization parameter matching library based on the error type and task type, determine the preset value range into which the error severity falls, query the optimization parameter value of the optimization parameter type associated with the preset value range, and determine the optimization parameter type and optimization parameter value as the optimization data of the agent to be optimized.

[0066] Step S104: Optimize the corresponding agent to be optimized based on the optimization data, and re-process the data to be managed based on the optimized multi-agent system.

[0067] In this process, after obtaining optimized data for the agent to be optimized through matching, this optimized data can be used to optimize the corresponding agent. The optimized multi-agent system then re-processes the data to be governed until no negative feedback data remains in the data governance result. For example, if the optimized data for the agent to be optimized is an adjustment of the sentiment threshold to 0.4, then the current sentiment threshold of the agent to be optimized is adjusted to 0.4, and the optimized multi-agent system re-processes the data to be governed until no negative feedback data remains in the data governance result.

[0068] As described above, by using a multi-agent system to perform data governance on the data to be governed, corresponding data governance results are obtained, and a task execution path graph corresponding to the data governance process is created in real time. The multi-agent system includes multiple agents that can independently execute data governance tasks. The data governance results are evaluated based on a preset set of evaluation indicators and a dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time an evaluation is performed. The task execution path graph is queried based on the negative feedback data to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Matching is performed based on the negative feedback data, the task type of the agent to be optimized, and an optimization parameter matching library to obtain the corresponding optimized data. The corresponding agent to be optimized is optimized based on the optimized data, and the data governance of the data to be governed is performed again based on the optimized multi-agent system. This solution solves the problem of relatively low data governance efficiency in existing technologies due to the lack of systematic mining and utilization of negative feedback data generated during the data governance process. It can optimize and iterate agents based on a negative feedback learning mechanism, performing data governance in a more efficient and superior way.

[0069] Figure 2 A flowchart of a multi-agent data governance method based on negative feedback optimization, including the creation of a task execution path graph, is provided for embodiments of this application. Figure 2 As shown, it specifically includes:

[0070] Step S201: Perform data governance on the data to be governed based on a multi-agent system to obtain the corresponding data governance results. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0071] Step S202: Record the data flow direction and data processing volume of each intelligent agent during the data governance process in real time, and generate the exclusive task node of each intelligent agent. Based on the data flow direction of each intelligent agent, generate the directed edges between the exclusive task nodes to obtain the execution path graph of the task to be adjusted.

[0072] In the data governance process, the data flow and processing volume of each agent are recorded in real time, generating dedicated task nodes for each agent. The data flow can be the direction of the agent's output data; for example, if agent A's output data is used as input data to agent B, then agent A's data flow is to agent B. The processing volume refers to the amount of data processed by the agent during task execution. This volume can be calculated as (basic data volume * data type coefficient) + model inference computation volume + tool interaction data volume + collaborative communication data volume. Using this data flow, directed edges between dedicated task nodes can be generated to obtain the execution path graph of the task to be adjusted. For example, in the data governance process, the agents are agent AG. By recording, the data flow of agent A is to agent B and agent C, agent 2's data flow is to agent D, agent C's data flow is to agent E and agent F, and agent E's data flow is to agent G. Connecting the dedicated task nodes of each agent through their corresponding data flows yields the execution path graph of the task to be adjusted. Figure 3 As shown, Figure 3 This is a schematic diagram of a task execution path graph to be adjusted provided in an embodiment of this application. AG respectively represent the dedicated task nodes of the agent AG, and each edge with an arrow (directed edge) represents the data flow direction of the agent.

[0073] Step S203: Adjust each dedicated task node in the task execution path graph to be adjusted according to the data processing volume to obtain the task execution path graph.

[0074] After generating the task execution path graph to be adjusted, the data processing volume of each agent during the data governance process can be used to adjust the dedicated task nodes in the task execution path graph to obtain the task execution path graph. Optionally, one adjustment method is to assign evaluation weights to the corresponding dedicated task nodes according to the data processing volume of each agent, and mark each evaluation weight at a preset position on the corresponding dedicated task node to obtain the task execution path graph. Here, the evaluation weight is used to characterize the importance of the agent corresponding to the task node requiring evaluation processing. For example, the intelligent agents executing the data governance process are referred to as intelligent agents AG. The data processing volume of intelligent agent AG is 80GB, 50GB, 75GB, 60GB, 40GB, 70GB, and 55GB respectively. The larger the data processing volume, the greater the corresponding evaluation weight. Evaluation weights are assigned to the corresponding dedicated task nodes according to the data processing volume of each intelligent agent. The evaluation weights of the dedicated task nodes of intelligent agent AG are 0.8, 0.5, 0.75, 0.6, 0.4, 0.7, and 0.55 respectively. Each evaluation weight is marked at a preset position on the corresponding dedicated task node to obtain the task execution path map, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of a task execution path graph provided in an embodiment of this application. 0.8 represents the evaluation weight of task node A, 0.5 represents the evaluation weight of task node B, 0.75 represents the evaluation weight of task node C, 0.6 represents the evaluation weight of task node D, 0.4 represents the evaluation weight of task node E, 0.7 represents the evaluation weight of task node F, and 0.55 represents the evaluation weight of task node G.

[0075] In another embodiment, one adjustment method is to determine the evaluation level of the corresponding dedicated task node based on the data processing volume of each intelligent agent, and adjust the color of the dedicated task node to the associated color of the corresponding evaluation level to obtain the task execution path map.

[0076] Step S204: Evaluate the data governance results based on the preset evaluation index set and the dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation process is performed.

[0077] Step S205: Query the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, evaluate the task processing path to obtain one or more agents to be optimized, and perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data.

[0078] Step S206: Optimize the corresponding agent to be optimized based on the optimization data, and re-process the data to be managed based on the optimized multi-agent system.

[0079] As described above, by recording the data flow and processing volume of each agent during the data governance process in real time, and generating dedicated task nodes for each agent, a directed edge graph of the task execution path to be adjusted is obtained by generating directed edges between these dedicated task nodes based on the data flow of each agent. The task execution path graph is then adjusted based on the processing volume of each dedicated task node in the graph. This solution generates the task execution path graph by recording the data flow and processing volume of each agent during the data governance process, which more intuitively displays the data governance process and the evaluation importance of each agent.

[0080] Figure 5 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization for determining negative feedback data, as provided in this application embodiment, is shown below. Figure 5 As shown, it specifically includes:

[0081] Step S301: Based on the multi-agent system, perform data governance on the data to be governed to obtain the corresponding data governance results, and create a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0082] Step S302: Perform integrity assessment, consistency assessment and timeliness assessment on each result data in the data governance results to obtain the corresponding integrity assessment value, consistency assessment value and timeliness assessment value.

[0083] The pre-defined set of evaluation indicators includes completeness indicators, consistency indicators, and timeliness indicators. These indicators allow for the evaluation of each data point in the data governance results, yielding corresponding completeness, consistency, and timeliness assessment values. The completeness assessment value characterizes the degree of completeness of the data, and its evaluation method can be (actual data volume / expected data volume) * 100%. The consistency assessment value reflects the degree of agreement between the actual data and the expected results. The timeliness assessment value reflects the proximity between the actual data generation time and the expected time. For example, if the data points in the data governance results are data point 1, data point 2, and data point 3, performing completeness, consistency, and timeliness assessments on data point 1, data point 2, and data point 3 respectively yields the following results: Data point 1 has a completeness assessment value of 98, a consistency assessment value of 96, and a timeliness assessment value of 100; data point 2 has a completeness assessment value of 99, a consistency assessment value of 97, and a timeliness assessment value of 95; and data point 3 has a completeness assessment value of 100, a consistency assessment value of 100, and a timeliness assessment value of 75.

[0084] Step S303: Compare the integrity assessment value, consistency assessment value, and timeliness assessment value with the corresponding dynamic thresholds in the dynamic threshold set. The result data with assessment values ​​less than the corresponding dynamic thresholds are determined as negative feedback data. The dynamic threshold set is updated every time an assessment is performed.

[0085] The dynamic threshold set contains standard values ​​corresponding to the integrity assessment value, consistency assessment value, and timeliness assessment value, respectively. Negative feedback data can be determined by comparing each dynamic threshold with the integrity assessment value, consistency assessment value, and timeliness assessment value of each result data. For example, if the integrity threshold is 95, the consistency threshold is 95, and the timeliness threshold is 90, and the data governance results show that result data 1 has integrity assessment values ​​of 98, consistency assessment value of 96, and timeliness assessment value of 100; result data 2 has integrity assessment values ​​of 99, consistency assessment value of 97, and timeliness assessment value of 95; and result data 3 has integrity assessment values ​​of 100, consistency assessment value of 100, and timeliness assessment value of 75, then result data 3 is determined to be negative feedback data because its timeliness assessment value is less than the timeliness threshold.

[0086] Step S304: Query the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, evaluate the task processing path to obtain one or more agents to be optimized, and perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data.

[0087] Step S305: Optimize the corresponding agent to be optimized based on the optimization data, and re-process the data to be managed based on the optimized multi-agent system.

[0088] As described above, by conducting integrity, consistency, and timeliness assessments on each data point in the data governance results, corresponding integrity, consistency, and timeliness assessment values ​​are obtained. These values ​​are then compared with the corresponding dynamic thresholds in the dynamic threshold set. Data points with assessment values ​​lower than the corresponding dynamic thresholds are identified as negative feedback data. This solution evaluates the data governance results using multiple assessment indicators and corresponding dynamic thresholds, allowing for the identification of negative feedback data from multiple dimensions, thus improving its rationality and accuracy.

[0089] Figure 6 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization for determining the agent to be optimized, as provided in this application embodiment, is shown below. Figure 6As shown, it specifically includes:

[0090] Step S401: Based on the multi-agent system, perform data governance on the data to be governed to obtain the corresponding data governance results, and create a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0091] Step S402: Evaluate the data governance results based on the preset evaluation index set and the dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation process is performed.

[0092] Step S403: Query the task nodes that output error data in the task execution path graph, and perform path backtracking in the task execution path graph based on the task nodes that output error data to obtain the task processing path.

[0093] The negative feedback data includes erroneous data and error types. Erroneous data refers to the erroneous data in the data governance results, and error types refer to the error classification of the erroneous data, such as integrity errors, timeliness errors, etc. The task execution path graph includes each directed connection of task nodes and their corresponding evaluation weights. The evaluation weights can represent the importance of the agent corresponding to the task node in terms of evaluation. Querying the task execution path graph reveals the task node that outputs the erroneous data. By tracing the path back through the task execution path graph from the task node that outputs the erroneous data, the task processing path for that erroneous data can be obtained. For example, the task node that outputs the erroneous data is queried as task node G in the task execution path graph, as shown below. Figure 4 As shown, the task processing path is obtained by backtracking the task node G in the task execution path graph as A→C→E→G.

[0094] Step S404: Determine the first task node in the task processing path that is associated with the task type and error type, and filter out the second task node in the task processing path whose evaluation weight is greater than the preset weight. Determine the first task node and the second task node as nodes to be evaluated, and perform evaluation processing based on the task execution logs and resource consumption monitoring information associated with each node to be evaluated to obtain one or more intelligent agents to be optimized.

[0095] The node task type is used to characterize the classification of tasks performed by the agent corresponding to the task node. The nodes to be evaluated can be determined using the node task type and evaluation weight of each task node. Evaluation processing can be performed using the task execution logs and resource consumption monitoring information associated with each node to be evaluated to obtain one or more agents to be optimized. The task execution log refers to a detailed record file of the entire process of an agent performing data governance tasks. Resource consumption monitoring information refers to real-time monitoring and recording data of the usage of various computing resources during the execution of tasks by the agent, such as CPU utilization and memory consumption. Optionally, one method for determining agents to be optimized is to perform anomaly detection on various records in the task execution logs to obtain log detection results for each node to be evaluated; determine the corresponding resource consumption threshold information based on the node task type of each node to be evaluated; compare the resource consumption monitoring information with the corresponding resource consumption threshold information to obtain the resource consumption detection results for each node to be evaluated; and identify the agents corresponding to nodes with abnormal log detection results or resource consumption detection results as agents to be optimized. For example, the error type of the erroneous data is a consistency error type, the node task type associated with the consistency error type is data cleaning, the task node performing the data cleaning task is task node D, the task processing path is A→B→C→D→E, and the corresponding evaluation weights are 0.8, 0.75, 0.4, 0.55, and 0.3 respectively. The preset weight is 0.6. The task nodes corresponding to the evaluation weights greater than the preset weight are task node A and task node B. Therefore, task nodes A, B, and D are the nodes to be evaluated. The node task type of task node A is data cleaning, and the corresponding resource consumption is... The threshold information is CPU utilization threshold of 50% and memory consumption threshold of 40%. Anomaly detection of all records of task node A yields no anomalies. The resource consumption monitoring information of task node A is CPU utilization of 60% and memory consumption of 45%. Both CPU utilization and memory consumption of task node A are greater than the corresponding thresholds, that is, the resource consumption detection result is abnormal. The agent corresponding to task node A is identified as the agent to be optimized. The same method is used to detect the log detection results and resource consumption detection results of task node B and task node D, and no anomalies are found.

[0096] In another embodiment, a method for determining an agent to be optimized may be to perform anomaly detection on various record information in the task execution log to obtain the log detection results of each node to be evaluated, compare the resource consumption monitoring information of each node to be evaluated with the preset resource consumption threshold information to obtain the resource consumption detection results of each node to be evaluated, and determine the agent corresponding to the node to be evaluated that has an abnormal log detection result or resource consumption detection result as the agent to be optimized.

[0097] Step S405: Based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library, the corresponding optimization data is obtained through matching processing. The corresponding agent to be optimized is optimized according to the optimization data, and the data to be managed is re-managed based on the optimized multi-agent system.

[0098] As described above, the process involves querying the task execution path graph for task nodes that output error data. Based on these error-producing nodes, a backtracking process is performed within the graph to obtain the task processing path. The first task node in this path, associated with both the task type and the error type, is identified. A second task node with an evaluation weight greater than a preset weight is then selected. Both the first and second task nodes are designated as nodes to be evaluated. Evaluation is then performed based on the task execution logs and resource consumption monitoring information associated with each node, resulting in one or more agents to be optimized. This approach improves the rationality and efficiency of identifying agents to be optimized by querying the task execution path graph for error data, determining the task nodes to be evaluated within those paths, and then monitoring their logs and resource consumption.

[0099] Figure 7 A flowchart illustrating a multi-agent data governance method based on negative feedback optimization for determining optimization data of an agent to be optimized, as provided in this application embodiment, is shown below. Figure 7 As shown, it specifically includes:

[0100] Step S501: Based on the multi-agent system, perform data governance on the data to be governed to obtain the corresponding data governance results, and create a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0101] Step S502: Evaluate the data governance results based on the preset evaluation index set and the dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation process is performed.

[0102] Step S503: Query the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, evaluate the task processing path to obtain one or more agents to be optimized, and determine the corresponding optimization parameter type in the optimization parameter matching library based on the error type and task type.

[0103] The negative feedback data includes the error type of the erroneous data. By matching this error type with the task type of the agent to be optimized, the corresponding matching optimization parameter type in the parameter matching library can be determined. For example, if the error type is a consistency error and the task type of the agent to be optimized is data cleaning, querying the parameter matching library for matching optimization parameter types that are both consistency error and data cleaning will return the learning rate from the model training parameters.

[0104] Step S504: When there are multiple agents to be optimized, the task types of the multiple agents to be optimized are combined to obtain a task type combination. Based on the task type combination, the corresponding error type influence distribution information is queried to obtain the influence of each task type on the error type. Based on the influence, the parameter values ​​of the corresponding optimization parameter types of the agents to be optimized are determined. Each optimization parameter type and parameter value is determined as the optimization data of the corresponding agents to be optimized.

[0105] In cases where multiple agents exist to be optimized, their task types are combined to obtain a task type combination. By querying the corresponding error type influence allocation information through this task type combination, the influence of each task type on the error type can be determined. This error type influence allocation information can be a pre-defined distribution of the influence of each task type on the error type within the task type combination. Using this influence, the parameter values ​​for the optimization parameter types of each agent to be optimized can be determined. These optimization parameter types and their values ​​constitute the optimization data for the corresponding agent. For example, the task type of agent 1 to be optimized is data cleaning, and the optimization parameter type is learning rate. The task type of agent 2 to be optimized is data integration, and the optimization parameter type is online learning frequency. Then the task type combination is (data cleaning, data integration), the error type is consistency error, and the corresponding consistency impact distribution information is found to be 40% for data cleaning and 60% for data integration. When the impact of data cleaning on consistency is 40%, the corresponding pre-set learning rate value is 0.01. When the impact of data integration on consistency is 60%, the corresponding pre-set online learning frequency value is to update in batches after accumulating 500 samples. The learning rate of 0.01 is determined as the optimized data for agent 1 to be optimized, and the online learning frequency of updating in batches after accumulating 500 samples is determined as the optimized data for agent 2 to be optimized.

[0106] Step S505: Optimize the corresponding agent to be optimized based on the optimization data, and re-process the data to be managed based on the optimized multi-agent system.

[0107] As described above, by determining the corresponding matching optimization parameter types in the optimization parameter matching library based on the error type and task type, when there are multiple agents to be optimized, the task types of multiple agents to be optimized are combined to obtain a task type combination. Based on the task type combination, the corresponding error type influence distribution information is queried to obtain the influence of each task type on the error type. The parameter values ​​of the corresponding optimization parameter types for the agents to be optimized are determined based on the influence. Each optimization parameter type and its value is then used as the optimization data for the corresponding agent. This scheme determines the corresponding matching optimization parameters in the database based on the error type and the agent's task type, and determines the values ​​of the optimization parameters based on the influence of each task type on the error type, thus ensuring the rationality and accuracy of the optimization data.

[0108] Figure 8 This application provides a module structure block diagram of a multi-agent data governance system based on negative feedback optimization. This system is used to execute the multi-agent data governance method based on negative feedback optimization provided in the above embodiments, and possesses corresponding functional modules and beneficial effects for executing the method. Figure 8 As shown, the system specifically includes:

[0109] The data governance module 101 is used to perform data governance on the data to be governed based on a multi-agent system to obtain the corresponding data governance results. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0110] The graph creation module 102 is used to create a task execution path graph corresponding to the data governance process in real time.

[0111] Evaluation and processing module 103 is used to evaluate and process the data governance results based on a preset set of evaluation indicators and a set of dynamic thresholds to obtain negative feedback data. The set of dynamic thresholds is updated every time the evaluation and processing is performed.

[0112] The query module 104 is used to query the task execution path map based on the negative feedback data to obtain the corresponding task processing path;

[0113] The evaluation and processing module 103 is further configured to evaluate the task processing path to obtain one or more intelligent agents to be optimized.

[0114] Matching processing module 105 is used to perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain corresponding optimization data;

[0115] Optimization module 106 is used to optimize the corresponding intelligent agent to be optimized based on the optimization data;

[0116] The data governance module 101 is also used to re-govern the data to be governed based on the optimized multi-agent system.

[0117] As described above, this scheme utilizes a multi-agent system to perform data governance on the data to be governed, obtaining corresponding data governance results and creating a task execution path graph corresponding to the data governance process in real time. The multi-agent system comprises multiple agents capable of independently executing data governance tasks. Based on a preset set of evaluation indicators and a dynamic threshold set, the data governance results are evaluated to obtain negative feedback data. The dynamic threshold set is updated every time an evaluation is performed. The task execution path graph is queried based on the negative feedback data to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Matching is performed based on the negative feedback data, the task type of the agent to be optimized, and an optimization parameter matching library to obtain corresponding optimization data. The corresponding agent to be optimized is then optimized based on the optimization data, and the data governance of the data to be governed is performed again based on the optimized multi-agent system. This scheme solves the problem of relatively low data governance efficiency in existing technologies due to the lack of systematic mining and utilization of negative feedback data generated during the data governance process. It enables iterative optimization of agents based on a negative feedback learning mechanism, achieving data governance in a more efficient and superior manner.

[0118] In one possible embodiment, the map creation module 102 is specifically used for:

[0119] The data flow direction and data processing volume of each intelligent agent during the data governance process are recorded in real time, and a dedicated task node for each intelligent agent is generated. Directed edges between the dedicated task nodes are generated based on the data flow direction of each intelligent agent to obtain the execution path graph of the task to be adjusted.

[0120] The task execution path map is obtained by adjusting each dedicated task node in the task execution path map to be adjusted based on the data processing volume.

[0121] In one possible embodiment, the map creation module 102 is further configured to:

[0122] Based on the data processing volume of each intelligent agent, an evaluation weight is assigned to the corresponding dedicated task node, and each evaluation weight is marked at a preset position on the corresponding dedicated task node to obtain a task execution path map.

[0123] In one possible embodiment, the evaluation processing module 103 is specifically used for:

[0124] The data in the data governance results are evaluated for completeness, consistency, and timeliness to obtain the corresponding completeness evaluation value, consistency evaluation value, and timeliness evaluation value.

[0125] The integrity assessment value, the consistency assessment value, and the timeliness assessment value are compared with the corresponding dynamic thresholds in the dynamic threshold set, and the result data with assessment values ​​less than the corresponding dynamic thresholds are determined as negative feedback data.

[0126] In one possible embodiment, the evaluation processing module 103 is further configured to:

[0127] Query the task node that outputs the error data in the task execution path graph, and obtain the task processing path by backtracking the task execution path graph based on the task node that outputs the error data.

[0128] Determine the first task node in the task processing path that is associated with the error type, and filter out the second task node in the task processing path whose evaluation weight is greater than a preset weight. Then, determine the first task node and the second task node as nodes to be evaluated.

[0129] One or more intelligent agents to be optimized are obtained by evaluating the task execution logs and resource consumption monitoring information associated with each of the nodes to be evaluated.

[0130] In one possible embodiment, the evaluation processing module 103 is further configured to:

[0131] Anomaly detection is performed on each record in the task execution log to obtain the log detection results for each node to be evaluated;

[0132] Based on the node task type of each node to be evaluated, the corresponding resource consumption threshold information is determined, and the resource consumption monitoring information is compared with the corresponding resource consumption threshold information to obtain the resource consumption detection result of each node to be evaluated.

[0133] The agents corresponding to the nodes to be evaluated that have abnormal log detection results or resource consumption detection results are identified as agents to be optimized.

[0134] In one possible embodiment, the matching processing module 105 is specifically used for:

[0135] The corresponding optimization parameter type in the optimization parameter matching library is determined based on the error type and the task type.

[0136] When there are multiple agents to be optimized, the task types of the multiple agents to be optimized are combined to obtain a task type combination. Based on the task type combination, the corresponding error type influence allocation information is queried to obtain the influence of each task type on the error type.

[0137] Based on the influence level, determine the parameter values ​​for the corresponding optimization parameter types of the agent to be optimized, and then define each optimization parameter type and its value as the optimization data for the corresponding agent to be optimized.

[0138] Figure 9 A schematic diagram of a multi-agent data governance device based on negative feedback optimization provided in this application embodiment is shown below. Figure 9 As shown, the device includes a processor 201, a memory 202, an input device 203, and an output device 204; the number of processors 201 in the device can be one or more. Figure 9 Taking a processor 201 as an example; the processor 201, memory 202, input device 203, and output device 204 in the device can be connected via a bus or other means. Figure 9 Taking a bus connection as an example, the memory 202, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions or modules corresponding to the multi-agent data governance method based on negative feedback optimization in this embodiment of the application. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 202, thereby realizing the aforementioned multi-agent data governance method based on negative feedback optimization. The input device 203 can be used to receive input digital or character information and generate key signal inputs related to user settings and function control of the device. The output device 204 may include a display screen or other display device.

[0139] This application also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a multi-agent data governance method based on negative feedback optimization, the method comprising:

[0140] The multi-agent system performs data governance on the data to be governed to obtain the corresponding data governance results, and creates a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks.

[0141] The data governance results are evaluated and processed based on a preset set of evaluation indicators and a dynamic set of thresholds to obtain negative feedback data. The dynamic set of thresholds is updated every time the evaluation process is performed.

[0142] Based on the negative feedback data, the task execution path graph is queried to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library, the corresponding optimization data is obtained through matching.

[0143] Based on the optimized data, the corresponding intelligent agents to be optimized are optimized, and the data to be managed is re-managed based on the optimized multi-agent system.

[0144] It is worth noting that in the embodiments of the multi-agent data governance method system based on negative feedback optimization described above, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this application.

[0145] Note that the above are merely preferred embodiments and the technical principles applied in this application. Those skilled in the art will understand that the embodiments of this application are not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the embodiments of this application. Therefore, although the embodiments of this application have been described in detail through the above embodiments, the embodiments of this application are not limited to the above embodiments. More other equivalent embodiments may be included without departing from the concept of the embodiments of this application, and the scope of the embodiments of this application is determined by the scope of the appended claims.

Claims

1. A multi-agent data governance method based on negative feedback optimization, characterized in that, The method includes: The multi-agent system performs data governance on the data to be governed to obtain the corresponding data governance results, and creates a task execution path graph corresponding to the data governance process in real time. The multi-agent system includes multiple agents that can independently perform data governance tasks. The data governance results are evaluated and processed based on a preset set of evaluation indicators and a dynamic set of thresholds to obtain negative feedback data. The dynamic set of thresholds is updated every time the evaluation process is performed. Based on the negative feedback data, the task execution path graph is queried to obtain the corresponding task processing path. The task processing path is evaluated to obtain one or more agents to be optimized. Based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library, the corresponding optimization data is obtained through matching. Based on the optimized data, the corresponding intelligent agents to be optimized are optimized, and the data to be managed is re-managed based on the optimized multi-agent system.

2. The multi-agent data governance method based on negative feedback optimization according to claim 1, characterized in that, The real-time creation of the task execution path graph corresponding to the data governance process includes: The data flow direction and data processing volume of each intelligent agent during the data governance process are recorded in real time, and a dedicated task node for each intelligent agent is generated. Directed edges between the dedicated task nodes are generated based on the data flow direction of each intelligent agent to obtain the execution path graph of the task to be adjusted. The task execution path map is obtained by adjusting each dedicated task node in the task execution path map to be adjusted based on the data processing volume.

3. The multi-agent data governance method based on negative feedback optimization according to claim 2, characterized in that, The step of adjusting each dedicated task node in the task execution path graph to be adjusted based on the data processing volume to obtain the task execution path graph includes: Based on the data processing volume of each intelligent agent, an evaluation weight is assigned to the corresponding dedicated task node, and each evaluation weight is marked at a preset position on the corresponding dedicated task node to obtain a task execution path map.

4. The multi-agent data governance method based on negative feedback optimization according to any one of claims 1-3, characterized in that, The preset evaluation index set includes completeness indicators, consistency indicators, and timeliness indicators. The negative feedback data obtained by evaluating the data governance results based on the preset evaluation index set and the dynamic threshold set includes: The data in the data governance results are evaluated for completeness, consistency, and timeliness to obtain the corresponding completeness evaluation value, consistency evaluation value, and timeliness evaluation value. The integrity assessment value, the consistency assessment value, and the timeliness assessment value are compared with the corresponding dynamic thresholds in the dynamic threshold set, and the result data with assessment values ​​less than the corresponding dynamic thresholds are determined as negative feedback data.

5. The multi-agent data governance method based on negative feedback optimization according to any one of claims 1-3, characterized in that, The negative feedback data includes error data and error types. The task execution path graph includes each directed connection of task nodes and their corresponding evaluation weights. The step of querying the task execution path graph based on the negative feedback data to obtain the corresponding task processing path, and evaluating the task processing path to obtain one or more agents to be optimized, including: Query the task node that outputs the error data in the task execution path graph, and obtain the task processing path by backtracking the task execution path graph based on the task node that outputs the error data. Determine the first task node in the task processing path that is associated with the error type, and filter out the second task node in the task processing path whose evaluation weight is greater than a preset weight. Then, determine the first task node and the second task node as nodes to be evaluated. One or more intelligent agents to be optimized are obtained by evaluating the task execution logs and resource consumption monitoring information associated with each of the nodes to be evaluated.

6. The multi-agent data governance method based on negative feedback optimization according to claim 5, characterized in that, The process of evaluating and processing the task execution logs and resource consumption monitoring information associated with each of the nodes to be evaluated to obtain one or more agents to be optimized includes: Anomaly detection is performed on each record in the task execution log to obtain the log detection results for each node to be evaluated; Based on the node task type of each node to be evaluated, the corresponding resource consumption threshold information is determined, and the resource consumption monitoring information is compared with the corresponding resource consumption threshold information to obtain the resource consumption detection result of each node to be evaluated. The agents corresponding to the nodes to be evaluated that have abnormal log detection results or resource consumption detection results are identified as agents to be optimized.

7. The multi-agent data governance method based on negative feedback optimization according to any one of claims 1-3, characterized in that, The negative feedback data includes the error type of the erroneous data. The corresponding optimization data is obtained by matching the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library, including: The corresponding optimization parameter type in the optimization parameter matching library is determined based on the error type and the task type. When there are multiple agents to be optimized, the task types of the multiple agents to be optimized are combined to obtain a task type combination. Based on the task type combination, the corresponding error type influence allocation information is queried to obtain the influence of each task type on the error type. Based on the influence degree, the parameter values ​​of the corresponding optimization parameter types of the agent to be optimized are determined, and each of the optimization parameter types and the parameter values ​​are determined as the optimization data of the corresponding agent to be optimized.

8. A multi-agent data governance system based on negative feedback optimization, characterized in that, include: The data governance module is used to perform data governance on the data to be governed based on a multi-agent system to obtain the corresponding data governance results. The multi-agent system includes multiple agents that can independently perform data governance tasks. The graph creation module is used to create task execution path graphs corresponding to the data governance process in real time. An evaluation and processing module is used to evaluate the data governance results based on a preset set of evaluation indicators and a dynamic threshold set to obtain negative feedback data. The dynamic threshold set is updated every time the evaluation and processing is performed. The query module is used to query the task execution path map based on the negative feedback data to obtain the corresponding task processing path; The evaluation processing module is also used to evaluate the task processing path to obtain one or more intelligent agents to be optimized. The matching processing module is used to perform matching processing based on the negative feedback data, the task type of the agent to be optimized, and the optimization parameter matching library to obtain the corresponding optimization data. The optimization module is used to optimize the corresponding intelligent agent to be optimized based on the optimization data; The data governance module is also used to re-govern the data to be governed based on the optimized multi-agent system.

9. A multi-agent data governance device based on negative feedback optimization, the device comprising: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the multi-agent data governance method based on negative feedback optimization as described in any one of claims 1-7.

10. A storage medium storing computer-executable instructions, which, when executed by a computer processor, are used to perform the multi-agent data governance method based on negative feedback optimization as described in any one of claims 1-7.

Images