Digital economic information management method and system

Abstract

This invention relates to the field of information management technology and discloses a digital economy information management method and system. The system includes: collecting raw digital economy information from multiple heterogeneous data sources, synchronizing and identifying the collected data in time to generate a raw dataset, and performing intelligent preprocessing on the raw dataset, including data cleaning, format standardization, outlier detection, and multi-source data fusion to generate a preprocessed dataset. By constructing a closed-loop management link from raw dataset collection, preprocessed dataset generation, analysis result output, decision data formation to execution feedback data acquisition, the problem of feedback lag and disconnection in traditional one-way processes is solved. This ensures that the effects of management actions are fed back to the upstream analysis and decision-making links in real time, thereby ensuring the continuity and real-time nature of the entire digital economy information management process and improving the overall response speed and collaborative efficiency of management.

Description

Technical Field

[0001] This invention relates to the field of information management technology, specifically to a digital economy information management method and system. Background Technology

[0002] Digital economy information management refers to the complete process and practice system of systematically collecting, intelligently processing, deeply analyzing, and value-creating multi-source, heterogeneous, and dynamic data resources generated in digital economy activities by using information technology and management methods.

[0003] Currently, in the practice of digital economy information management, the collection, processing, analysis, decision-making and execution of multi-source heterogeneous data are usually handled in stages by independent and loosely coupled system modules. There is a lack of efficient real-time feedback and adaptive adjustment mechanisms between modules, resulting in lag, disconnect and static problems in the value transformation chain from data to knowledge to action.

[0004] Therefore, a digital economy information management method and system are proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a digital economy information management method and system, which solves the problems mentioned in the background technology, namely the lack of efficient real-time feedback and adaptive adjustment mechanisms, which leads to lag, disconnection and static nature in the value transformation chain from data to knowledge to action.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a digital economy information management method and system, wherein the method includes the following steps: S1. Collect raw digital economy information from multiple heterogeneous data sources, and synchronize and identify the source of the collected data to generate the raw dataset. S2. Perform intelligent preprocessing on the original dataset. The intelligent preprocessing includes data cleaning, format standardization, outlier detection, and multi-source data fusion to generate a preprocessed dataset. S3. Based on the preprocessed dataset, perform in-depth analysis using an adaptive artificial intelligence analysis model to generate analysis results; S4. Based on the analysis results, a decision-making scheme is generated by combining business rules and historical information, and formatted into executable instructions to generate decision data; S5. Based on the decision data, the management actions are implemented by calling the interface and sending instructions through the automated execution engine, and the execution status is collected synchronously to generate execution feedback data. S6. Based on the execution feedback data and the analysis results, the parameters of the adaptive artificial intelligence analysis model are optimized and the knowledge base is updated to generate an optimized analysis model. S7. Based on the preprocessed dataset, a new round of in-depth analysis is performed using the optimized analysis model, and the business rule base and historical information are dynamically calibrated and incrementally stored in combination with the latest execution feedback data. S8. Based on the updated analysis model, business rule base, and historical information database, the method is driven into continuous closed-loop iteration and spiral evolution.

[0007] Preferably, generating the original dataset in S1 includes the following steps: S11. Obtain multi-source heterogeneous raw data streams in real time from IoT devices, enterprise business systems, public data platforms and mobile applications through the configured interface; S12. The acquired raw data stream is timestamped and identified by the data source to generate a raw dataset with a unified spatiotemporal reference.

[0008] Preferably, generating the preprocessed dataset in step S2 includes the following steps: S21. Perform data cleaning on the original dataset to remove duplicate records, fill in missing values, and correct format errors; S22. Standardize the format of the cleaned data and convert it into a unified data model predefined by the system. S23. Perform anomaly detection and processing on standardized data based on statistical models and machine learning models; S24. Apply a data fusion algorithm to associate and integrate the multi-source data after cleaning, standardization and anomaly processing to generate the preprocessed dataset.

[0009] Preferably, the data fusion algorithm in S24 adopts a fusion method based on Kalman filtering, including the following steps: S241. Timestamp alignment and measurement unit unification are performed on the same indicator data from different data sources; S242. Assign a weight to each data source based on its historical data accuracy, calculated dynamically. S243. Use the Kalman filter algorithm to perform optimal estimation of multi-source data and generate fused high-confidence data values. S244. Integrate the fused high-confidence data values ​​into the preprocessed dataset.

[0010] Preferably, generating decision data in step S4 includes the following steps: S41. Input the analysis results into the rule reasoning engine and perform matching and reasoning with the predefined rules in the business rule base; S42. Call the cases and statistical data in the historical information database to conduct a feasibility assessment and effect simulation of the preliminary decision-making scheme derived from rule reasoning; S43. The final decision scheme determined after evaluation and simulation is encoded into an instruction format that the automated execution engine can recognize and execute, thereby generating the decision data.

[0011] Preferably, generating the optimized analysis model in step S6 includes the following steps: S61. Compare the analysis results with the corresponding execution feedback data, and calculate at least one quantitative indicator among model prediction accuracy, decision execution success rate, resource utilization efficiency, and target achievement. S62. Based on the aforementioned indicators, the internal parameters of the adaptive artificial intelligence analysis model are adjusted using an optimization algorithm; S63. Extract new knowledge patterns and rules from the execution feedback data and update the knowledge base.

[0012] Preferably, step S62 involves using an optimization algorithm to adjust the internal parameters of the adaptive artificial intelligence analysis model, including the following steps: S621. Using the quantitative index as a loss function, calculate the error between the output of the adaptive artificial intelligence analysis model and the execution feedback data; S622. Calculate the gradient of the loss function with respect to the parameters of each layer of the model using the backpropagation algorithm; S623. Apply the gradient descent algorithm to iteratively update the model parameters based on the calculated gradient.

[0013] Preferably, the dynamic calibration and incremental storage in step S7 includes the following steps: S71. Using the preprocessed dataset and the execution feedback data as input, perform a new round of analysis using the optimized analysis model; S72. Based on the actual business effect reflected by the latest execution feedback data, adjust the triggering conditions, weights, and logical relationships of the relevant rules in the business rule base; S73. The latest execution feedback data, related analysis process and decision records are stored as new entries in the historical information database.

[0014] Preferably, adjusting the business rule base according to the actual business results in step S72 includes the following steps: S721. When the success rate of decision execution is lower than the preset threshold, the triggering conditions of the corresponding rule are automatically relaxed. S722. When resource utilization efficiency continuously deviates from the optimal range, adjust the weight parameters of the relevant resource allocation rules. S723. Record the version, time, and reason for each rule adjustment to form a rule change log.

[0015] Preferably, the system includes: The data acquisition and synchronization module obtains raw information from multiple heterogeneous data sources, aligns the data time stamps through the time synchronization unit, marks the data source through the source identification processing unit, and generates the raw dataset. The intelligent preprocessing module receives the original dataset, processes the data through a data cleaning unit and a format standardization unit, and integrates the data through an outlier detection unit and a multi-source data fusion unit to generate a preprocessed dataset. The adaptive analysis module receives the preprocessed dataset, extracts data features through the feature engineering unit, calls and trains an artificial intelligence analysis model using the adaptive model scheduling unit, and generates analysis results. The decision generation module receives the analysis results, generates a preliminary decision scheme by combining the rule reasoning unit with the business rule base, and verifies and optimizes it through the multi-objective optimization and simulation evaluation unit to generate decision data. The automated execution and feedback module receives the decision data, parses the task through the instruction parsing unit, executes the action through the interface calling unit and the signal sending unit, and generates execution feedback data through the status monitoring and acquisition unit. The model and knowledge optimization module receives the execution feedback data, calculates performance indicators through the effect comparison unit, updates the model through the parameter tuning unit, and updates the knowledge base through the knowledge base management unit. The rules and knowledge base maintenance module receives the latest execution feedback data, adjusts the business rule base through the rule calibration unit, and updates the historical information database through the data archiving unit. The closed-loop iterative control module integrates the updated model, rule base, and database, and drives the system to enter a new round of processing through iterative triggering units.

[0016] Compared with existing technologies, the present invention provides a digital economy information management method and system, which has the following beneficial effects: 1. In this invention, by constructing a closed-loop management link from the collection of raw datasets, the generation of preprocessed datasets, the output of analysis results, the formation of decision data, to the acquisition of execution feedback data, the problem of feedback lag and disconnection in the traditional one-way process is solved. It can ensure that the effect of management actions is fed back to the upstream analysis and decision-making links in real time, thereby ensuring the continuity and real-time nature of the entire process of digital economy information management and improving the overall response speed and collaborative efficiency of management.

[0017] 2. In this invention, by optimizing the parameters and updating the knowledge base of the adaptive artificial intelligence analysis model based on execution feedback data, and by dynamically calibrating and incrementally storing the business rule base and historical information in conjunction with the latest business results, it can continuously improve itself according to the actual execution results, solving the problem that static systems cannot adapt to environmental changes. This ensures the long-term accuracy of the analysis results and the continuous adaptability of decision data, and realizes the spiral evolution of management efficiency.

[0018] 3. In this invention, raw datasets are collected in real time from multiple heterogeneous data sources and generated synchronously. These datasets are then intelligently preprocessed to produce high-quality preprocessed datasets, providing a unified and reliable data foundation for subsequent in-depth analysis. At the same time, by combining business rule bases and historical information, the analysis results are quickly transformed into actionable decision data, solving the problems of difficulty in integrating multi-source data and decision delays. This improves the overall efficiency of digital economy information management from data perception to decision-making action. Attached Figure Description

[0019] Figure 1 This is a flowchart of a digital economy information management method according to the present invention; Figure 2 This is a diagram illustrating the architecture of a digital economy information management system according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figure 1-2 A digital economy information management method and system, characterized in that the method includes the following steps: S1. Collect raw digital economy information from multiple heterogeneous data sources, and synchronize and identify the source of the collected data to generate the raw dataset. S2. Perform intelligent preprocessing on the original dataset. Intelligent preprocessing includes data cleaning, format standardization, outlier detection, and multi-source data fusion to generate a preprocessed dataset. S3. Based on the preprocessed dataset, perform in-depth analysis using an adaptive artificial intelligence analysis model to generate analysis results; S4. Based on the analysis results, combined with business rules and historical information, generate a decision-making plan, format it into executable instructions, and generate decision data; S5. Based on the decision data, the automated execution engine calls the interface and sends instructions to implement management actions, and synchronously collects the execution status and generates execution feedback data. S6. Based on the execution feedback data and analysis results, perform parameter tuning and knowledge base update on the adaptive artificial intelligence analysis model to generate an optimized analysis model. S7. Based on the preprocessed dataset, a new round of in-depth analysis is conducted using the optimized analysis model, and the business rule base and historical information are dynamically calibrated and incrementally stored in combination with the latest execution feedback data. S8. Based on the updated analysis model, business rule base, and historical information database, the method is driven into continuous closed-loop iteration and spiral evolution.

[0022] Generating the original dataset in S1 involves the following steps: S11. Obtain multi-source heterogeneous raw data streams in real time from IoT devices, enterprise business systems, public data platforms and mobile applications through the configured interface; S12. Perform timestamp alignment and data source identification on the acquired raw data stream to generate the raw dataset.

[0023] Generating a preprocessed dataset in S2 involves the following steps: S21. Perform data cleaning on the original dataset to remove duplicate records, fill in missing values, and correct formatting errors. This includes the following steps: S211. Perform regular expression-based format validation on the original dataset, and filter and remove records that do not conform to the predefined data structure; Format validation is achieved through regular expression matching, and the conditions for passing the validation are: ; in, Represents a string variable. Indicated by rules A set of defined formal languages; S212. Perform imputation processing on missing values ​​based on the nearest neighbor algorithm and assign specific identifiers; The imputation process based on the nearest neighbor algorithm calculates missing values ​​using the following formula: ; in, Indicates the first missing data points The interpolation estimate, This represents the number of nearest neighbors for the interpolation. Indicates missing data points of The set of nearest neighbor observations Represents a set No. The nearest neighbor observation; S213. By calculating the Z-score of numerical data, identify and remove statistical outliers that exceed a preset threshold; The Z-score is calculated using the following formula: ; in, Represents the Z-score value. Represents the original data value. This represents the mean of the dataset. The standard deviation of the dataset; The outlier determination rule based on Z-score is as follows: ; in, The preset threshold; S22. Standardize the format of the cleaned data and convert it into a unified data model predefined by the system. S23. Based on statistical models and machine learning models, perform anomaly detection and processing on the standardized data, specifically including the following steps: S231. For time-series data, an unsupervised learning algorithm based on isolated forest is used for outlier detection. The Isolation Forest algorithm calculates anomaly scores for data points by constructing isolated trees. The formula for calculating anomaly scores is as follows: ; in, Representing data points Abnormal scores, Representing data points The expected value of the path length in all isolated trees. A constant representing the normalized path length. This indicates that the dataset size is relevant; S232. For transaction data, apply rule engine-based logic validation to identify records that violate business logic; S233. Isolate the detected abnormal data to the pending review queue and automatically correct it according to the preset strategy; The automatic correction strategy is the mean substitution method, and the correction formula is: ; in, The corrected data values. This represents the number of valid data points within the same metric that were not marked as an anomaly. For the values ​​of valid data points, To express summation; S24. Apply data fusion algorithms to associate and integrate multi-source data that has been cleaned, standardized and anomaly-handled to generate a preprocessed dataset.

[0024] The data fusion algorithm in S24 adopts a fusion method based on Kalman filtering, which includes the following steps: S241. Timestamp alignment and unit unification are performed on the same indicator data from different data sources; Timestamp alignment is achieved by calculating the time offset and shifting it: ; in, This is the original timestamp. For reference base time, This serves as the time base for the data source. For the aligned timestamps, This is due to time deviation; Units are standardized; conversion is performed using a linear scaling factor. ; in, The original value, For predefined unit conversion factors, Values ​​in standard units; S242. Assign a dynamic weight based on historical accuracy to the data from each data source; S243. Use the Kalman filter algorithm to perform optimal estimation of multi-source data and generate fused high-confidence data values. The Kalman filter algorithm includes state prediction and update steps. The state prediction formula is: ; in, Indicates time step Prior state estimation, Represents the state transition matrix. Indicates time step Posterior state estimation, Represents the control input matrix. Represents the control vector; The state update formula is: ; in, Indicates time step Posterior state estimation, Indicates time step Posterior state estimation, Indicates Kalman gain, Represents the observed value. Represents the observation matrix; S244. Integrate the merged high-confidence data values ​​into the preprocessed dataset.

[0025] Generating decision data in S4 includes the following steps: S41. Input the analysis results into the rule reasoning engine and perform matching and reasoning with the predefined rules in the business rule base; S42. Use cases and statistical data from the historical information database to conduct a feasibility assessment and effect simulation of the preliminary decision-making scheme derived from rule reasoning; S43. The final decision scheme determined after evaluation and simulation is encoded into an instruction format that can be recognized and executed by the automated execution engine to generate decision data.

[0026] Generating the optimized analysis model in S6 includes the following steps: S61. Compare the analysis results with the corresponding execution feedback data, and calculate at least one of the following quantitative indicators: model prediction accuracy, decision execution success rate, resource utilization efficiency, and target achievement. S62. Based on the indicators, the internal parameters of the adaptive artificial intelligence analysis model are adjusted using optimization algorithms. S63. Extract new knowledge patterns and rules from the execution feedback data and update the knowledge base.

[0027] S62 employs an optimization algorithm to adjust the internal parameters of the adaptive artificial intelligence analysis model, including the following steps: S621, using a quantitative index as the loss function, calculate the output and execution feedback data of the adaptive artificial intelligence analysis model. The error between; The error is calculated using the following mean squared error loss function: ; in, Represents the loss function. Indicates the number of samples. Indicates the first The true value of each sample Indicates the first Predicted values ​​for each sample; S622. Calculate the gradient of the loss function with respect to the parameters of each layer of the model using the backpropagation algorithm; The gradient is calculated using the chain rule in the backpropagation algorithm, and its formula is as follows: ; in, Represents the loss function For the parameters of the first layer model gradient, This represents the partial derivative calculated via backpropagation; S623. Apply the gradient descent algorithm to iteratively update the model parameters based on the calculated gradient; The parameter updates for the gradient descent algorithm are performed using the following formula: ; in, Indicates the first Model parameters of the next iteration , Indicates the learning rate. Represents the loss function For model parameters The gradient.

[0028] Dynamic calibration and incremental storage in S7 include the following steps: S71. Using the preprocessed dataset and execution feedback data as input, a new round of analysis is performed using the optimized analysis model; S72. Based on the actual business results reflected in the latest execution feedback data, adjust the triggering conditions, weights, and logical relationships of relevant rules in the business rule base; S73. Store the latest execution feedback data, related analysis processes, and decision records as new entries in the historical information database.

[0029] In S72, adjusting the business rule base based on actual business results includes the following steps: S721. When the success rate of decision execution is lower than the preset threshold, the triggering conditions of the corresponding rule are automatically relaxed. The trigger condition adjustment uses a linear adjustment formula: ; in, This indicates the adjusted trigger threshold. Indicates the original trigger threshold. Indicates adjusting the step size. Indicates the current success rate of decision execution; S722. When resource utilization efficiency continuously deviates from the optimal range, adjust the weight parameters of the relevant resource allocation rules. S723. Record the version, time, and reason for each rule adjustment to form a rule change log.

[0030] The system includes: The data acquisition and synchronization module obtains raw information from multiple heterogeneous data sources, aligns the data time stamps through the time synchronization unit, marks the data source through the source identification processing unit, and generates the raw dataset. The intelligent preprocessing module receives the raw dataset, processes the data through the data cleaning unit and the format standardization unit, and integrates the data through the outlier detection unit and the multi-source data fusion unit to generate a preprocessed dataset. The adaptive analysis module receives the preprocessed dataset, extracts data features through the feature engineering unit, calls and trains the artificial intelligence analysis model using the adaptive model scheduling unit, and generates analysis results. The decision generation module receives the analysis results, generates preliminary decision schemes through the rule reasoning unit and the business rule base, and then verifies and optimizes them through the multi-objective optimization and simulation evaluation unit to generate decision data. The automated execution and feedback module receives decision data, parses the task through the instruction parsing unit, executes the action through the interface calling unit and the signal sending unit, and generates execution feedback data through the status monitoring and acquisition unit. The model and knowledge optimization module receives execution feedback data, calculates performance indicators through the effect comparison unit, updates the model through the parameter tuning unit, and updates the knowledge base through the knowledge base management unit. The rules and knowledge base maintenance module receives the latest execution feedback data, adjusts the business rule base through the rule calibration unit, and updates the historical information database through the data archiving unit. The closed-loop iterative control module integrates the updated model, rule base, and database, and drives the system to enter a new round of processing through iterative triggering units.

[0031] The operational steps of this digital economy information management method and system are as follows: Step 1: Data Acquisition and Synchronization After the system starts up, it first collects heterogeneous data from multiple sources. The data acquisition and synchronization module obtains raw digital economy information from heterogeneous data sources such as IoT devices and enterprise business systems, aligns the data timestamps through the time synchronization unit, and marks the data source through the source identification processing unit to generate a standardized raw dataset.

[0032] Step 2: Intelligent Preprocessing and Data Integration Next, the intelligent preprocessing module performs deep cleaning and integration on the original dataset. This module removes duplicate records and outliers through the data cleaning unit, converts it into a unified data pattern through the format standardization unit, and then processes it through the outlier detection unit and the multi-source data fusion unit to finally generate a high-quality preprocessed dataset.

[0033] Step 3: Adaptive Intelligent Analysis Once the data is ready, the adaptive analysis module begins to work. This module first extracts features from the preprocessed dataset through the feature engineering unit to construct a feature vector set. Subsequently, the adaptive model scheduling unit dynamically selects and combines models from the algorithm library according to the data features and analysis objectives to construct an adaptive artificial intelligence analysis model. The model is then trained and used for prediction, outputting analysis results that include quantitative indicators and predicted trends.

[0034] Step 4: Decision Generation and Solution Output Based on the generated analysis results, the decision generation module initiates the decision-making process. The rule reasoning unit matches and reasons with the rules in the business rule base to generate a preliminary solution. The multi-objective optimization and simulation evaluation unit then combines cases in the historical information database to verify the feasibility. Finally, the optimized decision solution is encoded into an instruction format that can be recognized by the automated execution engine, generating executable decision data.

[0035] Step 5: Automated Execution and Feedback Collection The decision data is transmitted to the automated execution and feedback module for execution. After the instruction parsing unit of this module parses the task content, the interface calling unit and signal sending unit execute the specific management actions. At the same time, the status monitoring and acquisition unit tracks the execution status and resource consumption indicators throughout the process and generates execution feedback data in real time.

[0036] Step Six: Model and Rule Optimization Once the system enters the optimization phase, the model and knowledge optimization module receives execution feedback data and calculates performance indicators through the effect comparison unit. The parameter tuning unit updates the parameters of the adaptive artificial intelligence analysis model, and the knowledge base management unit updates the knowledge base synchronously. On the other hand, the rules and knowledge base maintenance module adjusts the business rule base according to the latest business results through the rule calibration unit and incrementally stores the historical information database through the data archiving unit, thus completing the dynamic calibration of the knowledge system.

[0037] Step 7: Closed-loop iteration and spiral evolution Finally, the closed-loop iterative control module starts a new cycle. This module configures the updated model, rule base, and database as new initial parameters. Through iterative triggering units, the system is driven to re-execute the complete process from data collection to optimization and update. After each iteration, the system dynamically adjusts the strategy based on the evaluation results. Through multiple cycles, the performance is spiraled and evolved, ultimately enabling the entire system to have the ability to continuously evolve and adapt.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, 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. Without further limitations, 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 said element.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital economy information management method, characterized in that: The method includes the following steps: S1. Collect raw digital economy information from multiple heterogeneous data sources, and synchronize and identify the source of the collected data to generate the raw dataset. S2. Perform intelligent preprocessing on the original dataset. The intelligent preprocessing includes data cleaning, format standardization, outlier detection, and multi-source data fusion to generate a preprocessed dataset. S3. Based on the preprocessed dataset, perform in-depth analysis using an adaptive artificial intelligence analysis model to generate analysis results; S4. Based on the analysis results, a decision-making scheme is generated by combining business rules and historical information, and formatted into executable instructions to generate decision data; S5. Based on the decision data, the management actions are implemented by calling the interface and sending instructions through the automated execution engine, and the execution status is collected synchronously to generate execution feedback data. S6. Based on the execution feedback data and the analysis results, the parameters of the adaptive artificial intelligence analysis model are optimized and the knowledge base is updated to generate an optimized analysis model. S7. Based on the preprocessed dataset, a new round of in-depth analysis is performed using the optimized analysis model, and the business rule base and historical information are dynamically calibrated and incrementally stored in combination with the latest execution feedback data. S8. Based on the updated analysis model, business rule base, and historical information database, the method is driven into continuous closed-loop iteration and spiral evolution.

2. The digital economy information management method according to claim 1, characterized in that: The process of generating the original dataset in S1 includes the following steps: S11. Obtain multi-source heterogeneous raw data streams in real time from IoT devices, enterprise business systems, public data platforms and mobile applications through the configured interface; S12. The acquired raw data stream is timestamped and identified by the data source to generate the raw dataset.

3. The digital economy information management method according to claim 2, characterized in that: The process of generating the preprocessed dataset in S2 includes the following steps: S21. Perform data cleaning on the original dataset to remove duplicate records, fill in missing values, and correct format errors; S22. Standardize the format of the cleaned data and convert it into a unified data model predefined by the system. S23. Perform anomaly detection and processing on standardized data based on statistical models and machine learning models; S24. Apply a data fusion algorithm to associate and integrate the multi-source data after cleaning, standardization and anomaly processing to generate the preprocessed dataset.

4. The digital economy information management method according to claim 3, characterized in that: The data fusion algorithm in S24 adopts a fusion method based on Kalman filtering, which includes the following steps: S241. Timestamp alignment and measurement unit unification are performed on the same indicator data from different data sources; S242. Assign a weight to each data source based on its historical data accuracy, calculated dynamically. S243. Use the Kalman filter algorithm to perform optimal estimation of multi-source data and generate fused high-confidence data values. S244. Integrate the fused high-confidence data values ​​into the preprocessed dataset.

5. The digital economy information management method according to claim 1, characterized in that: The generation of decision data in S4 includes the following steps: S41. Input the analysis results into the rule reasoning engine and perform matching and reasoning with the predefined rules in the business rule base; S42. Call the cases and statistical data in the historical information database to conduct a feasibility assessment and effect simulation of the preliminary decision-making scheme derived from rule reasoning; S43. The final decision scheme determined after evaluation and simulation is encoded into an instruction format that the automated execution engine can recognize and execute, thereby generating the decision data.

6. The digital economy information management method according to claim 1, characterized in that: The process of generating the optimized analysis model in step S6 includes the following steps: S61. Compare the analysis results with the corresponding execution feedback data, and calculate at least one quantitative indicator among model prediction accuracy, decision execution success rate, resource utilization efficiency, and target achievement. S62. Based on the aforementioned indicators, the internal parameters of the adaptive artificial intelligence analysis model are adjusted using an optimization algorithm; S63. Extract new knowledge patterns and rules from the execution feedback data and update the knowledge base.

7. A digital economy information management method according to claim 6, characterized in that: The S62 step involves using an optimization algorithm to adjust the internal parameters of the adaptive artificial intelligence analysis model, including the following steps: S621. Using the quantitative index as a loss function, calculate the error between the output of the adaptive artificial intelligence analysis model and the execution feedback data; S622. Calculate the gradient of the loss function with respect to the parameters of each layer of the model using the backpropagation algorithm; S623. Apply the gradient descent algorithm to iteratively update the model parameters based on the calculated gradient.

8. The digital economy information management method according to claim 1, characterized in that: The dynamic calibration and incremental storage in S7 includes the following steps: S71. Using the preprocessed dataset and the execution feedback data as input, perform a new round of analysis using the optimized analysis model; S72. Based on the actual business effect reflected by the latest execution feedback data, adjust the triggering conditions, weights, and logical relationships of the relevant rules in the business rule base; S73. The latest execution feedback data, related analysis process and decision records are stored as new entries in the historical information database.

9. A digital economy information management method according to claim 8, characterized in that: The S72 step of adjusting the business rule base based on actual business results includes the following steps: S721. When the success rate of decision execution is lower than the preset threshold, the triggering conditions of the corresponding rule are automatically relaxed. S722. When resource utilization efficiency continuously deviates from the optimal range, adjust the weight parameters of the relevant resource allocation rules. S723. Record the version, time, and reason for each rule adjustment to form a rule change log.

10. A digital economy information management system, used to implement the digital economy information management method according to any one of claims 1-9, characterized in that: The system includes: The data acquisition and synchronization module obtains raw information from multiple heterogeneous data sources, aligns the data time stamps through the time synchronization unit, marks the data source through the source identification processing unit, and generates the raw dataset. The intelligent preprocessing module receives the original dataset, processes the data through a data cleaning unit and a format standardization unit, and integrates the data through an outlier detection unit and a multi-source data fusion unit to generate a preprocessed dataset. The adaptive analysis module receives the preprocessed dataset, extracts data features through the feature engineering unit, calls and trains an artificial intelligence analysis model using the adaptive model scheduling unit, and generates analysis results. The decision generation module receives the analysis results, generates a preliminary decision scheme by combining the rule reasoning unit with the business rule base, and verifies and optimizes it through the multi-objective optimization and simulation evaluation unit to generate decision data. The automated execution and feedback module receives the decision data, parses the task through the instruction parsing unit, executes the action through the interface calling unit and the signal sending unit, and generates execution feedback data through the status monitoring and acquisition unit. The model and knowledge optimization module receives the execution feedback data, calculates performance indicators through the effect comparison unit, updates the model through the parameter tuning unit, and updates the knowledge base through the knowledge base management unit. The rules and knowledge base maintenance module receives the latest execution feedback data, adjusts the business rule base through the rule calibration unit, and updates the historical information database through the data archiving unit. The closed-loop iterative control module integrates the updated model, rule base, and database, and drives the system to enter a new round of processing through iterative triggering units.

Images