Data asset desensitization arrangement and listing transaction system and method based on AI algorithm

Abstract

The invention discloses a data asset desensitization arrangement and listing transaction system and method based on an AI algorithm, and relates to the technical field of data processing and financial transaction. Aiming at the problems of poor desensitization adaptability, low data arrangement efficiency, inaccurate transaction value quantification and high risk in the prior art, the method comprises the following five core flows: data asset acquisition and preprocessing, and standardized original data set construction; performing AI driving sensitive information identification and adaptive desensitization, accurately identifying sensitive information through an AI hybrid model, dynamically matching and verifying a desensitization strategy, and outputting a compliant desensitization data set; aI intelligent arrangement and data quality optimization are carried out to realize data classification, duplicate removal, standardization and quality improvement, and high-quality transaction-level data assets are formed; aI dynamic pricing and supply and demand intelligent matching are carried out, and multi-dimensional features are fused to realize precise pricing and efficient supply and demand docking; compliance verification and transaction whole-course tracing are realized, and compliance verification, non-tampering tracing and risk monitoring are realized based on the block chain and the AI.

Description

Technical Field

[0001] This invention belongs to the field of financial transaction and data processing technology, and in particular relates to a data asset desensitization, sorting and listing trading system and method based on AI algorithm. Background Technology

[0002] Data asset desensitization, processing, and listing technologies are core technologies that enable compliant circulation and value realization of data assets by performing privacy-protected processing and structural optimization of raw data, combined with market demand. However, existing data asset desensitization, processing, and listing technologies suffer from several drawbacks. They ignore the differences in data types, privacy levels, and transaction scenarios, and the use of fixed desensitization rules leads to over-desensitization (damaging data usability) or under-desensitization (posing a risk of privacy leakage). Furthermore, they lack the ability to intelligently identify implicit sensitive information. Existing data asset desensitization, processing, and listing technologies are unable to adapt to the format differences and quality defects of multi-source heterogeneous data. The lack of intelligent classification, deduplication, and standardization mechanisms results in low data processing efficiency and low transaction value. Moreover, the absence of AI-driven precise pricing, efficient matching, and full-process traceability mechanisms leads to low transaction efficiency and high risk. Summary of the Invention

[0003] To address the above issues and overcome the shortcomings of existing technologies, this invention provides a data asset de-identification, processing, and listing system and method based on AI algorithms. Addressing the problems of existing data asset de-identification, processing, and listing technologies neglecting differences in data types, privacy levels, and transaction scenarios, employing fixed de-identification rules leading to over- or under-identification, and lacking intelligent identification capabilities for implicit sensitive information, this solution introduces an AI hybrid recognition model (integrating BERT text recognition network, CNN image feature extraction network, and time-series data LSTM analysis network) to achieve accurate identification and privacy level determination of multiple types of explicit and implicit sensitive information. An adaptive de-identification strategy engine is designed to dynamically match de-identification methods and intensity based on data type, privacy level, and transaction scenario. An AI de-identification effect verification model is introduced to perform real-time verification from both compliance and usability dimensions, ensuring that de-identified data is compliant and retains its core usability. This addresses the shortcomings of existing data asset de-identification technologies. Existing data processing and listing technologies are unable to adapt to the format differences and quality defects of multi-source heterogeneous data. The lack of intelligent classification, deduplication, and standardization mechanisms leads to low data processing efficiency and low transaction value. Furthermore, the absence of AI-driven precise pricing, efficient matching, and end-to-end traceability mechanisms results in low transaction efficiency and high risk. This solution employs an improved K-Means clustering algorithm to achieve intelligent classification and deduplication of multi-source heterogeneous data, uses the Transformer model to complete format standardization conversion, and combines a gradient boosting tree model to optimize data quality. It constructs an AI dynamic pricing model (integrating random forest regression and LSTM time series prediction network) to achieve precise pricing of data assets by incorporating multi-dimensional features. It designs an AI supply and demand matching engine (based on the Siamese network) to mine the potential needs of both supply and demand sides and improve matching efficiency. Finally, it introduces a blockchain + AI traceability mechanism to achieve full-process compliance verification and tamper-proof traceability, thereby reducing transaction risks.

[0004] The technical solution adopted by this invention is as follows: The method for data asset de-identification, organization, and listing based on AI algorithms provided by this invention includes the following steps: Step S1: Data asset collection and preprocessing; Step S2: AI-driven sensitive information identification and adaptive desensitization; Step S3: AI-powered intelligent data organization and quality optimization; Step S4: AI-driven dynamic pricing and intelligent matching of supply and demand; Step S5: Compliance verification and full transaction traceability.

[0005] Further, in step S1, the data asset acquisition and preprocessing involves obtaining multi-source heterogeneous raw data and performing preprocessing operations to obtain a standardized raw dataset. The multi-source heterogeneous raw data includes text data, image data, time-series data, structured data, etc. The preprocessing includes data format parsing, coarse noise filtering, and data standardization. Data format parsing uses an AI parsing model to identify the storage formats (JSON, CSV, XML, binary, etc.) of data from different sources and completes a unified format conversion. Coarse noise filtering uses an improved 3σ criterion combined with the isolated forest algorithm to remove abnormal data (datasets with a data missing rate exceeding 50% or values ​​exceeding a reasonable range). Data standardization uses Z-Score standardization to map numerical data to the [-1,1] range to eliminate dimensional differences.

[0006] Furthermore, in step S2, the AI-driven sensitive information identification and adaptive desensitization specifically includes: Intelligent identification of sensitive information; based on standardized raw datasets, an AI hybrid recognition model is constructed. For text data, a BERT pre-trained model is used to extract semantic features to identify explicit sensitive information such as names, ID numbers, and mobile phone numbers, as well as implicit sensitive information derived through association. For image data, a CNN network is used to extract visual features to identify sensitive image regions such as faces and license plates. For time-series data, an LSTM network is used to analyze data change patterns and identify sensitive time-series features. The model outputs the type of sensitive information and its privacy level (level 1-5, with level 1 being the lowest and level 5 being the highest). Adaptive de-identification strategy matching; constructing a de-identification strategy library (including replacement, masking, AES encryption, generalization, differential privacy, etc.), and dynamically matching the optimal de-identification strategy based on data type, privacy level, and transaction scenario (scientific research, commerce, government affairs, etc.) through an AI decision tree model; for example, ID card numbers with level 5 privacy are de-identified using AES encryption, and address information with level 3 privacy is de-identified using generalized methods (retaining the province and city, masking the district and county and detailed address). De-identification effect verification: An AI verification model is used to verify the de-identified data from two dimensions: compliance (whether it complies with laws and regulations) and usability (whether it retains core statistical features). Compliance verification is performed by calculating semantic similarity to verify whether the de-identified data is compliant, and usability verification is performed by calculating the data feature retention rate. If the compliance does not meet the standard, the de-identification strategy is readjusted. If the usability is lower than the threshold (default 80%), the de-identification parameters are optimized until the requirements of both dimensions are met, and a compliant de-identified dataset is output.

[0007] Furthermore, in step S3, the AI ​​intelligent sorting and data quality optimization specifically includes: Intelligent classification and deduplication: Based on the compliant and anonymized dataset, an improved K-Means clustering algorithm is used to construct clustering feature vectors based on data content and usage characteristics, dividing the compliant and anonymized data into multiple categories such as finance, medical care, and transportation; data fingerprints are generated through the SimHash algorithm, and the Hamming distance of the fingerprints is calculated. If the distance is ≤3, it is determined to be duplicate data and is removed. Format standardization and redundancy removal; the Transformer model is used to learn the target transaction format (based on industry standards and trading platform specifications) to complete the unified format conversion of compliant and anonymized data; the AI ​​semantic analysis model is used to identify redundant fields (invalid annotations and duplicate records without actual transaction value) to achieve intelligent removal of redundant information; Data quality optimization; constructing a data quality assessment index system (completeness, accuracy, consistency, timeliness); using a gradient boosting tree model to intelligently complete missing values ​​in compliant and anonymized data (using probability prediction for categorical data and time-series prediction for numerical data); identifying and correcting numerically abnormal data through an anomaly detection model (IsolationForest); verifying the logical consistency of data fields (such as the number of digits in an ID card number, the format of a mobile phone number, etc.) through an AI logic matching model, forming standardized, high-quality transaction-grade data assets.

[0008] Furthermore, in step S4, the AI ​​dynamic pricing and intelligent supply-demand matching specifically include: AI-driven dynamic pricing; based on standardized, high-quality transaction-grade data assets, a multi-dimensional pricing feature system is constructed (data quality score, privacy level, scarcity, application scenario value, timeliness); a random forest regression model combined with an LSTM time series prediction network is used to train a data value pricing model and output a benchmark price for data assets; the floating price is dynamically adjusted based on the market supply and demand coefficient (the ratio of the current number of demanders to the current number of suppliers) to form the final transaction quote; AI-powered intelligent supply and demand matching; collects characteristic information from both supply and demand sides (industry, purpose, and budget of the demand side; data type, pricing range, and delivery method of the supply side), constructs a 128-dimensional supply and demand feature vector; uses the Siamese network to calculate the cosine similarity of the supply and demand feature vectors as the matching degree; prioritizes recommending transaction combinations with a matching degree higher than the threshold (default 75%); supports personalized filtering functions to meet the customized needs of both supply and demand sides.

[0009] Furthermore, in step S5, the compliance verification and full transaction traceability specifically include: Secondary compliance verification: Before the transaction, the AI ​​compliance verification model is used to verify the residual rate of sensitive information (threshold ≤ 0.01%) and the regulatory compliance degree (threshold ≥ 95%) of the compliant and de-identified dataset corresponding to the standardized high-quality transaction-grade data assets. If both indicators meet the standards, the transaction is allowed to proceed. The entire transaction process is traceable; a transaction traceability system is built based on consortium blockchain technology, with nodes including data suppliers, demanders, trading platforms, and regulatory agencies; smart contracts are used to automatically execute the transaction process (after the demander makes payment, data delivery authorization is triggered, the supplier uploads the compliant and anonymized dataset to the encrypted storage node, and the demander obtains access access); key information such as anonymized processing logs, data organization records, transaction contracts, payment vouchers, and delivery records are hashed to generate summaries and then stored on the blockchain to achieve tamper-proof traceability; Post-transaction risk monitoring; employing an LSTM anomaly detection model to monitor in real time the access behavior of demanders to standardized high-quality transaction-grade data assets (access frequency, access time, data export behavior, etc.), and constructing a normal behavior feature database; when a behavior deviation of ≥30% is detected, an early warning mechanism is triggered, sending early warning information to the trading platform and the supplier, and freezing data access permissions; based on on-chain traceability records, a complete chain of evidence is provided for rights protection against violations.

[0010] The data asset desensitization, sorting, and listing trading system based on AI algorithms provided by this invention includes a data asset acquisition and preprocessing module, an AI sensitive information identification and desensitization module, an AI intelligent sorting and optimization module, an AI pricing and matching trading module, and a compliance traceability module; The data asset acquisition and preprocessing module acquires multi-source heterogeneous raw data and constructs a standardized raw dataset through data format parsing, coarse noise filtering, and standardization processing. The AI ​​sensitive information identification and desensitization module is based on a standardized original dataset. It identifies sensitive information and privacy level through an AI hybrid recognition model, matches an adaptive desensitization strategy and completes the desensitization. After effect verification, it outputs a compliant desensitized dataset. The AI ​​intelligent sorting and optimization module is based on a compliant and de-identified dataset. Through AI clustering, standardization, and quality optimization algorithms, it achieves intelligent data classification, deduplication, format unification, and quality improvement, forming standardized, high-quality transaction-grade data assets. The AI ​​pricing and matching transaction module is based on standardized, high-quality transaction-grade data assets. It outputs transaction prices through an AI dynamic pricing model and combines an AI supply and demand matching engine to complete intelligent supply and demand matching, supporting automated execution of the transaction process. The compliance traceability module, based on the output of the AI ​​pricing and matching transaction module, enables secondary compliance verification before transactions, full traceability during transactions, and risk monitoring after transactions, thereby ensuring transaction compliance and security.

[0011] The beneficial effects achieved by the present invention using the above solution are as follows: (1) In view of the problems that existing data asset desensitization and listing technologies ignore the differences in data types, privacy levels and transaction scenarios, adopt fixed desensitization rules leading to over- or under-desensitization, and lack of intelligent identification capabilities for implicit sensitive information, this solution introduces an AI hybrid recognition model to achieve accurate identification and privacy level determination of multiple types of explicit and implicit sensitive information; designs an adaptive desensitization strategy engine to dynamically match desensitization methods and intensity; and introduces an AI desensitization effect verification model to verify from the dual dimensions of compliance and usability, ensuring that desensitized data is compliant and retains core use value, thus solving the adaptability and accuracy defects of traditional desensitization methods.

[0012] (2) In view of the problems that existing data asset desensitization and listing technologies cannot adapt to the format differences and quality defects of multi-source heterogeneous data, lack intelligent classification, deduplication and standardization mechanisms, resulting in low data processing efficiency and low transaction value, and the lack of AI-driven accurate pricing, efficient matching and full traceability mechanisms, resulting in low transaction efficiency and high risk, this solution adopts an improved K-Means+Transformer+quality optimization model to achieve intelligent data classification, deduplication, standardization and quality improvement; constructs an AI dynamic pricing model to achieve accurate pricing by integrating multi-dimensional features; designs an AI supply and demand matching engine to improve matching efficiency; and introduces a blockchain+AI traceability mechanism to achieve full compliance verification and tamper-proof traceability of transactions, effectively improving data processing efficiency, transaction value and security, and reducing transaction risks. Attached Figure Description

[0013] The accompanying drawings are provided to further understand the present invention and form part of the specification. They are used together with the embodiments of the present invention to explain the invention and do not constitute a limitation thereof.

[0014] Figure 1 This is a schematic diagram of the core data asset processing flow of a data asset de-identification, sorting, and listing trading system and method based on AI algorithms proposed in this invention; Figure 2 This is a schematic diagram of the core sub-process of AI desensitization in a data asset desensitization, sorting, and listing trading system and method based on AI algorithms proposed in this invention. Figure 3 This is a schematic diagram illustrating the entire data asset trading process of a data asset desensitization, sorting, and listing trading system and method based on AI algorithms proposed in this invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0016] Example 1, see Figures 1-2 The present invention provides a method for data asset de-identification, organization, and listing based on AI algorithms. The method includes the following steps: Step S1: Data asset collection and preprocessing; acquire multi-source heterogeneous raw data, and construct a standardized raw dataset through data format parsing, coarse noise filtering, and standardization processing; Step S2: AI-driven sensitive information identification and adaptive desensitization; Based on the standardized original dataset, identify sensitive information and privacy level, match adaptive desensitization strategy and complete desensitization, and output compliant desensitized dataset after effect verification; Step S3: AI-powered intelligent data processing and quality optimization; Based on compliant and anonymized datasets, intelligent classification, deduplication, format standardization, and quality optimization are performed to form standardized, high-quality transaction-grade data assets; Step S4: AI dynamic pricing and intelligent supply and demand matching; Based on standardized, high-quality transaction-grade data assets, the transaction price is output through an AI pricing model, and intelligent matching is completed by combining supply and demand characteristics; Step S5: Compliance Verification and Full Transaction Traceability; Based on AI dynamic pricing and intelligent supply and demand matching results, complete the second compliance verification before the transaction, full traceability during the transaction, and risk monitoring after the transaction to achieve compliant and secure transactions.

[0017] Example 2, based on the above examples, in step S1, the multi-source heterogeneous raw data includes transaction flow data from the financial industry, medical case data from the medical industry, traffic condition time-series data from the transportation industry, user behavior text data from e-commerce platforms, and surveillance image data; data format parsing uses an AI format recognition model (based on CNN + attention mechanism), supporting automatic parsing of more than 10 mainstream data formats and unified JSON format conversion; coarse noise filtering uses an improved 3σ criterion (reducing the outlier judgment threshold from...). Adjusted to (Adapting to data distribution characteristics) Combined with the Isolation Forest algorithm, outlier data with missing data rates exceeding 50% and values ​​outside the reasonable range are removed; data standardization adopts the Z-Score standardization formula: ,in, This is the original data. The characteristic mean, The standard deviation of the feature is used to map the numerical data to the interval [-1,1] to eliminate the difference in units and finally construct the standardized original dataset.

[0018] Example 3, based on the above examples, in step S2, the AI-driven sensitive information identification and adaptive desensitization specifically includes: Intelligent identification of sensitive information: Based on standardized raw datasets, text data uses a BERT-base pre-trained model, freezing the parameters of the first 8 layers and fine-tuning the last 4 layers. The input text fragment length is set to 128. The Softmax activation function outputs the probability distribution of sensitive information types (ID card number, mobile phone number, name, etc.) and privacy levels, and the highest probability value is taken as the final privacy level. Image data uses a ResNet50 network, with pre-trained weights based on the ImageNet dataset. After fine-tuning, it identifies sensitive areas such as faces and license plates, and the privacy level is set to level 4 by default. Temporal data uses an LSTM network (256 hidden layer dimensions, 100 iterations) to identify sensitive temporal features (such as continuous user consumption behavior data). The privacy level is determined based on the scope of impact of data leakage. Adaptive de-identification strategy matching; constructing a de-identification strategy library, including replacement (replacing the middle 4 digits of the phone number with *), masking (masking the last 6 digits of the ID card number), AES encryption (key length 256 bits), generalization (generalizing the specific address to "Haidian District, Beijing"), differential privacy (noise intensity ε=0.1), etc.; using the C4.5 decision tree model, with data type, privacy level, and transaction scenario as input features, outputting the optimal de-identification strategy; for example, for medical case text data with privacy level 5, a combination of "AES encryption + differential privacy" de-identification strategy is used. De-identification effectiveness verification; compliance verification employs an AI-powered regulatory matching model, building a regulatory knowledge base based on the "Personal Information Protection Law" and the "Data Security Law," and verifies the compliance of de-identified data through semantic similarity calculation (cosine similarity threshold 0.85); usability verification uses data feature retention rate calculation. ,like If the encryption is reduced (e.g., encryption is changed to generalization), the desensitization process is re-executed, and a compliant desensitized dataset is finally output.

[0019] Example 4, based on the above examples, in step S3, the AI ​​intelligent sorting and data quality optimization specifically includes: Intelligent classification and deduplication: Based on a compliant and anonymized dataset, an improved K-Means clustering algorithm is used. Initial cluster centers are selected using a density peak algorithm to avoid local optima. Cluster feature vectors include data field types, semantic keywords, and usage labels. The number of clusters is determined by the elbow rule (the point where the sum of squared errors within a cluster drops sharply). Deduplication uses the SimHash algorithm to generate 64-bit data fingerprints. The Hamming distance of the fingerprint is calculated, and if the distance is ≤3, it is considered duplicate data and is removed. Format standardization and redundancy removal: The Transformer model (6 encoder layers and 8 long attention heads) is used to learn the standard data format of the trading platform (field names, data types, precision requirements, etc.) and convert compliant and anonymized data into a unified format; Redundancy removal uses the BERT semantic analysis model to calculate the semantic importance score of fields and remove redundant fields (meaningless labeled fields and duplicate statistical fields) with a score lower than 0.3. Data quality optimization; data integrity completion uses the Gradient Boosting Tree (XGBoost) model, inputting the relevant field features of compliant and anonymized data to predict missing values; for example, completing missing spending power tags based on user age and occupation characteristics; data accuracy correction uses the Isolation Forest algorithm (outlier threshold of 0.9) to identify numerical anomalies in compliant and anonymized data (transaction amounts exceeding reasonable ranges, etc.), and corrects them to reasonable values ​​based on historical data trends; data consistency verification uses an AI logical matching model to ensure logical consistency across different fields of the same compliant and anonymized data asset (number of digits in ID card number, mobile phone number format, etc.), ultimately forming standardized, high-quality transaction-grade data assets.

[0020] Example 5, based on the above examples, in step S4, the AI ​​dynamic pricing and intelligent supply-demand matching specifically includes: AI-driven dynamic pricing; based on standardized, high-quality transaction-grade data assets, a pricing feature system is constructed, including data quality score (weight 0.3), privacy level (weight 0.25), scarcity (inverse of industry coverage, weight 0.2), application scenario value (weight 0.15), and timeliness (inverse of the number of days since data generation, weight 0.1); a random forest regression model (100 decision trees, maximum depth 15) combined with an LSTM temporal network (5 input feature dimensions, 1 output dimension) is used to train the pricing model based on historical transaction data, outputting a benchmark price. Based on market supply and demand coefficients Adjust the floating price, final transaction price: ,in The value range is [0.5, 2]; AI-powered intelligent supply and demand matching; collecting demand-side characteristics (industry, purpose, budget range, data type requirements) and supply-side characteristics (data type, pricing range, delivery method, update frequency) to construct a 128-dimensional supply and demand feature vector; using a Siamese network (a two-branch CNN structure with shared weights) to calculate the cosine similarity of the supply and demand feature vectors as the matching degree. ;when When the data is in use, it is included in the recommendation list and pushed to both supply and demand sides in descending order of matching degree; it supports personalized filtering function, and demanders can set filtering conditions such as budget limit, data update frequency, etc., and the system adjusts the recommendation results in real time. Example 6, based on the above examples, in step S5, the compliance verification and full transaction traceability specifically include: Secondary compliance verification: Based on AI dynamic pricing and intelligent supply and demand matching results, before the transaction, the AI ​​compliance verification model verifies again the residual rate of sensitive information (threshold ≤ 0.01%) and regulatory compliance (threshold ≥ 95%) of the compliant de-identified dataset corresponding to the standardized high-quality transaction-level data assets. If both indicators meet the standards, the transaction is allowed; otherwise, it is rejected and the direction of de-identification optimization is suggested. The entire transaction process is traceable. A transaction traceability system is built based on consortium blockchain technology, with nodes including data suppliers, demanders, trading platforms, and regulatory agencies. Smart contracts (developed using Solidity) are used to automatically execute the transaction process: after the demander makes payment, the smart contract triggers data delivery authorization; the supplier uploads a standardized, high-quality, compliant, anonymized dataset corresponding to the transaction-grade data assets to the encrypted storage node, and the demander obtains access permissions; key information such as anonymized processing logs, data processing records, transaction contracts, payment vouchers, and delivery records are hashed and stored on the blockchain, ensuring immutability. Post-transaction risk monitoring; employing an LSTM anomaly detection model to monitor in real time the access behavior of demanders to standardized high-quality transaction-grade data assets (access frequency, access time, data export behavior, etc.), and constructing a normal behavior feature database; when a behavior deviation of ≥30% is detected (such as a large amount of data being exported outside of working hours), an early warning mechanism is triggered, sending early warning information to the trading platform and the supplier, and freezing data access permissions; based on on-chain traceability records, a complete chain of evidence is provided for rights protection against violations.

[0021] Example 7, based on the above examples, provides an AI-based data asset desensitization, processing, and listing system, which includes a data asset acquisition and preprocessing module, an AI sensitive information identification and desensitization module, an AI intelligent processing and optimization module, an AI pricing and matching trading module, and a compliance traceability module. The data asset acquisition and preprocessing module acquires multi-source heterogeneous raw data through API interfaces, database synchronization, file uploads, etc., performs data format parsing, coarse noise filtering, and standardization processing, and outputs a standardized raw dataset. The AI ​​sensitive information identification and desensitization module is based on the standardized raw dataset output by the data asset collection and preprocessing module. It identifies sensitive information and privacy level through an AI hybrid recognition model, matches an adaptive desensitization strategy and completes the desensitization. After effect verification, it outputs a compliant desensitized dataset. The AI ​​intelligent sorting and optimization module is based on the compliant de-identified dataset output by the AI ​​sensitive information identification and de-identification module. Through AI clustering, standardization, and quality optimization algorithms, it achieves intelligent data classification, deduplication, format unification, and quality improvement, forming standardized, high-quality transaction-grade data assets. The AI ​​pricing and matching transaction module is based on standardized, high-quality transaction-grade data assets output by the AI ​​intelligent sorting and optimization module. It outputs transaction prices through an AI dynamic pricing model and combines with an AI supply and demand matching engine to complete intelligent supply and demand matching, supporting automated execution of the transaction process. The compliance traceability module, based on the pricing and matching results output by the AI ​​pricing and matching transaction module, enables secondary compliance verification before transactions, full traceability during transactions, and risk monitoring after transactions, thereby ensuring transaction compliance and security.

[0022] It should be noted that, in this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, system, 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, system, or apparatus.

[0023] 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 method for data asset de-identification, organization, and listing based on AI algorithms, characterized in that: The method includes the following steps: Step S1: Data asset collection and preprocessing; acquire multi-source heterogeneous raw data and construct a standardized raw dataset; Step S2: AI-driven sensitive information identification and adaptive desensitization; Based on the standardized original dataset, complete sensitive information identification, desensitization strategy matching and effect verification, and output a compliant desensitized dataset; Step S3: AI-powered intelligent organization and data quality optimization; Based on the compliant and anonymized dataset, classification, deduplication, standardization, and quality optimization are performed to form standardized, high-quality transaction-grade data assets; Step S4: AI dynamic pricing and intelligent matching of supply and demand; Based on standardized, high-quality transaction-grade data assets, realize data asset pricing and intelligent matching of supply and demand; Step S5: Compliance verification and full transaction traceability; Based on pricing and matching results, complete secondary compliance verification, transaction traceability and risk monitoring to achieve compliant transactions.

2. The method for data asset de-identification, organization, and listing based on AI algorithms according to claim 1, characterized in that: In step S1, the data asset collection and preprocessing specifically includes: Obtain heterogeneous raw data from multiple sources, perform data format parsing, coarse noise filtering, and data standardization to obtain a standardized raw dataset; The multi-source heterogeneous raw data includes text data, image data, time-series data, and structured data; The preprocessing includes data format parsing, coarse noise filtering, and data standardization.

3. The method for data asset de-identification, organization, and listing based on AI algorithms according to claim 2, characterized in that: In step S2, the AI-driven sensitive information identification and adaptive desensitization specifically includes: Intelligent identification of sensitive information; constructing an AI hybrid identification model to identify explicit sensitive information, implicit sensitive information, sensitive image regions and sensitive temporal features, and outputting the type of sensitive information and its privacy level; Adaptive de-identification strategy matching; construct a de-identification strategy library, and dynamically match the optimal de-identification strategy based on data type, privacy level and transaction scenario; De-identification effectiveness verification: Verify the de-identified data from both compliance and usability perspectives, and output a compliant de-identified dataset.

4. The method for data asset de-identification, organization, and listing based on AI algorithms according to claim 3, characterized in that: In step S3, the AI ​​intelligent sorting and data quality optimization specifically includes: Intelligent classification and deduplication; based on compliant and anonymized datasets, clustering algorithms are used to classify data, and duplicate data is removed through data fingerprinting; Format standardization and redundancy removal; complete the unified format conversion of compliant and anonymized data, and identify and remove redundant fields; Data quality optimization: missing values ​​in compliant and anonymized data are filled in, outliers are corrected, and the logical consistency of data fields is verified to form standardized, high-quality transaction-grade data assets.

5. The method for data asset de-identification, organization, and listing based on AI algorithms according to claim 4, characterized in that: In step S4, the AI ​​dynamic pricing and intelligent supply-demand matching specifically include: AI-driven dynamic pricing; constructing a multi-dimensional pricing feature system, training a data value pricing model based on an AI model, outputting a benchmark price, and adjusting it in conjunction with market supply and demand to form the final transaction price; AI-powered intelligent supply and demand matching; collects feature information from both supply and demand sides, constructs supply and demand feature vectors, calculates the supply and demand matching degree, recommends transaction combinations, and supports personalized filtering.

6. The method for data asset de-identification, organization, and listing based on AI algorithms according to claim 5, characterized in that: In step S5, the compliance verification and full transaction traceability specifically include: Secondary compliance verification; pre-transaction verification of the residual rate of sensitive information and regulatory compliance of the compliant de-identified dataset corresponding to standardized high-quality transaction-grade data assets; Full transaction traceability; a transaction traceability system is built based on blockchain technology, which executes the transaction process through smart contracts and stores key information on the blockchain; Post-trade risk monitoring; real-time monitoring of data access behavior, construction of a normal behavior characteristic database, triggering abnormal alerts and freezing access permissions.

7. A data asset de-identification, processing, and listing system based on AI algorithms, used to implement the data asset de-identification, processing, and listing method based on AI algorithms as described in any one of claims 1-6, characterized in that: It includes a data asset collection and preprocessing module, an AI sensitive information identification and desensitization module, an AI intelligent sorting and optimization module, an AI pricing and matching transaction module, and a compliance traceability module; The data asset acquisition and preprocessing module acquires multi-source heterogeneous raw data and constructs a standardized raw dataset through format parsing, coarse noise filtering, and standardization processing. The AI-based sensitive information identification and desensitization module, based on a standardized raw dataset, completes sensitive information identification, desensitization strategy matching, and effect verification, and outputs a compliant desensitized dataset. The AI ​​intelligent sorting and optimization module, based on compliant and de-identified datasets, achieves data classification, deduplication, standardization, and quality optimization, forming standardized, high-quality transaction-grade data assets. The AI ​​pricing and matching transaction module is based on standardized, high-quality transaction-grade data assets, and completes dynamic pricing and intelligent matching of supply and demand for data assets. The compliance traceability module, based on pricing and matching results, enables secondary compliance verification, full transaction traceability, and post-transaction risk monitoring.

Images