Intelligent analysis and decision support system of artificial intelligence-based data asset management platform

Abstract

The application belongs to the technical field of data asset management, and provides an intelligent analysis and decision support system of a data asset management platform based on artificial intelligence, which comprises: a data asset data acquisition and processing module, which is used for acquiring and obtaining preprocessed data asset data; an intelligent analysis module, which is used for performing data asset value evaluation and data asset risk analysis on the preprocessed data asset data to obtain value evaluation results and risk analysis results; and a target management strategy generation module, which is used for generating a target management strategy based on the value evaluation results and the risk analysis results. The application can automatically process and analyze a large amount of data asset data by integrating artificial intelligence algorithms and models, and significantly improves the efficiency and accuracy of data asset management.

Description

Technical Field

[0001] This invention relates to the field of data asset management technology, and in particular to an intelligent analysis and decision support system for a data asset management platform based on artificial intelligence. Background Technology

[0002] With the rapid development of information technology, data assets have become an important strategic resource for enterprises and organizations. Data assets include, but are not limited to, digital documents, images, videos, audio, databases, software code, and encrypted assets. Traditional data asset management methods mainly rely on manual classification, storage, and retrieval, which have the following technical problems: First, data processing efficiency is low. Traditional methods are difficult to cope with the needs of efficient processing and analysis of massive amounts of data assets, and manual processing is slow and prone to errors. Second, the analysis dimensions are limited. Existing systems can usually only provide basic data statistics and simple analysis, lacking multi-dimensional and in-depth intelligent analysis capabilities. Third, decision support is insufficient. Most data asset management platforms only provide data display functions and cannot provide managers with scientific decision-making suggestions based on data analysis results. Fourth, real-time performance is poor. Traditional systems cannot respond to dynamic changes in the status of data assets in a timely manner and cannot provide real-time analysis and decision support. Fifth, security risks are high. The security management of data assets mainly relies on static permission settings and lacks dynamic security protection mechanisms based on behavioral analysis.

[0003] Therefore, it is necessary to provide an intelligent analysis and decision support system for data asset management platforms based on artificial intelligence. Summary of the Invention

[0004] This invention provides an intelligent analysis and decision support system for a data asset management platform based on artificial intelligence. By integrating artificial intelligence algorithms and models, it can automatically process and analyze large amounts of data asset data, significantly improving the efficiency and accuracy of data asset management.

[0005] This invention provides an intelligent analysis and decision support system for a data asset management platform based on artificial intelligence, comprising:

[0006] The data asset acquisition and processing module is used to acquire and process pre-processed data asset data.

[0007] The intelligent analysis module is used to perform data asset value assessment and data asset risk analysis on preprocessed data assets, and obtain value assessment results and risk analysis results.

[0008] The goal management strategy generation module is used to generate goal management strategies based on value assessment results and risk analysis results.

[0009] Furthermore, preprocessed data asset data is collected and acquired, including:

[0010] Based on configured API interfaces, database connectors, and file parsers, it collects data asset data in various structural forms from the data repository; performs quality inspection, standardization processing, and metadata extraction on the data asset data to generate preprocessed data asset data; the various structural forms include structured, semi-structured, and unstructured data.

[0011] Furthermore, based on the configured API interface, database connector, and file parser, data asset data in various structural forms is collected from the data repository, including:

[0012] By configuring the API gateway component and utilizing the web crawler component, data asset information from public data sources can be retrieved.

[0013] Capture real-time data asset transaction data on the blockchain using a blockchain listener;

[0014] Data asset data is composed of data asset information and data asset transaction data.

[0015] Furthermore, the data asset data undergoes quality inspection, standardization processing, and metadata extraction to generate preprocessed data asset data, including:

[0016] A detection mechanism that employs syntax layer detection, semantic layer detection, and business rule layer detection is adopted to detect the data quality of data assets;

[0017] Standardize and convert the data assets that have passed quality inspection to generate data assets in a unified intermediate format.

[0018] A deep learning model is used to process data assets in a unified intermediate format to generate descriptive metadata and structural metadata, and then metadata is extracted and generated based on the descriptive metadata and structural metadata.

[0019] Furthermore, the intelligent analysis module includes a feature extraction unit, a value assessment unit, and a risk analysis unit;

[0020] The feature extraction unit is used to extract data features from preprocessed data asset data based on text feature extractors and image feature extractors, and to generate statistical features, frequency domain features and nonlinear features using a time-series feature builder, and to establish mapping relationships between statistical features, frequency domain features and nonlinear features using a cross-modal correlator.

[0021] The value assessment unit is used to set three assessment indicators—utility value, market value, and strategic value—based on data characteristics and mapping relationships. The weights of the assessment indicators are determined using the analytic hierarchy process (AHP). Based on the assessment indicators and their weights, the XGBoost model is used to predict the value score. Based on the obtained value score prediction results, the data asset data is valued to obtain the value assessment results.

[0022] The risk analysis unit is used to detect technical vulnerabilities based on the OWASP application security assessment standard using a static risk analyzer; to build a user behavior baseline using an LSTM network using a dynamic behavior analyzer; to achieve real-time connection with third-party threat intelligence sources by combining with a threat intelligence integration component; and to simulate the path of risk propagation among multiple data assets using a risk transmission analysis model to generate risk analysis results.

[0023] Furthermore, the target management strategy generation module includes a preliminary management strategy generation unit and a target management strategy optimization unit;

[0024] The preliminary management strategy generation unit is used to generate preliminary management strategies based on the value assessment results and risk analysis results;

[0025] The target management strategy optimization unit is used to optimize the initial management strategy using reinforcement learning algorithms to generate a target management strategy.

[0026] Furthermore, based on the value assessment results and risk analysis results, preliminary management strategies are generated, including:

[0027] Based on the retrieval of similar historical cases and applicable scenarios, combined with the results of value assessment and risk analysis, scenario management strategies are generated.

[0028] Based on the established business rule tree, combined with the value assessment results and risk analysis results, business management rules are formulated, and business management strategies are generated.

[0029] The multi-objective genetic algorithm NSGA-II is used to optimize the scenario management strategy and business management strategy in multiple ways, and generate a preliminary management strategy.

[0030] Furthermore, reinforcement learning algorithms are used to optimize the initial management strategy and generate a target management strategy, including:

[0031] A hierarchical reinforcement learning framework is adopted to construct an interaction mechanism between four parallel evaluation units and the central decision-maker. The four parallel evaluation units include a feasibility evaluation unit, an economic evaluation unit, a risk evaluation unit, and a sensitivity analysis unit. The feasibility evaluation unit is used to detect and evaluate the resource constraints and technical limitations of the initial management strategy; the economic evaluation unit is used to calculate the ROI and NPV indicators of the initial management strategy; the risk evaluation unit is used to predict the implementation risks of the initial management strategy; and the sensitivity analysis unit is used to calculate parameter sensitivity based on the Sobol index. The formula for calculating the NPV indicator is as follows:

[0032]

[0033] In the above formula, θ represents the industry benchmark rate of return, ε represents the sensitivity to achieving the return target, and r t Represents the time-varying cost of capital. Represents a correction to cash flow; β t R represents the technology maturity moderating factor. t Represents time-varying cash flow; t represents the number of investment periods; T represents the total investment periods;

[0034] Based on a central decision-maker, a multi-objective proximal strategy optimization algorithm is used to optimize the evaluation results of four parallel evaluation units to obtain an objective management strategy.

[0035] Furthermore, it also includes a security protection module; the security protection module includes:

[0036] Encryption control is implemented for fine-grained data access using the configured attribute-based encryption component.

[0037] Use a behavior audit tracker to record complete operation logs and generate audit reports;

[0038] Anomaly detectors are used to identify abnormal access behaviors to data assets using the Isolation Forest algorithm;

[0039] Based on the threat level detected, protective measures are automatically triggered; the threat level is determined based on real-time detection using statistical rules, deep detection based on machine learning, and correlation detection based on graph neural networks.

[0040] Furthermore, it also includes a feedback learning module, which is used to optimize the functions of the intelligent analysis module and the goal management strategy generation module based on the collected user feedback on the goal management strategy; the feedback learning module includes a feedback content analysis unit, a targeted improvement mechanism construction unit, and a feedback database construction unit;

[0041] The feedback content analysis unit is used to classify the collected user feedback on the target management strategy in multiple dimensions, obtain a feedback classification dataset, and generate a feedback hotspot distribution map based on the classification dataset; the multi-dimensional feedback classification includes data quality, analysis logic, and decision applicability.

[0042] The targeted improvement mechanism construction unit is used to take targeted improvement measures for the feedback content in the feedback hotspot distribution map. Specifically, for data quality feedback, data cleaning is carried out according to the set data cleaning process; for analysis logic feedback, the feature importance weights are adjusted according to the set adjustment range; for decision applicability feedback, the constraint condition check is strengthened, the cost-benefit calculation parameters are adjusted, or the risk aversion coefficient is increased.

[0043] The feedback database construction unit is used to generate a knowledge graph of the feedback classification dataset relative to the improvement measures based on the feedback classification dataset and the improvement measures, and to construct the feedback database based on the knowledge graph;

[0044] It also includes, based on the feedback database, in the process of generating target management strategies based on value assessment results and risk analysis results, first performing a self-check of the health of the generated target management strategies. If the health is less than the set health threshold, then based on the set causal reasoning model, the core variables affecting the generation of target management strategies are located, and counterfactual decisions are made in combination with counterfactual simulation methods. Based on the counterfactual decision results, the optimization parameters of the initial management strategy of the target management strategy generation module are adjusted in a Bayesian optimization manner.

[0045] Compared with existing technologies, this invention has the following advantages and beneficial effects: By integrating artificial intelligence algorithms and models, it can automatically process and analyze large amounts of data asset data, significantly improving the efficiency and accuracy of data asset management; it can update the value assessment of data assets in real time, helping users to understand the asset status in a timely manner and providing data support for investment decisions; through in-depth analysis of data asset risks, the system can identify potential risk factors, provide early warnings, and reduce investment risks; the target management strategy generation module, based on the results of value assessment and risk analysis, tailors management strategies for users to achieve optimized allocation and value-added of data assets.

[0046] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0047] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0048] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0049] Figure 1 This is a schematic diagram of the intelligent analysis and decision support system structure of an AI-based data asset management platform;

[0050] Figure 2 This is a schematic diagram of the intelligent analysis module structure;

[0051] Figure 3 A schematic diagram of the module structure for generating target management strategies. Detailed Implementation

[0052] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0053] This invention provides an intelligent analysis and decision support system for a data asset management platform based on artificial intelligence, such as... Figure 1 As shown, it includes:

[0054] The data asset acquisition and processing module is used to acquire and process pre-processed data asset data.

[0055] The intelligent analysis module is used to perform data asset value assessment and data asset risk analysis on preprocessed data assets, and obtain value assessment results and risk analysis results.

[0056] The goal management strategy generation module is used to generate goal management strategies based on value assessment results and risk analysis results.

[0057] The working principle of the above technical solution is as follows: In order to realize an intelligent analysis and decision support system for an artificial intelligence-based data asset management platform, this invention first uses a data asset data acquisition and processing module to automatically collect and acquire a large amount of data asset data from various data sources and preprocess it to ensure the accuracy and consistency of the data. Subsequently, the intelligent analysis module uses artificial intelligence algorithms to conduct in-depth value assessment and risk analysis on the preprocessed data asset data. This step not only considers the current market value of the data assets, but also predicts their future growth potential and potential risks. Based on these assessment and analysis results, the target management strategy generation module can automatically generate a set of targeted target management strategies, aiming to maximize the value of data assets and reduce potential risks.

[0058] The beneficial effects of the above technical solution are as follows: By integrating artificial intelligence algorithms and models, the solution provided in this embodiment can automatically process and analyze large amounts of data asset data, significantly improving the efficiency and accuracy of data asset management; it can update the value assessment of data assets in real time, helping users to understand the asset status in a timely manner and providing data support for investment decisions; through in-depth analysis of data asset risks, the system can identify potential risk factors, provide early warnings, and reduce investment risks; the target management strategy generation module can tailor management strategies for users based on the results of value assessment and risk analysis, thereby achieving optimized allocation and value-added of data assets.

[0059] In one embodiment, acquiring preprocessed data asset data includes:

[0060] Based on configured API interfaces, database connectors, and file parsers, it collects data asset data in various structural forms from the data repository; performs quality inspection, standardization processing, and metadata extraction on the data asset data to generate preprocessed data asset data; the various structural forms include structured, semi-structured, and unstructured data.

[0061] The working principle of the above technical solution is as follows: First, it connects to various data sources through configured API interfaces. These data sources may include relational databases, NoSQL databases, cloud storage services, etc., ensuring the collection of data assets stored in different locations. The database connector is specifically used for efficient data interaction with relational databases to extract the required structured data, while the file parser performs format parsing and content extraction for data assets not stored in databases, such as Excel files, CSV files, and JSON files, supporting the collection of semi-structured and unstructured data. After data collection, the raw data assets undergo quality inspection. This step mainly checks the completeness, accuracy, and timeliness of the data to ensure that subsequent analysis is based on a reliable data foundation. Next, the data is standardized, including operations such as unifying data formats, converting data units, and filling missing values, so that data from different sources and in different formats can be analyzed and processed under a unified standard. At the same time, the system also extracts metadata, adding descriptive information such as data source, collection time, and data type to each data asset, facilitating subsequent data management and use.

[0062] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, through the above-mentioned steps of quality inspection, standardization processing and metadata extraction, the system finally generates preprocessed data asset data, providing a high-quality data foundation for subsequent intelligent analysis and decision support.

[0063] In one embodiment, based on configured API interfaces, database connectors, and file parsers, data asset data in various structural formats is collected from a data repository, including:

[0064] By configuring the API gateway component and utilizing the web crawler component, data asset information from public data sources can be retrieved.

[0065] Capture real-time data asset transaction data on the blockchain using a blockchain listener;

[0066] Data asset data is composed of data asset information and data asset transaction data.

[0067] The working principle of the above technical solution is as follows: First, the configured API interface interacts with different data repositories according to preset rules and parameters. These repositories may include cloud storage, local databases, third-party data service platforms, etc., to obtain the data assets stored therein. This data may exist in the form of relational databases, or in the form of NoSQL databases, document databases, etc., allowing the system to flexibly handle various data structures. Second, the database connector is specifically used to connect to and access relational databases, efficiently extracting the required data assets through SQL queries or other database access protocols. This process ensures that the system can fully utilize existing database management systems to achieve rapid data collection and integration. Simultaneously, the web crawler component, as part of the API gateway, automatically accesses publicly available data on the Internet according to preset crawling rules and strategies. Sources such as news websites, social media platforms, and financial websites are used to extract information related to data assets. This information may include the latest prices, trading volumes, and market dynamics of data assets, providing investors with real-time market references. Blockchain listeners are specifically used to monitor data asset transaction data on the blockchain network. By subscribing to transaction events on the blockchain, the listener can capture real-time transaction information for each data asset, including key data such as the trading parties, transaction volume, and transaction price. This data is crucial for analyzing market trends and assessing investment value of data assets. Finally, the collected data asset information and transaction data are integrated and processed to form a complete data asset dataset. These datasets provide a rich and accurate data foundation for subsequent intelligent analysis and decision support, enabling the platform to provide users with more precise and valuable investment advice and market analysis.

[0068] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, it is possible to achieve comprehensive and efficient collection and integration of data assets from multiple sources and in multiple forms, which solves the problems of data silos and data fragmentation in traditional data asset management; through various technical means such as configured API interfaces, database connectors, web crawler components and blockchain listeners, the system can flexibly cope with various complex data environments and ensure the comprehensiveness and accuracy of the data.

[0069] In one embodiment, data asset data undergoes quality inspection, standardization processing, and metadata extraction to generate preprocessed data asset data, including:

[0070] A detection mechanism that employs syntax layer detection, semantic layer detection, and business rule layer detection is adopted to detect the data quality of data assets;

[0071] Standardize and convert the data assets that have passed quality inspection to generate data assets in a unified intermediate format.

[0072] A deep learning model is used to process data assets in a unified intermediate format to generate descriptive metadata and structural metadata, and then metadata is extracted and generated based on the descriptive metadata and structural metadata.

[0073] The working principle of the above technical solution is as follows: First, syntax layer detection ensures that the data format and encoding conform to the specifications, avoiding data parsing failures due to format errors; semantic layer detection focuses on the meaning and contextual relationships of data fields to ensure logical consistency and accuracy of the data; business rule layer detection further verifies and filters the data according to specific business logic and data requirements to ensure the business relevance and compliance of the data; during the data standardization and transformation process, the system converts data assets from different sources and formats into a unified intermediate format according to preset data mapping rules and transformation algorithms. This not only eliminates processing obstacles caused by data format differences but also facilitates subsequent data analysis and processing; the application of deep learning models is to more deeply mine and understand the information and features in the data. Through the processing of data asset data, the model can automatically generate descriptive metadata and structural metadata. Descriptive metadata mainly describes the basic information such as the nature, source, and purpose of the data, helping users quickly understand the overview of the data; structural metadata records in detail the relationships and hierarchical structure between data, providing the possibility for efficient data retrieval and analysis. Based on this metadata, the system can further extract and integrate valuable information, providing strong support for intelligent analysis and decision support.

[0074] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, the multi-level data detection mechanism effectively ensures data quality, reduces the risk of data errors and anomalies, and provides a reliable foundation for subsequent data analysis and decision-making; on the other hand, the application of data standardization conversion and deep learning models not only solves the problems caused by data format differences, but also deeply mines the information and features in the data, providing the possibility for in-depth data utilization; in addition, the solution can automatically generate rich metadata, providing users with convenient data retrieval and analysis methods, further improving the efficiency and value of data use.

[0075] In one embodiment, such as Figure 2 As shown, the intelligent analysis module includes a feature extraction unit, a value assessment unit, and a risk analysis unit;

[0076] The feature extraction unit is used to extract data features from preprocessed data asset data based on text feature extractors and image feature extractors, and to generate statistical features, frequency domain features and nonlinear features using a time-series feature builder, and to establish mapping relationships between statistical features, frequency domain features and nonlinear features using a cross-modal correlator.

[0077] The value assessment unit is used to set three assessment indicators—utility value, market value, and strategic value—based on data characteristics and mapping relationships. The weights of the assessment indicators are determined using the analytic hierarchy process (AHP). Based on the assessment indicators and their weights, the XGBoost model is used to predict the value score. Based on the obtained value score prediction results, the data asset data is valued to obtain the value assessment results.

[0078] The risk analysis unit is used to detect technical vulnerabilities based on the OWASP application security assessment standard using a static risk analyzer; to build a user behavior baseline using an LSTM network using a dynamic behavior analyzer; to achieve real-time connection with third-party threat intelligence sources by combining with a threat intelligence integration component; and to simulate the path of risk propagation among multiple data assets using a risk transmission analysis model to generate risk analysis results.

[0079] The working principle of the above technical solution is as follows: The intelligent analysis module first uses text and image feature extraction technology through the feature extraction unit to deeply mine key information in the preprocessed data asset data. These features not only cover statistical features, frequency domain features, and nonlinear features, but also establish complex mapping relationships between them through cross-modal correlators, providing a solid foundation for subsequent value assessment and risk analysis. In the value assessment unit, the system combines three dimensions: utility value, market value, and strategic value. The weights of each assessment indicator are scientifically determined using the analytic hierarchy process (AHP). Subsequently, utilizing the efficient predictive capabilities of the XGBoost model, the system can accurately predict the value of the data asset. The system scores data assets to achieve a comprehensive assessment of their value. This process not only improves the accuracy and objectivity of the assessment but also provides strong data support for asset management and decision-making. The risk analysis unit focuses on the security and stability of data assets. Through the collaborative work of static risk analyzers and dynamic behavior analyzers, the system can promptly detect technical vulnerabilities and abnormal behaviors, effectively preventing potential security risks. At the same time, real-time integration with third-party threat intelligence sources further enhances the system's early warning and response capabilities. In addition, the application of risk transmission analysis models enables the system to simulate the propagation path of risks among multiple data assets, providing a scientific basis for developing targeted risk prevention measures.

[0080] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, the intelligent analysis module achieves comprehensive, in-depth and intelligent analysis of data asset data through the close cooperation of the three units of feature extraction, value assessment and risk analysis, providing strong technical support for the intelligent analysis and decision support system of the data asset management platform.

[0081] In one embodiment, such as Figure 3 As shown, the target management strategy generation module includes a preliminary management strategy generation unit and a target management strategy optimization unit;

[0082] The preliminary management strategy generation unit is used to generate preliminary management strategies based on the value assessment results and risk analysis results;

[0083] The target management strategy optimization unit is used to optimize the initial management strategy using reinforcement learning algorithms to generate a target management strategy.

[0084] The working principle of the above technical solution is as follows: The preliminary management strategy generation unit receives the output results from the value assessment unit and the risk analysis unit. These two results reflect the value score and potential risk of the data asset, respectively. Based on this information, the preliminary management strategy generation unit will automatically generate a series of preliminary management strategy suggestions using preset decision logic and rules. These strategy suggestions aim to balance the maximization of data asset value and risk control, and may include investment strategies, maintenance strategies, security protection strategies, etc. Subsequently, the target management strategy optimization unit takes the output from the preliminary management strategy generation unit and further optimizes these strategies using reinforcement learning algorithms. The reinforcement learning algorithm continuously adjusts the strategy parameters by simulating the execution effect of different strategies in the actual environment in order to maximize long-term returns or minimize risks. This optimization process makes the target management strategy more in line with the actual scenario and can maintain high adaptability and competitiveness in a complex and ever-changing market environment.

[0085] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, it is possible to achieve precise management and efficient decision-making of data assets, providing strong support for the digital transformation and sustainable development of enterprises.

[0086] In one embodiment, based on the value assessment results and risk analysis results, a preliminary management strategy is generated, including:

[0087] Based on the retrieval of similar historical cases and applicable scenarios, combined with the results of value assessment and risk analysis, scenario management strategies are generated.

[0088] Based on the established business rule tree, combined with the value assessment results and risk analysis results, business management rules are formulated, and business management strategies are generated.

[0089] The multi-objective genetic algorithm NSGA-II is used to optimize the scenario management strategy and business management strategy in multiple ways, and generate a preliminary management strategy.

[0090] The working principle of the above technical solution is as follows: To generate a preliminary management strategy, historical similar cases and suitable scenarios are first retrieved. By searching for historical cases similar to the current data asset situation, these cases may involve different market environments, asset types, value assessments, and risk analysis results. After comparative analysis, the cases that best match the current situation can be extracted. Based on the management strategies of these cases, combined with the current value assessment and risk analysis results, a targeted scenario management strategy is generated. At the same time, the business rule tree is set based on the actual business needs and processes of the enterprise. The business rules are organized and expressed in a tree structure. The system can automatically traverse the business rule tree and match the corresponding business rules according to the value assessment and risk analysis results, thereby formulating a business management strategy. The application of the multi-objective genetic algorithm NSGA-II is to solve the potential conflicts and contradictions between the scenario management strategy and the business management strategy. This algorithm can optimize the strategy while maintaining a balance of multiple objectives (such as value maximization, risk control, etc.), ensuring that the generated preliminary management strategy not only meets the business needs of the enterprise, but also maintains high adaptability and competitiveness in a complex and ever-changing market environment.

[0091] The beneficial effects of the above technical solution are as follows: The solution provided in this embodiment can significantly improve the intelligence level of data asset management. First, by searching and comparing historical similar cases, the management strategy that best matches the current situation can be quickly located, greatly shortening the strategy formulation time and improving work efficiency. Second, the setting of the business rule tree enables the system to automatically match business rules, avoiding the subjectivity and uncertainty of human judgment and improving the accuracy and scientific nature of strategy formulation. Finally, the application of the multi-objective genetic algorithm NSGA-II can optimize the strategy while maintaining the balance of multiple objectives, ensuring that the generated preliminary management strategy meets the business needs of the enterprise.

[0092] In one embodiment, a reinforcement learning algorithm is used to optimize the initial management strategy and generate a target management strategy, including:

[0093] A hierarchical reinforcement learning framework is adopted to construct an interaction mechanism between four parallel evaluation units and the central decision-maker. The four parallel evaluation units include a feasibility evaluation unit, an economic evaluation unit, a risk evaluation unit, and a sensitivity analysis unit. The feasibility evaluation unit is used to detect and evaluate the resource constraints and technical limitations of the initial management strategy; the economic evaluation unit is used to calculate the ROI and NPV indicators of the initial management strategy; the risk evaluation unit is used to predict the implementation risks of the initial management strategy; and the sensitivity analysis unit is used to calculate parameter sensitivity based on the Sobol index. The formula for calculating the NPV indicator is as follows:

[0094]

[0095] In the above formula, θ represents the industry benchmark rate of return, ε represents the sensitivity to achieving the return target, and r t Represents the time-varying cost of capital. Represents a correction to cash flow; β t R represents the technology maturity moderating factor. t R represents time-varying cash flow. t The 't' represents the rate of return on investment; 't' represents the number of investment periods; and 'T' represents the total number of investment periods.

[0096] Based on a central decision-maker, a multi-objective proximal strategy optimization algorithm is used to optimize the evaluation results of four parallel evaluation units to obtain an objective management strategy.

[0097] The working principle of the above technical solution is as follows: First, the preliminary management strategy is preset by the system or generated by the user, serving as the starting point for the intelligent analysis and decision support system. After receiving the preliminary management strategy, the system inputs it into a hierarchical reinforcement learning framework. Four parallel evaluation units—feasibility evaluation unit, economic evaluation unit, risk evaluation unit, and sensitivity analysis unit—begin a comprehensive evaluation of the preliminary management strategy. The feasibility evaluation unit determines whether the strategy has the basic conditions for implementation by detecting resource constraints and technical limitations. If the strategy exceeds resource constraints or cannot be implemented technically, adjustment suggestions are fed back to the central decision-maker. The economic evaluation unit focuses on the economic benefits of the preliminary management strategy, calculating ROI (Return on Investment) and NPV (Net Present Value) indicators to evaluate the financial feasibility of the strategy. If the NPV is negative or the ROI is lower than the expected value, the financial evaluation unit will adjust the strategy accordingly. The system establishes industry benchmarks and provides improvement suggestions to the central decision-maker. The risk assessment unit predicts potential risks during strategy implementation, including market, technological, and policy risks, providing the central decision-maker with risk warnings and response strategies. The sensitivity analysis unit calculates parameter sensitivity based on the Sobol index, identifying the key factors that have the greatest impact on strategy outcomes and providing a scientific basis for strategy adjustments. After receiving the evaluation results from the four parallel evaluation units, the central decision-maker uses a multi-objective proximal strategy optimization algorithm to optimize the strategy, achieving a target management strategy that satisfies resource constraints and technological limitations while possessing good economic benefits and low risk. The entire intelligent analysis and decision support system forms a closed loop, gradually approaching the optimal management strategy through continuous iteration and optimization, providing powerful decision support for data asset management.

[0098] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, the level of intelligence in data asset management can be significantly improved; through the collaborative work of the hierarchical reinforcement learning framework and four parallel evaluation units, the system can conduct a comprehensive and in-depth evaluation of the preliminary management strategy, ensuring that the strategy is fully considered in terms of feasibility, economy, risk control and sensitivity of key factors. This not only helps to avoid blind decision-making and waste of resources, but also effectively improves the implementation effect and return on investment of the strategy.

[0099] In one embodiment, a security protection module is further included; the security protection module includes:

[0100] Encryption control is implemented for fine-grained data access using the configured attribute-based encryption component.

[0101] Use a behavior audit tracker to record complete operation logs and generate audit reports;

[0102] Anomaly detectors are used to identify abnormal access behaviors to data assets using the Isolation Forest algorithm;

[0103] Based on the threat level detected, protective measures are automatically triggered; the threat level is determined based on real-time detection using statistical rules, deep detection based on machine learning, and correlation detection based on graph neural networks.

[0104] The working principle of the above technical solution is as follows: This invention also includes a security protection module. The attribute-based encryption component ensures that only users with the corresponding attributes can access specific data, effectively preventing data leakage. The behavior audit tracker records all user operations, providing detailed evidence for the tracing and auditing of security events. The anomaly detector monitors the access behavior of data assets in real time, and triggers protective measures immediately upon detecting anomalies, effectively resisting potential security threats. The determination of threat levels is combined with real-time detection using statistical rules, which can quickly identify abnormal behaviors that deviate from normal access patterns. Deep detection based on machine learning improves the accuracy and robustness of detection by continuously learning normal and abnormal behavior patterns. Correlation detection based on graph neural networks can uncover potential connections between different abnormal behaviors, further enhancing the system's security protection capabilities. The entire security protection module's workflow forms a closed loop, ensuring the security of data assets through real-time monitoring, early warning, and protection.

[0105] The beneficial effects of the above technical solution are as follows: the solution provided in this embodiment can significantly improve the security protection level of the data asset management platform; the application of the attribute-based encryption component realizes fine-grained access control of sensitive data, avoiding the risk of data leakage due to improper permission management; the recording function of the behavior audit tracker provides reliable evidence for tracing security events, helping to discover and deal with potential security risks in a timely manner; the real-time monitoring and rapid response mechanism of the anomaly detector effectively prevents abnormal access behavior and reduces the impact of security threats; at the same time, the application of deep detection based on machine learning and correlation detection based on graph neural networks not only improves the accuracy and robustness of detection, but also can uncover the potential connections between different abnormal behaviors, providing strong support for formulating more precise security protection strategies.

[0106] In one embodiment, a feedback learning module is also included, which is used to optimize the functions of the intelligent analysis module and the goal management strategy generation module based on the collected user feedback on the goal management strategy; the feedback learning module includes a feedback content analysis unit, a targeted improvement mechanism construction unit, and a feedback database construction unit;

[0107] The feedback content analysis unit is used to classify the collected user feedback on the target management strategy in multiple dimensions, obtain a feedback classification dataset, and generate a feedback hotspot distribution map based on the classification dataset; the multi-dimensional feedback classification includes data quality, analysis logic, and decision applicability.

[0108] The targeted improvement mechanism construction unit is used to take targeted improvement measures for the feedback content in the feedback hotspot distribution map. Specifically, for data quality feedback, data cleaning is carried out according to the set data cleaning process; for analysis logic feedback, the feature importance weights are adjusted according to the set adjustment range; for decision applicability feedback, the constraint condition check is strengthened, the cost-benefit calculation parameters are adjusted, or the risk aversion coefficient is increased.

[0109] The feedback database construction unit is used to generate a knowledge graph of the feedback classification dataset relative to the improvement measures based on the feedback classification dataset and the improvement measures, and to construct the feedback database based on the knowledge graph;

[0110] It also includes, based on the feedback database, in the process of generating target management strategies based on value assessment results and risk analysis results, first performing a self-check of the health of the generated target management strategies. If the health is less than the set health threshold, then based on the set causal reasoning model, the core variables affecting the generation of target management strategies are located, and counterfactual decisions are made in combination with counterfactual simulation methods. Based on the counterfactual decision results, the optimization parameters of the initial management strategy of the target management strategy generation module are adjusted in a Bayesian optimization manner.

[0111] The working principle of the above technical solution is as follows: The feedback learning module continuously receives and analyzes user feedback, thereby continuously optimizing the intelligent analysis module and the target management strategy generation module. First, the feedback content analysis unit meticulously categorizes user feedback. These data categories cover not only data quality and analysis logic but also decision applicability, ensuring a comprehensive understanding of the feedback content. By generating a feedback hotspot distribution map, the system can intuitively identify the feedback points that users are most concerned about, providing a clear direction for subsequent improvements. Next, the targeted improvement mechanism construction unit takes specific and effective improvement measures for these hotspot feedbacks. For data quality feedback, the system ensures the accuracy and completeness of the data through a preset data cleaning process. For analysis logic feedback, the system flexibly adjusts the feature importance weights according to a preset adjustment range to optimize... The system analyzes the logic; for feedback applicable to decision-making, it strengthens constraint checks, adjusts cost-benefit calculation parameters, or increases risk aversion coefficients to ensure that decision-making strategies are closer to actual needs. Furthermore, the feedback database construction unit builds a knowledge graph based on feedback classification datasets and improvement measures. This graph not only records the relationships between feedback and improvement measures but also provides valuable experience for subsequent decision-making. Based on this feedback database, when generating target management strategies, the system first performs a self-check of the generated health level to ensure the effectiveness and feasibility of the strategy. If the generated health level is lower than a set threshold, the system activates a causal reasoning model to locate the core variables affecting strategy generation and combines counterfactual simulation methods to make counterfactual decisions. This process not only improves the accuracy and efficiency of strategy generation but also enhances the system's adaptability.

[0112] The beneficial effects of the above technical solution are as follows: By introducing a feedback learning module, the solution provided in this embodiment enables a rapid response to user needs and market changes, providing strong support for the intelligent upgrade of the data asset management platform.

[0113] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent analysis and decision support system for an artificial intelligence based data asset management platform, characterized in that, Comprise: Data asset data collection processing module, for collecting and obtaining pre-processing data asset data; Intelligent analysis module, for data asset value evaluation and data asset risk analysis on pre-processing data asset data, obtaining value evaluation results and risk analysis results; Target management strategy generation module, for generating target management strategy based on value evaluation results and risk analysis results; The target management strategy generation module includes a preliminary management strategy generation unit and a target management strategy optimization unit; The preliminary management strategy generation unit is used for generating a preliminary management strategy based on the value evaluation results and the risk analysis results; The target management strategy optimization unit is used for optimizing the preliminary management strategy by using a reinforcement learning algorithm to generate a target management strategy; It also includes a feedback learning module for optimizing the functions of the intelligent analysis module and the target management strategy generation module according to the feedback collected from users on the target management strategy; The feedback learning module includes a feedback content analysis unit, a targeted improvement mechanism construction unit and a feedback database construction unit; The feedback content analysis unit is used for multi-dimensional feedback classification of the feedback collected from users on the target management strategy, obtaining a feedback classification dataset, and generating a feedback hotspot distribution map according to the classification dataset; multi-dimensional feedback classification includes data quality, analysis logic and decision applicability; The targeted improvement mechanism construction unit is used for taking targeted improvement measures for the feedback content in the feedback hotspot distribution map, specifically: for data quality feedback, data cleaning is performed according to the set data cleaning process; for analysis logic feedback, the feature importance weight is adjusted according to the set adjustment range; For decision applicability feedback, perform strong constraint condition check, adjust cost benefit calculation parameter or increase risk avoidance coefficient; The feedback database construction unit is used for generating a knowledge graph of the feedback classification dataset relative to the improvement measures based on the feedback classification dataset and the improvement measures, and constructing a feedback database based on the knowledge graph; It also includes, based on the feedback database, in the process of generating the target management strategy based on the value evaluation results and the risk analysis results, first performing a self-check of the generation health degree of the target management strategy, if the generation health degree is less than the set health degree threshold, based on the set causal reasoning model, locate the core variable affecting the generation of the target management strategy, combine the counterfactual simulation method to make counterfactual decision, according to the counterfactual decision result, adjust the optimization parameters of the preliminary management strategy of the target management strategy generation module based on the Bayesian optimization method.

2. The intelligent analysis and decision support system of the artificial intelligence-based data asset management platform according to claim 1, wherein, Collect and obtain pre-processing data asset data, including: Based on the configured API interface, database connector and file parser, collect data asset data in multiple structural forms from the data storage; perform quality detection, standardization processing and metadata extraction on the data asset data to generate pre-processing data asset data; multiple structural forms include structured, semi-structured and unstructured.

3. The intelligent analysis and decision support system of the artificial intelligence-based data asset management platform according to claim 2, wherein, Based on the configured API interface, database connector and file parser, collect data asset data in multiple structural forms from the data storage, including: Through the configured API gateway component, the data asset information in the public data source is captured by using the web crawler component; Through the blockchain listener, the data asset transaction data on the blockchain is captured in real time; Based on the data asset information and the data asset transaction data, the data asset data is composed.

4. The intelligent analysis and decision support system of the artificial intelligence-based data asset management platform according to claim 3, wherein, The data asset data is subjected to quality detection, standardized processing and metadata extraction to generate preprocessed data asset data, including: A detection mechanism of syntax layer detection, semantic layer detection and business rule layer detection is adopted to detect the quality of the data asset data; The data asset data subjected to quality detection is subjected to standardized conversion to generate data asset data in a unified intermediate format; A deep learning model is adopted to process the data asset data in the unified intermediate format to generate descriptive metadata and structural metadata, and the metadata is extracted based on the descriptive metadata and the structural metadata.

5. The intelligent analysis and decision support system of the artificial intelligence based data asset management platform according to claim 1, wherein, The intelligent analysis module includes a feature extraction unit, a value evaluation unit and a risk analysis unit; The feature extraction unit is configured to extract data features from the preprocessed data asset data based on a text feature extractor and an image feature extractor, and to generate statistical features, frequency domain features and nonlinear features by using a time series feature constructor, and to establish a mapping relationship between the statistical features, the frequency domain features and the nonlinear features by using a cross-modal correlator; The value evaluation unit is configured to set three evaluation indexes of utility value, market value and strategic value based on the data features and the mapping relationship, to determine the weights of the evaluation indexes by using an analytic hierarchy process, to perform value score prediction by using an XGBoost model based on the evaluation indexes and the weights of the evaluation indexes, to evaluate the value of the data asset data according to the obtained value score prediction result, and to obtain a value evaluation result; The risk analysis unit is configured to perform technical vulnerability detection according to an OWASP application security evaluation standard by using a static risk analyzer, and to construct a user behavior baseline by using an LSTM network by using a dynamic behavior analyzer; The threat intelligence integration component is combined to realize real-time docking with third-party threat intelligence sources; A risk transmission analysis model is used to simulate the path of risk transmission among multiple data asset data to generate a risk analysis result.

6. The intelligent analysis and decision support system of the artificial intelligence-based data asset management platform according to claim 1, wherein, Based on the value evaluation result and the risk analysis result, a preliminary management strategy is generated, including: Based on the retrieval of historical similar cases and adaptive scenes, the scene management strategy is generated based on the value evaluation result and the risk analysis result; According to the set business rule tree, the business management strategy is generated by combining the value evaluation result and the risk analysis result to formulate the business management rules; A multi-objective genetic algorithm NSGA-II is adopted to perform multi-objective optimization on the scene management strategy and the business management strategy to generate a preliminary management strategy.

7. The intelligent analysis and decision support system of the artificial intelligence-based data asset management platform according to claim 1, wherein, A reinforcement learning algorithm is adopted to optimize the preliminary management strategy to generate a target management strategy, including: An interactive mechanism of four parallel evaluation units and a central decision maker is constructed by using a hierarchical reinforcement learning framework; the four parallel evaluation units include a feasibility evaluation unit, an economy evaluation unit, a risk evaluation unit and a sensitivity analysis unit; the feasibility evaluation unit is used for detecting resource constraints and technical limitations of the preliminary management strategy; the economy evaluation unit is used for calculating ROI and NPV indexes of the preliminary management strategy; the risk evaluation unit is used for predicting the implementation risk of the preliminary management strategy; the sensitivity analysis unit is used for calculating parameter sensitivity based on Sobol index; wherein, the calculation formula of the NPV index is: In the above formula, represents the industry benchmark return rate, represents the yield threshold sensitivity, represents the time-varying capital cost rate, represents the correction of cash flow; represents the technology maturity adjustment factor, represents the time-varying cash flow; t represents the number of investment cycles; T represents the total investment cycle; Based on the central decision maker, a multi-objective proximal policy optimization algorithm is used to optimize the evaluation results of the four parallel evaluation units to obtain a target management strategy.

8. The intelligent analysis and decision support system of the artificial intelligence based data asset management platform according to claim 1, wherein, It also includes a security protection module; the security protection module includes: Fine-grained data access is controlled by using the set attribute-based encryption component; Complete operation logs are recorded and audit reports are generated by using the behavior audit tracker; The abnormal detector is used to identify abnormal access behavior of data assets data by using the isolation forest algorithm; According to the threat level obtained by detection, the protection measures are automatically triggered; wherein, the threat level is determined based on real-time detection of statistical rules, deep detection based on machine learning and correlation detection based on graph neural network.

Images