Data outbound security assessment management system and method based on artificial intelligence technology

Abstract

The invention provides a data outbound security assessment management system and method based on an artificial intelligence technology, and the system comprises a data collection module which is used for receiving a data outbound application submitted by a user, and automatically collecting to-be-outbound data and related metadata; the risk assessment module is used for receiving the data and the metadata collected by the data acquisition module, performing risk assessment on the data collected by the data acquisition module based on a pre-trained risk assessment model, identifying potential security threats and generating a risk assessment report; and the decision support module is used for extracting and generating a data outbound security policy suggestion from a security policy library according to the risk level and the threat description in the risk assessment report. According to the invention, the evaluation efficiency and accuracy are obviously improved; closed-loop intelligent support from risk identification to risk management and control is provided for the user, the capability of coping with complex data outbound scenes is greatly enhanced, and compliance risks are reduced.

Description

Technical Field

[0001] This application relates to the field of data security technology, specifically to a data export security assessment and management system and method based on artificial intelligence technology. Background Technology

[0002] With the acceleration of global economic integration and the rapid development of digital technologies, data, as a key production factor, is increasingly flowing across borders. Enterprises inevitably need to transfer data from their home country to other countries or regions when conducting international trade, cloud collaboration, transnational R&D, and customer service.

[0003] However, the process of data leaving the country is accompanied by significant security risks, primarily including: the risk of data being stolen or tampered with during transmission; the risk of data being illegally accessed or leaked due to inadequate security measures during storage at the destination; and the risk of data being used by the destination country for purposes beyond its original authorization, violating the data sovereignty and laws and regulations of the source country. Countries worldwide are increasingly emphasizing the security supervision of data leaving the country, and have successively introduced strict laws and regulations, such as the EU's General Data Protection Regulation (GDPR), China's Personal Information Protection Law, and the Measures for Security Assessment of Data Leaving the Country, requiring security assessments to be conducted before data leaves the country.

[0004] Currently, the industry's management of data export security assessments mainly relies on the following two methods: Manual Review: This is the most traditional assessment method. The company's security experts or compliance team manually review each batch of data to be exported, considering factors such as type, content, sensitivity, and the legal environment of the destination country, based on laws, regulations, and internal checklists. This determines whether export is permitted and what security measures are required. This method heavily relies on the expertise and experience of the reviewers and has significant limitations: First, it is extremely inefficient, struggling to handle large-scale, high-frequency data export demands and easily becoming a bottleneck for business development. Second, it is highly subjective and prone to errors; manual review struggles to maintain consistent standards, and fatigue or cognitive limitations can lead to oversights and security vulnerabilities when dealing with complex or novel data scenarios. Finally, it lacks in-depth analytical capabilities, failing to perform correlation analysis on massive amounts of data and making it difficult to uncover hidden and complex potential threat patterns.

[0005] Automated approaches based on rule engines: To address the efficiency issues of manual review, some systems have adopted automated assessments based on predefined rules. The system solidifies certain compliance requirements into rules (e.g., "Personal ID numbers destined for country A must be encrypted"). While this approach improves efficiency to some extent, it lacks flexibility and intelligence. The rule base is costly to build and maintain, requiring continuous manual updates as laws, regulations, and business scenarios change; it cannot handle unknown threats, failing to identify undefined, novel, or complex combinations of security risks; and its assessment dimensions are limited, typically only allowing Boolean judgments (yes / no), making it difficult to accurately quantify and classify risks, and unable to provide decision-makers with nuanced insights.

[0006] In summary, both purely manual review and rule-based automated systems generally suffer from core shortcomings when facing increasingly complex data export scenarios, including low assessment efficiency, poor accuracy, lack of adaptive learning capabilities, and insufficient ability to identify complex risks. Therefore, there is an urgent need in this field for a technical solution that combines automated processing with intelligent analysis to achieve efficient, accurate, quantifiable, and evolving assessment and management of data export security risks. Summary of the Invention

[0007] The problem addressed by this invention is how to improve the evaluation efficiency, accuracy, and adaptive learning capabilities in data export scenarios.

[0008] To address the aforementioned issues, this invention provides a data export security assessment and management system, method, electronic device, and storage medium based on artificial intelligence technology.

[0009] In a first aspect, the present invention provides a data export security assessment and management system based on artificial intelligence technology, comprising: The data collection module is used to receive user-submitted data export applications and automatically collect the data to be exported and its related metadata. The risk assessment module is used to receive the data and metadata collected by the data acquisition module, and to perform risk assessment on the data collected by the data acquisition module based on a pre-trained risk assessment model, identify potential security threats, and generate a risk assessment report. The decision support module is used to extract and generate security policy recommendations for data export from the security policy library based on the risk level and threat description in the risk assessment report.

[0010] Optionally, the data acquisition module further includes a data preprocessing unit for cleaning, format conversion, and feature extraction of the collected data. Feature extraction includes quantization encoding of metadata, and the quantization encoding method includes one-hot encoding, numerical quantization, or rating quantization.

[0011] Optionally, the risk assessment module includes a risk assessment model, which is trained using historical data and used to identify and quantify risks in the input data features; wherein, the training process of the risk assessment model includes: Historical data is cleaned, standardized, and its features are extracted, and security risk labels are added to the data; The model is trained using preprocessed data, and the prediction error is minimized through an optimization algorithm. Use a validation dataset to evaluate the model's accuracy and generalization ability.

[0012] Optionally, the risk assessment model is a Transformer model, which includes an encoder composed of a multi-head self-attention mechanism and a feedforward neural network, used to perform in-depth analysis of data features and output risk prediction results.

[0013] Optionally, the hyperparameters of the Transformer model include: an input embedding dimension of 512, 8 heads in the multi-head self-attention mechanism, a hidden layer dimension of 2048 in the feedforward neural network, 6 encoding layers N, and a Dropout rate of 0.1; the model is trained using a cross-entropy loss function and an Adam optimizer, with the learning rate initially set to 0.001 and dynamically adjusted using a cosine annealing scheduler.

[0014] Optionally, the risk assessment report output of the risk assessment module includes a threat description, quantitative indicators, risk level, and recommended countermeasures, and is presented in the form of visual charts.

[0015] Optionally, the security policy library of the decision support module is built based on historical data outbound cases, industry best practices, and laws and regulations, and supports dynamic updates; The security policy library categorizes and stores policies based on data export scenarios, data types, and the laws and regulations of the destination country.

[0016] Optionally, the decision support module also includes a strategy conflict resolution mechanism, which selects a strategy based on risk coverage, strategy priority, or user-assisted decision-making when multiple strategies conflict.

[0017] Secondly, the present invention provides a data export security assessment and management method based on artificial intelligence technology, which, based on the system described in the first aspect, includes the following steps: Receive user-submitted data export applications and collect the data to be exported and its related metadata; The data collected by the data acquisition module is assessed based on a pre-trained risk assessment model to identify potential security threats and generate a risk assessment report. This is used to extract and generate security policy recommendations for data export from the security policy library based on the risk level and threat description in the risk assessment report.

[0018] Optionally, the risk assessment step uses a risk assessment model to perform risk assessment, and the training process of the risk assessment model includes: Collect historical data outbound records and their security assessment results, publicly available data breach incidents, and simulated data outbound scenarios to construct a training dataset; Preprocessing of training data includes cleaning, standardization, feature extraction, and data labeling; Train the risk assessment model using training data and optimize the model parameters; Use the validation dataset to validate the model performance and adjust the model hyperparameters based on the validation results; The trained model is deployed for real-time risk assessment.

[0019] Thirdly, embodiments of the present invention provide an electronic device, including a processor, a communication interface, a memory, and a bus, wherein the processor, the communication interface, and the memory communicate with each other through the bus, and the processor can call logical instructions in the memory to execute the steps of the method provided in the second aspect.

[0020] Fourthly, embodiments of the present invention provide a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the data export security assessment and management method based on artificial intelligence technology as described in the second aspect.

[0021] The beneficial effects of the data export security assessment and management system based on artificial intelligence technology of the present invention are as follows: This invention effectively overcomes the shortcomings of existing technologies, such as low efficiency and strong subjectivity, by constructing a management system that integrates automatic data collection, artificial intelligence risk assessment, and intelligent decision support. The system utilizes a machine learning model trained on massive amounts of data to achieve automated, high-precision identification and quantification of data export risks, significantly improving assessment efficiency and accuracy. Simultaneously, its decision support module can dynamically generate precise security policy recommendations based on assessment results and possesses a policy conflict resolution mechanism. This provides users with closed-loop intelligent support from risk identification to risk control, greatly enhancing the ability to cope with complex data export scenarios and reducing compliance risks. Attached Figure Description

[0022] Figure 1 This is a structural block diagram of the data export security assessment and management system based on artificial intelligence technology in an embodiment of the present invention; Figure 2This is a flowchart of the data export security assessment and management system based on artificial intelligence technology in an embodiment of the present invention; Figure 3 This is a flowchart of the data export security assessment and management method based on artificial intelligence technology in an embodiment of the present invention; Figure 4 This is a structural block diagram of the electronic device in an embodiment of the present invention. Detailed Implementation

[0023] To better understand the purpose, technical solution, and advantages of this application, the application is described and explained below in conjunction with the accompanying drawings and embodiments.

[0024] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0025] like Figure 1 and Figure 2 As shown in the figure, an embodiment of the present invention provides a data export security assessment and management system based on artificial intelligence technology, comprising: The data collection module is used to receive user-submitted data export applications and automatically collect the data to be exported and its related metadata. The data acquisition module also includes a data preprocessing unit, which is used to clean, convert the format and extract features from the collected data. Feature extraction includes quantization encoding of metadata, and the quantization encoding methods include one-hot encoding, numerical quantization or rating quantization.

[0026] Data Acquisition Module: Responsible for collecting outbound data and its related metadata. First, it receives data requests. Users submit outbound data requests through the system interface, including basic information such as data type, quantity, source, and destination. Next, it determines the data scope. Based on the user's data request, the system automatically determines the scope of data to be collected, including specific datasets, data tables, or data items. Then, it collects metadata. In addition to the data itself, the system also collects data-related metadata, such as data generation time, generator, data format, data size, and storage location. Before formal data collection, the system preprocesses the data, including data cleaning and format conversion, to ensure accuracy and consistency. Finally, the system automatically collects data from the data source through preset data interfaces or channels and transmits it to the risk assessment module for further processing.

[0027] The risk assessment module receives data and metadata collected by the data acquisition module, performs risk assessment on the data collected by the data acquisition module based on a pre-trained risk assessment model, identifies potential security threats, and generates a risk assessment report. The risk assessment module includes a risk assessment model, which is trained using historical data and used to identify and quantify risks based on the characteristics of the input data. The training process of the risk assessment model includes: Historical data is cleaned, standardized, and its features are extracted, and security risk labels are added to the data; The model is trained using preprocessed data, and the prediction error is minimized through an optimization algorithm. Use a validation dataset to evaluate the model's accuracy and generalization ability.

[0028] The risk assessment model is a Transformer model, which includes an encoder. The encoder consists of a multi-head self-attention mechanism and a feedforward neural network, used to perform in-depth analysis of data features and output risk prediction results.

[0029] The hyperparameters of the Transformer model include: an input embedding dimension of 512, 8 heads in the multi-head self-attention mechanism, a hidden layer dimension of 2048 in the feedforward neural network, 6 encoding layers N, and a Dropout rate of 0.1. The model is trained using the cross-entropy loss function and the Adam optimizer, with the learning rate initially set to 0.001 and dynamically adjusted using a cosine annealing scheduler.

[0030] The risk assessment report output from the risk assessment module includes threat descriptions, quantitative indicators, risk levels, and recommended countermeasures, presented in the form of visual charts.

[0031] Specifically, the risk assessment module, based on artificial intelligence technology, performs risk assessments on the collected data to identify potential security threats. First, data preprocessing is performed to clean the raw data, removing invalid, redundant, or erroneous data. Data standardization is then applied to ensure that the data format and data type meet the requirements of subsequent analysis. Feature extraction is performed to extract key features relevant to the risk assessment. Next, risk identification is conducted using machine learning algorithms (such as classification and clustering algorithms) to perform pattern recognition on the preprocessed data, identifying potential security threat patterns. Convolutional neural networks are used for deep analysis of complex data structures to uncover hidden security risks. Natural language processing technology is used to perform semantic analysis on text data to identify sensitive information related to data security. Then, risk quantification is implemented to quantitatively assess the identified security threats, determining their probability of occurrence and potential severity. Based on the quantification results, security threats are categorized and managed, with high-risk items prioritized. Finally, a risk assessment report is generated, integrating the results of risk identification and quantification. The report should include a detailed threat description, quantitative indicators, risk level, and recommended countermeasures. Risk assessment reports are presented in a visual manner, such as charts and graphs, to help users understand the assessment results intuitively.

[0032] It should be noted that the training process for the risk assessment model is as follows: 1. Data Preprocessing Data cleaning: Cleaning the collected raw data (data comes from historical data export records and their security assessment results, publicly available data breach events and their analysis reports, simulated data export scenarios and their corresponding security risk labels) to remove invalid, duplicate or erroneous data and ensure data quality.

[0033] Data standardization: Transforming data into a unified format and standard to facilitate subsequent analysis and processing.

[0034] Feature extraction: Based on the evaluation requirements, key features are extracted from the data, such as data type, data volume, data flow, and laws and regulations of the destination country. These features will be used as input to the machine learning model.

[0035] Data labeling: Label each piece of data with a security risk, such as low risk, medium risk, high risk, etc.

[0036] 2. Model Selection and Training Model Selection: Based on the specific needs of data export security assessment, for classification problems, Support Vector Machine (SVM) and Random Forest models are selected; for sequence data, a Transformer-based model is selected for risk assessment. The Transformer model consists of an encoder and a decoder, but in the risk assessment task of this invention, the encoder part is mainly used to extract data features and perform risk assessment. The encoder consists of multiple identical layers stacked together, each layer including two sub-layers: a multi-head self-attention mechanism and a feedforward neural network.

[0037] The system consists of: an input layer that receives preprocessed data features; hidden layers, a multi-layered neural network structure used for automatic feature extraction and combination; an output layer that outputs the security risk prediction result; and a training algorithm that uses backpropagation and gradient descent to minimize the loss function (such as cross-entropy loss) between the predicted result and the true label.

[0038] 3. Model Training: The selected model is trained using the preprocessed data. During training, the model parameters are continuously adjusted to enable the model to more accurately identify security threats in the data.

[0039] The hyperparameters are as follows: input embedding dimension: d_model = 512; number of heads in the multi-head self-attention mechanism: num_heads = 8; hidden layer dimension of the feedforward neural network: d_ff = 2048; number of encoding layers: N = 6; dropout rate: dropout = 0.1.

[0040] The Transformer model is trained using the cross-entropy loss function to optimize model parameters. The specific formula is as follows: ; in, It's a real label. It is the probability predicted by the model. It refers to the number of samples.

[0041] Model validation: The trained model is validated using an independent validation dataset, such as accuracy, recall, and F1 score, to evaluate the model's accuracy and generalization ability.

[0042] The tuning parameters are as follows: Learning rate: The Adam optimizer is used, with an initial learning rate set to lr = 0.001, which is dynamically adjusted based on the loss changes during training. Batch size: batch_size = 32; Number of training epochs: 10, with early stopping based on the performance of the validation set to avoid overfitting. Learning rate decay: A cosine annealing learning rate scheduler is used for learning rate decay.

[0043] 4. Deployment and Testing: Deploy the trained model into the data export security assessment and management system, and conduct actual testing and verification.

[0044] In addition, it is necessary to explain the specific feature dimensions and feature quantization methods used in the model input: In the data export security assessment and management system, the feature dimensions of the model input mainly include the following aspects: Data type: Describes the type of data, such as text, images, videos, structured data, etc.

[0045] Data volume: The total amount of data to be exported, usually measured in bytes, kilobytes (KB), megabytes (MB), or gigabytes (GB).

[0046] Data flow: From which source country or region does the data flow to which destination country or region?

[0047] Laws and regulations of the destination country: The legal and regulatory requirements of the destination country or region regarding data protection, such as the EU's GDPR and China's Personal Information Protection Law.

[0048] Data sensitivity: The degree of sensitivity of the data, such as whether it contains personal privacy information, trade secrets, etc.

[0049] Data transmission method: The technical means used for data transmission, such as encrypted transmission, plaintext transmission, etc.

[0050] Data access permissions: Setting access permissions for data at the destination, such as whether public access is allowed or access is restricted to specific users.

[0051] Data storage location: The location where the data is stored at its destination, such as cloud storage, local server, etc.

[0052] Data generation time: The time when the data was generated is used to assess the freshness and timeliness of the data.

[0053] Data generator: The creator or source of data, used to trace the origin of the data.

[0054] For example, the feature quantization method is as follows: Data type: One-hot encoding is used for quantization. For example, text, image, and video types are encoded as [1, 0, 0], [0, 1, 0], and [0, 0, 1], respectively.

[0055] Data size: expressed directly in numerical form, such as 1024KB, 5MB, etc.

[0056] Data flow direction: Quantify using country or region codes, such as CN for China and US for the United States. Then construct a binary vector based on the flow direction, such as [1, 0] for flow from China to the United States (assuming only two countries are considered).

[0057] The laws and regulations of the destination country: scores are quantified based on the strictness of the laws and regulations, such as GDPR being rated 5 points (the highest), and China's Personal Information Protection Law being rated 4 points, etc.

[0058] Data sensitivity: graded and quantified, such as low sensitivity is rated as 1 point, medium sensitivity as 2 points, and high sensitivity as 3 points.

[0059] Data transmission method: One-hot encoding is used for quantization, such as encrypted transmission encoding as [1, 0] and plaintext transmission encoding as [0, 1].

[0060] Data access permissions: The degree of openness of the permissions is scored and quantified, such as public access is scored as 1 point, access only to specific users is scored as 2 points, and completely restricted access is scored as 3 points.

[0061] Data storage location: The security and compliance of the storage location are scored and quantified, such as cloud storage being scored 2 points (assuming that cloud storage has certain security measures) and local server being scored 3 points (assuming that local server has stricter security controls).

[0062] Data generation time: Quantified using timestamps or time differences, such as the number of days or hours since the current time.

[0063] Data generator: Quantified using the generator's unique identifier or code, such as company name, personal ID, etc.

[0064] In addition, it should be noted that the output format of the risk assessment report should include the following key information and be presented in a structured manner for easy understanding and use by users: Threat Description: Provide a detailed description of the identified potential security threats, such as data breach risks, unauthorized access risks, etc.

[0065] Quantitative indicators include the probability of a security threat occurring (e.g., a probability value between 0 and 1) and the potential degree of harm (e.g., low, medium, high, or specific scores).

[0066] Risk Level: Based on quantitative indicators, security threats are classified into different risk levels, such as low risk, medium risk, and high risk.

[0067] Recommended countermeasures: For each identified security threat, provide specific countermeasures or recommendations, such as data anonymization, encrypted transmission, and access control restrictions.

[0068] Visual charts: To help users understand the assessment results intuitively, visual charts such as bar charts, pie charts, and line charts can be generated to show the distribution of different risk levels and the probability of security threats occurring.

[0069] The decision support module is used to extract and generate security policy recommendations for data export from the security policy library based on the risk level and threat description in the risk assessment report.

[0070] For example, the risk assessment report format is as follows: 1. Threat Description: Data breach risk; Quantitative indicator: Occurrence probability 0.3, high potential harm level; Risk level: High risk; Recommended countermeasures: De-identify sensitive data, use strong encryption algorithms for data transmission, and restrict data access permissions.

[0071] 2. Threat Description: Risk of unauthorized access; Quantitative indicators: Probability of occurrence 0.1, potential harm level medium; Risk level: Medium risk; Recommended countermeasures: Implement a strict identity verification mechanism, regularly review access logs, and promptly detect and block unauthorized access.

[0072] In this embodiment, the security policy library of the decision support module is built based on historical data outbound cases, industry best practices and laws and regulations, and supports dynamic updates; The security policy library categorizes and stores data based on the data export scenario, data type, and laws and regulations of the destination country.

[0073] The decision support module also includes a strategy conflict resolution mechanism, which selects a strategy based on risk coverage, strategy priority, or user-assisted decision-making when multiple strategies conflict.

[0074] Based on the risk assessment results (as recorded in the risk assessment report), security strategy recommendations for data export are provided. First, a risk assessment report is received from the risk assessment module. This report details the potential security risks, risk levels, and possible impact scope of the data to be exported. Next, the risk assessment report is analyzed to identify key risk points and trends. Based on the analysis results of the risk assessment report and according to a pre-set security strategy library, security strategy recommendations for the current data export request are generated. These recommendations include data anonymization, encrypted transmission, access control restrictions, and monitoring measures after data export. Finally, the optimized security strategy is presented to the user in an intuitive and easy-to-understand manner through the system interface. This includes a textual description of the strategy recommendations, implementation steps, expected effects, and the potential degree of risk reduction. User decision support functions are also provided, such as risk assessment comparison, analysis of the advantages and disadvantages of strategy recommendations, and assessment of implementation difficulty, helping users better understand and choose the most suitable security strategy for their current data export needs.

[0075] It is important to emphasize the rules for building the security policy library and the mechanism for resolving policy conflicts.

[0076] The security policy library is a core component of the decision support module. It is built upon a multi-dimensional framework, including historical data on cross-border data transfer cases, industry best practices, legal and regulatory requirements, and security threat intelligence. Its construction rules primarily include the following aspects: Categorized Construction: Based on different scenarios of data export (such as business cooperation, scientific research exchange, emergency rescue, etc.), data types (such as sensitive personal information, trade secrets, government data, etc.), and the legal and regulatory requirements of the destination country, security strategies are categorized and constructed.

[0077] Dynamic updates: Regularly collect and analyze the latest data export security incidents, legal and regulatory changes, and technological development trends, and dynamically update the security strategy library to ensure the timeliness and effectiveness of the strategies.

[0078] Expert review: Industry experts, legal counsel, and security technicians are invited to review the security strategy library to ensure the scientific validity, rationality, and operability of the strategies.

[0079] User feedback: Establish a user feedback mechanism to collect user evaluations and suggestions on security policies in practical applications, and continuously optimize the policy library.

[0080] In the decision support module, when multiple security policies conflict, an effective conflict resolution mechanism needs to be established to ensure that the ultimately selected security policy minimizes the risk of data leakage. The specific mechanism is as follows: Risk coverage priority: When multiple strategies conflict, the strategy with the highest risk coverage is selected first. Risk coverage refers to the types and number of security threats that a strategy can cover. By quantitatively evaluating the risk coverage of each strategy, the optimal strategy is selected.

[0081] Policy Priority Setting: Prioritize different types of security policies. Policies mandated by laws and regulations have the highest priority, followed by industry best practices, and finally user-defined policies. In conflict resolution, policies are selected according to priority.

[0082] User decision support: When the risk coverage and priority of conflicting strategies are similar, provide user decision support functions, such as displaying the advantages and disadvantages of each strategy, implementation difficulty, expected results, etc., to help users make choices based on the actual situation.

[0083] Automatic adjustment and optimization: Based on user selections and feedback, the strategy conflict resolution mechanism is automatically adjusted and optimized to improve its adaptability and accuracy.

[0084] For ease of understanding, the following are examples of security strategy recommendations based on different data export scenarios and risk levels: Scenario 1: Business cooperation data leaving the country; Risk level: Medium risk; Security strategy recommendations: 1. Data anonymization: Sensitive data, such as names, ID numbers, and contact information, is anonymized to reduce the risk of data leakage.

[0085] 2. Encrypted transmission: Data is encrypted during transmission using encryption protocols such as SSL / TLS to ensure data security during transmission.

[0086] 3. Access control restrictions: Authenticate and control the permissions of users accessing the data to ensure that only authorized users can access the data.

[0087] 4. Monitoring measures after data leaves the country: Establish a monitoring mechanism for data after it leaves the country, and regularly check the usage and security status of the data.

[0088] Scenario 2: Scientific research data export; Risk level: Low risk; Security strategy recommendations: 1. Data anonymization: Anonymize data involving personal privacy to ensure that personal privacy is not leaked during data exchange.

[0089] 2. Encrypted storage: Data is stored using encrypted storage technology in the destination country to prevent unauthorized access.

[0090] 3. Partner qualification verification: Verify the qualifications of partners to ensure they have the corresponding data security protection capabilities.

[0091] Scenario 3: Emergency rescue data export; Risk level: High risk; Security strategy recommendations: 1. Strict data anonymization and encryption: All sensitive data undergoes strict anonymization and encryption processes to ensure absolute data security during transmission and storage.

[0092] 2. Real-time monitoring and emergency response: Establish a real-time monitoring mechanism to monitor the entire data export process and formulate emergency response plans to deal with possible security incidents.

[0093] 3. Multi-departmental collaboration: Establish a collaborative mechanism with relevant departments (such as security departments, legal departments, etc.) to jointly address security risks during the data export process.

[0094] In this system, the modules are connected via data interfaces to ensure smooth data transmission and processing. To guarantee the reliability and efficiency of data transmission, this invention defines a unified transmission protocol and data format.

[0095] Transmission protocol: HTTP / HTTPS Protocol: The system primarily uses HTTP (Hypertext Transfer Protocol) or HTTPS (Hypertext Transfer Protocol Secure) as the basic protocol for data transmission between modules. HTTPS adds an SSL / TLS encryption layer on top of HTTP to ensure data security during transmission and prevent data theft or tampering.

[0096] RESTful API: Data interaction between modules is achieved through RESTful API. RESTful API is based on the HTTP protocol and uses standard HTTP methods (such as GET, POST, PUT, DELETE, etc.) for CRUD operations on data. It is characterized by its simplicity, extensibility, and maintainability.

[0097] Data format: JSON format: Data transmitted between modules within the system primarily uses JSON (JavaScript Object Notation) format. JSON is a lightweight data-interchange format that is easy for humans to read and write, and also easy for machines to parse and generate. It supports nested data structures, making it convenient to represent complex data relationships.

[0098] Data field definition: From the data acquisition module to the risk assessment module: The transmitted data includes basic information such as the type, quantity, source, and destination of the data to be exported, as well as metadata (such as generation time, generator, data format, data size, and storage location). This data is encapsulated in the form of JSON objects, with each field corresponding to a specific data item.

[0099] From the risk assessment module to the decision support module: the transmitted data mainly consists of risk assessment reports, including detailed threat descriptions, quantitative indicators, risk levels, and recommended countermeasures. This information is also encapsulated in JSON objects for easy parsing and processing by the decision support module.

[0100] From the decision support module to the monitoring and feedback module, the transmitted data includes generated security policy recommendations, along with textual descriptions, implementation steps, and expected results. This information is also encapsulated in JSON format to ensure data integrity and consistency.

[0101] Combination Figure 2 As shown, users submit a data application for export through the system user interface. The data collection module automatically collects the data and its related metadata. The risk assessment module uses artificial intelligence technology to analyze and evaluate the data and generate a risk assessment report. The decision support module provides security strategy suggestions for data export based on the risk assessment report. Users make decisions based on the suggestions and initiate the data export process.

[0102] like Figure 3 As shown, this embodiment of the invention also provides a data export security assessment and management method based on artificial intelligence technology, including the following steps: S1. Receive user-submitted data export applications and collect the data to be exported and its related metadata; S2. Based on the pre-trained risk assessment model, the data collected by the data acquisition module is used to conduct risk assessment, identify potential security threats, and generate a risk assessment report. S3. Based on the risk level and threat description in the risk assessment report, extract and generate security policy recommendations for data export from the security policy library.

[0103] The training process for the risk assessment model includes: Collect historical data outbound records and their security assessment results, publicly available data breach incidents, and simulated data outbound scenarios to construct a training dataset; Preprocessing of training data includes cleaning, standardization, feature extraction, and data labeling; Train the risk assessment model using training data and optimize the model parameters; Use the validation dataset to validate the model performance and adjust the model hyperparameters based on the validation results; The trained model is deployed for real-time risk assessment.

[0104] like Figure 4 As shown in the figure, an electronic device provided by an embodiment of the present invention includes: a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute the following method: Receive user-submitted data export applications and collect the data to be exported and its related metadata; The data collected by the data acquisition module is assessed based on a pre-trained risk assessment model to identify potential security threats and generate a risk assessment report. This is used to extract and generate security policy recommendations for data export from the security policy library based on the risk level and threat description in the risk assessment report.

[0105] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0106] This invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0107] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data export security assessment and management system based on artificial intelligence technology, characterized in that, include: The data collection module is used to receive user-submitted data export applications and automatically collect the data to be exported and its related metadata. The risk assessment module is used to receive the data and metadata collected by the data acquisition module, and to perform risk assessment on the data collected by the data acquisition module based on a pre-trained risk assessment model, identify potential security threats, and generate a risk assessment report. The decision support module is used to extract and generate security policy recommendations for data export from the security policy library based on the risk level and threat description in the risk assessment report.

2. The data export security assessment and management system based on artificial intelligence technology according to claim 1, characterized in that, The data acquisition module also includes a data preprocessing unit, which is used to clean, convert the format and extract features from the collected data. Feature extraction includes quantization encoding of metadata, and the quantization encoding method includes one-hot encoding, numerical quantization or rating quantization.

3. The data export security assessment and management system based on artificial intelligence technology according to claim 1, characterized in that, The risk assessment module includes a risk assessment model, which is trained using historical data and used to identify and quantify risks based on the characteristics of the input data. The training process of the risk assessment model includes: Historical data is cleaned, standardized, and its features are extracted, and security risk labels are added to the data; The model is trained using preprocessed data, and the prediction error is minimized through an optimization algorithm. Use a validation dataset to evaluate the model's accuracy and generalization ability.

4. The data export security assessment and management system based on artificial intelligence technology according to claim 3, characterized in that, The risk assessment model is a Transformer model, which includes an encoder composed of a multi-head self-attention mechanism and a feedforward neural network, used to perform in-depth analysis of data features and output risk prediction results.

5. The data export security assessment and management system based on artificial intelligence technology according to claim 4, characterized in that, The hyperparameters of the Transformer model include: an input embedding dimension of 512, 8 heads in the multi-head self-attention mechanism, a hidden layer dimension of 2048 in the feedforward neural network, 6 encoding layers N, and a Dropout rate of 0.

1. The model is trained using the cross-entropy loss function and the Adam optimizer, with the learning rate initially set to 0.001 and dynamically adjusted using a cosine annealing scheduler.

6. The data export security assessment and management system based on artificial intelligence technology according to claim 1, characterized in that, The risk assessment report output of the risk assessment module includes threat description, quantitative indicators, risk level, and recommended countermeasures, presented in the form of visual charts.

7. The data export security assessment and management system based on artificial intelligence technology according to claim 1, characterized in that, The security policy library of the decision support module is built based on historical data of outbound cases, industry best practices and laws and regulations, and supports dynamic updates; The security policy library categorizes and stores policies based on data export scenarios, data types, and the laws and regulations of the destination country.

8. The data export security assessment and management system based on artificial intelligence technology according to claim 7, characterized in that, The decision support module also includes a strategy conflict resolution mechanism, which selects a strategy based on risk coverage, strategy priority, or user-assisted decision-making when multiple strategies conflict.

9. A data export security assessment and management method based on artificial intelligence technology, characterized in that, The system according to any one of claims 1 to 8 includes the following steps: Receive user-submitted data export applications and collect the data to be exported and its related metadata; The data collected by the data acquisition module is assessed based on a pre-trained risk assessment model to identify potential security threats and generate a risk assessment report. Based on the risk level and threat description in the risk assessment report, security policy recommendations for data export are extracted from the security policy library and generated.

10. The method according to claim 9, characterized in that, The risk assessment step uses a risk assessment model to perform risk assessment, and the training process of the risk assessment model includes: Collect historical data outbound records and their security assessment results, publicly available data breach incidents, and simulated data outbound scenarios to construct a training dataset; Preprocessing of training data includes cleaning, standardization, feature extraction, and data labeling; Train the risk assessment model using training data and optimize the model parameters; Use the validation dataset to validate the model performance and adjust the model hyperparameters based on the validation results; The trained model is deployed for real-time risk assessment.

Images