Self-adaptive cross-border scene data grading and classifying method

Abstract

The invention provides a self-adaptive cross-border scene data grading and classification method, which belongs to the technical field of data grading and classification, and comprises the following steps: S1, carrying out hierarchical grading according to the association depth of data in a privacy association model to form a detailed privacy data grading system; s2, constructing a privacy association model by using a graph chain, and establishing relationships between data items and laws and regulations and enterprise strategies; s3, designing a data attribute privacy management model including a classification dimension and a grading dimension, and realizing accurate and intelligent grading of cross-border data privacy; s4, constructing an expandable cross-border scene portrait model, and depicting the full view of a cross-border business scene by using a data modeling means; s5, providing a cross-border scene dynamic expansion technology and a cross-border scene dynamic mapping technology so as to ensure the dynamic and precise cross-border scene portrait among the normal domains; and S6, forming a privacy scene dynamic association method, and training an association relationship between the tag model and the dynamic association model.

Description

Technical Field

[0001] The present invention belongs to the technical field of data grading and classification, and specifically relates to a data grading and classification method for adaptive cross-border scenarios. Background Art

[0002] Accurate data classification is the key to personal information security and the cornerstone of data rational use and data security governance. Domestic and foreign researchers use natural language processing (NLP), machine learning (ML) and other technologies to automatically identify and classify data, including: (1) Natural language processing (NLP) to identify sensitive information. Domestic and foreign researchers use NLP technology to parse and manage text data, identify and classify sensitive information, such as personal identity information (PII), medical records and financial details. Mehdy and other scholars

[13] developed a model that combines linguistic principles with artificial neural networks, which significantly improved the ability to identify privacy-leaking information in natural language text. At the same time, Verma and other researchers introduced NLP technology into the field of email security, using text classification technology to successfully identify potential phishing attacks and strengthen the security line of digital communications. (2) Machine learning (ML) predicts the type of data. Domestic and foreign researchers use machine learning (ML) algorithms to learn patterns from data and make predictions, which are used to identify sensitive information in data and perform semantic understanding. For example, the C2aaS model proposed by Munwar Ali, through data classification and access control technology, realizes differentiated protection measures for data of different security levels, effectively improving data security. Beijing Big Data Center classifies and grades data according to different dimensions such as data subject, application scenario, sensitive attributes and importance, and verifies the feasibility of the scheme in a company's express delivery business case. The SPEDAC resource proposed by Gambarelli G et al., through automated classification technology, realizes efficient identification and classification of sensitive personal data, effectively improving the accuracy and efficiency of data privacy protection. In the field of data security analysis, Li Z et al. conducted a comprehensive evaluation of the Mal-API-2019 dataset, using a variety of machine learning methods for malware detection, and verified the effectiveness and reliability of the dataset in malware analysis and detection research.

[0003] The above research not only promotes the innovation of data privacy classification theory, but also provides valuable reference for practical applications, laying an important technical foundation for the data classification research of the present invention.

[0004] Domestic and foreign research on adaptive management technology mainly includes the implementation of adaptive hierarchical dynamic risk assessment, adaptive hierarchical policy generation algorithm and dynamic data hierarchical management framework and tools: (1) Adaptive risk data security protection technology. Domestic and foreign researchers have proposed a variety of adaptive risk data protection models and methods, such as dynamic risk assessment models based on machine learning and risk prediction methods based on Bayesian networks. These methods can dynamically adjust data protection measures according to the actual use and risk level of the data, thereby achieving targeted security protection. (2) Adaptive risk data protection generation algorithm. Domestic and foreign researchers have proposed a variety of adaptive policy generation algorithms, such as policy generation methods based on genetic algorithms and policy optimization methods based on reinforcement learning

[23] . These methods can automatically generate corresponding data protection policies and algorithms based on the risk assessment results of the data to ensure the security and privacy of the data in different scenarios. (3) Dynamic data hierarchical management framework and tools. Domestic and foreign researchers have proposed a variety of hierarchical management frameworks and tools, such as role-based hierarchical management systems and trust-based hierarchical management models. These tools can dynamically adjust data access rights and protection measures according to the sensitivity and usage scenarios of the data, thereby achieving refined management of the data.

[0005] Although domestic and foreign regulations and standards for data classification have been issued, and significant progress has been made in data classification technology and adaptive association technology, the following deficiencies still exist in the hierarchical management of cross-border data privacy security:

[0006] (1) Data classification and grading regulations are only valid for the current jurisdiction. In cross-border business scenarios, data will flow between multiple jurisdictions. Due to differences in laws, regulations, culture, and standards among countries, cross-border data privacy grading management requires coordination of relevant laws and regulations in multiple jurisdictions.

[0007] (2) The granularity of data classification needs to be improved. Domestic and foreign data classification technology only stays at the data level, but data is related to each other. Privacy information can be obtained through data mining and analysis. The granularity of cross-border data classification needs to be improved to the data combination level. Summary of the invention

[0008] The purpose of the present invention is to provide a data grading and classification method that is adaptive to cross-border scenarios, aiming to solve the problem in the prior art that data grading and classification regulations are only effective for the current jurisdiction.

[0009] To achieve the above object, the present invention provides the following technical solution: a data grading and classification method for adaptive cross-border scenarios, comprising the following steps:

[0010] S1. Perform hierarchical classification based on the correlation depth of data in the privacy association model to form a detailed privacy data classification system;

[0011] S2. Use graph chains to build privacy association models and establish relationships between data items, regulations, and corporate policies to achieve dynamic updates of cross-border privacy classification;

[0012] S3. Design a data attribute privacy management model, including classification and grading dimensions, to achieve accurate and intelligent grading of cross-border data privacy;

[0013] S4. Build a scalable cross-border scenario portrait model and use data modeling to depict the full picture of cross-border business scenarios;

[0014] S5. Provide cross-border scenario dynamic expansion technology and cross-border scenario dynamic mapping technology to ensure the dynamic and accurate cross-border scenario portraits between jurisdictions;

[0015] S6. Form a dynamic association method for privacy scenarios, and form a dynamic association model between cross-border business scenarios and privacy grading models by training the association relationship between the label model and the dynamic association model.

[0016] In order to make the data grading and classification method for adaptive cross-border scenarios achieve a more refined effect of data privacy management, as a preferred embodiment of the present invention, the hierarchical grading includes:

[0017] Level 1 privacy data: information directly related to privacy, which is the most sensitive;

[0018] Secondary privacy data: including ID number, social security number, driver's license number, ID photo, etc. Although this type of data does not directly identify the identity, it is closely related to the personal identity;

[0019] Subsequent privacy levels: arranged in descending order based on how closely they are linked to a person's identity.

[0020] The data privacy association classification model includes: Class A privacy, Class B privacy, Class C privacy, and Class D privacy.

[0021] In order to enable this adaptive data grading and classification method for cross-border scenarios to more accurately identify and manage data related to specific regulations or corporate policies, as a preferred embodiment of the present invention, the Class A privacy: data items associated with regulations; Class B privacy: data items associated with corporate policies; Class C privacy: data items associated with personal privacy; Class D privacy: data items associated with privacy data.

[0022] In order to make this data grading and classification method for adaptive cross-border scenarios achieve clear classification and grading standards and realize the systematization and standardization of data privacy management, as a preferred embodiment of the present invention, the data attribute privacy management model includes classification dimension and grading dimension, and divides data attributes into four categories according to their associated objects - laws and regulations, enterprises, individuals, and privacy data. According to the closeness of data association with privacy, it is subdivided into N levels to form a clear hierarchy of privacy levels.

[0023] In order to enable the data grading and classification method for adaptive cross-border scenarios to provide a clear structured view for cross-border business scenarios, and facilitate the understanding and management of the complexity of cross-border data, as a preferred embodiment of the present invention, the cross-border scenario static portrait model includes:

[0024] Core data model: covers key entities such as individuals, legal terms and organizations;

[0025] Business domain data model: involving business details such as social services and international trade;

[0026] Cross-border business code set: contains codes for specific matters such as network and vehicle identification.

[0027] In order to enable the adaptive cross-border scenario data classification method to achieve the effect that the cross-border scenario portrait can be updated in time with the changes in the business and maintain the effectiveness and applicability of the model, the cross-border scenario dynamic portrait technology includes:

[0028] Cross-border scenario dynamic expansion technology: supports the addition, deletion and modification of dynamic elements in cross-border scenarios;

[0029] Cross-border scenario dynamic mapping technology: supports the mapping of dynamic elements of cross-border scenarios between legal jurisdictions.

[0030] In order to enable the adaptive data grading and classification method for cross-border scenarios to achieve dynamic association between cross-border business scenarios and privacy grading, and improve the flexibility and response speed of privacy management, as a preferred embodiment of the present invention, the privacy scenario dynamic association method also includes: (1) forming a basic data set of dynamic portraits of cross-border data business scenarios; (2) forming a basic data set of cross-border data privacy precise grading model; (3) forming a basic matching data set of cross-border business scenarios and privacy grading; (4) forming a cross-border scenario label data set; (5) a cross-border scenario-data association model based on label matching.

[0031] In order to enable the data grading and classification method for adaptive cross-border scenarios to achieve the system's ability to adaptively adjust data grading and classification to cope with the ever-changing cross-border scenario effects, as a preferred embodiment of the present invention, the dynamic association adaptive model includes: creating an adaptive scenario matching label model and using a rough set grading and classification method based on inter-domain distance metric learning to predict and calculate new cross-border data grading and classification.

[0032] In order to enable the data grading and classification method for adaptive cross-border scenarios to achieve the effect of expanding the application scope of the method, as a preferred embodiment of the present invention, the data grading and classification method is used to: (1) support personal cross-border data grading privacy computing processing; (2) help enterprises and institutions in cross-border data grading practices; (3) provide reference for regulatory agencies to formulate policies and standards.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. This adaptive cross-border scenario data classification method supports the hierarchical privacy computing and processing of personal cross-border data. It can support hierarchical processing of relevant tracks, personal identities, phone calls and other privacy data in different scenarios such as customs clearance, refueling, shopping, etc. according to different privacy requirements of users. Achieve a balance between privacy security and cross-border services; help enterprises and institutions in cross-border data classification practices, help enterprises accurately identify compliance points and potential risks in the cross-border data process, customize highly adaptable data classification protection strategies, and help institutions effectively manage cross-border data resources, optimize data flow architecture, and improve business efficiency and competitiveness.

[0035] 2. The data classification method for adaptive cross-border scenarios provides a reference for regulatory agencies to formulate policies and standards. The cross-border data classification of the present invention helps regulatory agencies to conduct compliance testing and risk assessment. By accurately analyzing the complexity and risk points of cross-border data flows, it provides decision-making support tools for the supervision of cross-border business of individuals and institutions, improves the effectiveness and foresight of regulatory measures, and ensures that while protecting personal privacy and maintaining national security, it stimulates the vitality and innovation potential of the digital economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0037] Figure 1 It is the overall technical roadmap of the present invention;

[0038] Figure 2 A schematic diagram of the relationship between data items and personal identities in the present invention;

[0039] Figure 3This is the cross-border data-related privacy classification diagram in the present invention;

[0040] Figure 4 A diagram showing the relationship between the data items in the present invention and the Macau Personal Privacy Protection Act;

[0041] Figure 5 This is a cross-border data-related privacy classification diagram in the present invention;

[0042] Figure 6 It is a hierarchical authorization diagram of the data attribute privacy management model in the present invention;

[0043] Figure 7 A diagram of a privacy scenario dynamic association method and an adaptive data grading and classification method of the present invention;

[0044] Figure 8 A privacy scenario dynamic association method of the present invention;

[0045] Fig. 9 The dynamic association adaptive method of the present invention;

[0046] Fig.10 This is a combined data item table for confirming the personal identity of residents in the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Example

[0049] See also Figure 1-Figure 10 , the present invention provides the following technical solution: a data grading and classification method for adaptive cross-border scenarios, comprising the following steps:

[0050] S1. Perform hierarchical classification based on the correlation depth of data in the privacy association model to form a detailed privacy data classification system;

[0051] S2. Use graph chains to build privacy association models and establish relationships between data items, regulations, and corporate policies to achieve dynamic updates of cross-border privacy classification;

[0052] S3. Design a data attribute privacy management model, including classification and grading dimensions, to achieve accurate and intelligent grading of cross-border data privacy;

[0053] S4. Build a scalable cross-border scenario portrait model and use data modeling to depict the full picture of cross-border business scenarios;

[0054] S5. Provide cross-border scenario dynamic expansion technology and cross-border scenario dynamic mapping technology to ensure the dynamic and accurate cross-border scenario portraits between jurisdictions;

[0055] S6. Form a dynamic association method for privacy scenarios, and form a dynamic association model between cross-border business scenarios and privacy grading models by training the association relationship between the label model and the dynamic association model.

[0056] In this embodiment, the hierarchical classification includes:

[0057] Level 1 privacy data: information directly related to privacy, which is the most sensitive;

[0058] Secondary privacy data: including ID number, social security number, driver's license number, ID photo, etc. Although this type of data does not directly identify the identity, it is closely related to the personal identity;

[0059] Subsequent privacy levels: arranged in descending order based on how closely they are linked to a person's identity.

[0060] In this embodiment: the data privacy association classification model includes: Class A privacy, Class B privacy, Class C privacy, and Class D privacy.

[0061] In this embodiment: the Class A privacy: data items related to laws and regulations; Class B privacy: data items related to corporate policies; Class C privacy: data items related to personal privacy; Class D privacy: data items related to privacy data.

[0062] In this embodiment: the data attribute privacy management model includes a classification dimension and a hierarchical dimension, which divides data attributes into four categories according to their associated objects - laws and regulations, enterprises, individuals, and privacy data, and subdivides them into N levels based on the closeness of the data association with privacy, forming a clearly hierarchical privacy level.

[0063] In this embodiment: the cross-border scene static portrait model includes:

[0064] Core data model: covers key entities such as individuals, legal terms and organizations;

[0065] Business domain data model: involving business details such as social services and international trade;

[0066] Cross-border business code set: contains codes for specific matters such as network and vehicle identification.

[0067] In this embodiment: the cross-border scene dynamic portrait technology includes:

[0068] Cross-border scenario dynamic expansion technology: supports the addition, deletion and modification of dynamic elements in cross-border scenarios;

[0069] Cross-border scenario dynamic mapping technology: supports the mapping of dynamic elements of cross-border scenarios between legal jurisdictions.

[0070] In this embodiment: the privacy scenario dynamic association method also includes: (1) forming a basic data set of dynamic portraits of cross-border data business scenarios; (2) forming a basic data set of cross-border data privacy precise classification model; (3) forming a basic matching data set of cross-border business scenarios and privacy classification; (4) forming a cross-border scenario label data set; (5) a cross-border scenario-data association model based on label matching.

[0071] In this embodiment: the dynamic association adaptive model includes: creating a model of adaptive scene matching labels and using a rough set hierarchical classification method of inter-domain distance metric learning to predict and calculate new cross-border data hierarchical classification.

[0072] In this embodiment: the data classification and categorization method is used to: (1) support the classification and privacy computing processing of personal cross-border data; (2) assist enterprises and institutions in cross-border data classification practices; and (3) provide a reference for regulatory agencies to formulate policies and standards.

[0073] In this embodiment, Figure 1 As shown, the overall technical solution of the present invention mainly includes the following 7 parts:

[0074] 1. Data attribute association classification model

[0075] The core goal of the data association model is to use graph chains to accurately describe the relationship between data and privacy. This study uses a directed acyclic graph (DAG) architecture to map the intrinsic associations between cross-border data. Among them, if data item A is sufficient to lock B's identity, a graph chain from A to B is constructed, leading to the privacy data endpoint defined by cross-border laws. Taking the Macau "Personal Data Protection Law" as an example, the law explicitly stipulates that any information related to a specific or identifiable individual, including sound and image records, is considered protected personal data. Therefore, this study starts from personal identity information and draws the following Figure 5 The data network diagram shown.

[0076] like Figure 2 As shown, confirmation of personal identity can be achieved through a variety of information combinations, the specific combinations are shown in Table 10. Any combination mentioned in the table is capable of uniquely identifying a resident, which means that the use of these combinations may violate the Macau Personal Privacy Protection Law. When dealing with cross-border data classification, if the data set contains information combinations such as "name and address", then these data can be directly linked to personal identity; but if there is only a single piece of information "name", it is not enough to directly link to personal identity information.

[0077] This model can accurately distinguish data content that is closely related to personal privacy. It not only highlights the correlation between data points, but also reveals the depth of privacy that may be leaked when seemingly unrelated information is combined. It ensures that the cross-border data processing process at the fine-grained level of "relationships between data" complies with legal regulations and effectively maintains the security of personal information, thereby providing strong support for cross-border data management and privacy protection.

[0078] In this embodiment, Figure 3 As shown, in order to solve the privacy problem of data-data association, the present invention proposes to perform hierarchical classification based on the association depth of data in the privacy association model. Specifically, as shown below:

[0079] Level 1 privacy data: information directly related to privacy, which is the most sensitive;

[0080] Secondary privacy data: including ID number, social security number, driver's license number, ID photo, etc. Although this type of data does not directly identify the identity, it is closely related to the personal identity;

[0081] Subsequent privacy levels: Arranged in descending order according to the degree of association with personal identity, forming a detailed privacy data classification system.

[0082] This hierarchical approach helps to more finely control and protect personal privacy, ensuring that different levels of data receive corresponding levels of security measures.

[0083] 2. Data Privacy Association Classification Model

[0084] An effective cross-border data privacy classification strategy must be dynamically adaptable and can be adjusted in time with changes in laws, regulations, corporate policies and other factors to maintain its relevance and effectiveness. This invention uses graph chains to build a privacy association model, establish the relationship between data items and laws, and corporate policies, and dynamically update the corresponding cross-border privacy classification when data and laws change. Figure 4 As shown, the Macao Personal Privacy Protection Law establishes a relationship with personal identity information. When the legal text is updated, the corresponding links in the graph chain will be changed in a timely manner to ensure that the classification logic is seamlessly connected with the latest legal requirements, thereby achieving accuracy and timeliness in cross-border data privacy management.

[0085] The analysis shows that cross-border data privacy is related to four dimensions: regulations, enterprises, individuals, and privacy data. Figure 5 As shown, the present invention divides cross-border data into the following four categories:

[0086] Category A privacy: data items related to regulations;

[0087] Category B privacy: data items related to corporate policies;

[0088] Category C privacy: data items related to personal privacy;

[0089] Category D privacy: data items associated with private data.

[0090] In this embodiment, 3. Data privacy association classification model:

[0091] In combination with the above classification and grading model, the present invention designs a data attribute privacy management model such as Figure 6 As shown, it contains two dimensions:

[0092] Classification dimension: Data attributes are divided into four categories according to the objects they are associated with - regulations, enterprises, individuals, and privacy data.

[0093] Hierarchical dimension: Based on the closeness of data-related privacy, it is subdivided into N levels to form a clearly defined privacy level.

[0094] This design realizes accurate and intelligent classification of cross-border data privacy to meet the specific needs of various entities in cross-border scenarios, and implements authorized access control with precise classification and classification.

[0095] In this embodiment, 4. Cross-border scene static portrait model

[0096] This section mainly uses data modeling to depict the overall picture of cross-border business scenarios, in order to accurately grasp the privacy needs of cross-border business, and provide digital model support for subsequent cross-border scenario dynamic profiling technology, including the following contents:

[0097] Build an extensible cross-border scenario portrait model. Build an extensible three-layer cross-border scenario digital model radiating from the center: the first is the core data model, covering key entities such as individuals, legal terms and organizations; the second is the business field data model, involving business details such as social services and international trade; the third is the cross-border business code set, including specific codes such as network and vehicle identification.

[0098] Construct objects in the cross-border scenario portrait model. Use graph chain and knowledge graph technology to depict the relationship network and its strength between regulations, subjects and data objects in cross-border scenarios.

[0099] In this embodiment, 5. Cross-border scene dynamic portrait technology

[0100] This part is mainly responsible for accurately identifying and refining the dynamic factors of cross-border scenarios, and adjusting the cross-border scenario portrait model using dynamic data modeling technology, including:

[0101] Cross-border scenario dynamic expansion technology supports the addition, deletion and modification of dynamic elements in cross-border scenarios, and provides four types of cross-border scenario dynamic expansion technologies for objects in cross-border scenarios: adding attributes, deleting attributes, adding associations, and replacing associations;

[0102] Cross-border scenario dynamic mapping technology supports the mapping of dynamic elements of cross-border scenarios between legal domains. According to the similarity of dynamic factors in different legal and regulatory regions, it provides four types of cross-border scenario dynamic mapping technologies: precise matching, approximate matching, associated matching and all-inclusive matching, to ensure the dynamic and precise cross-border scenario portraits between legal domains.

[0103] In this embodiment, 6. Privacy scenario dynamic association method

[0104] like Figure 7 As shown, the present invention forms a cross-border data business scenario dynamic portrait model and a cross-border data privacy accurate classification model in the aforementioned stage, and after importing the basic data set, a basic dynamic association model is formed. By labeling the cross-border data business scenario dynamic portrait model of the basic data set, the association relationship between the label model and the dynamic association model is trained, thereby forming a dynamic association model between the cross-border business scenario and the privacy classification model. When a new cross-border business scenario comes, the cross-border data privacy classification required for the scenario is obtained by importing the cross-border business scenario label. The specific steps are as follows:

[0105] (1) Form a basic data set of dynamic portraits of cross-border data business scenarios. Import the basic data of cross-border business scenarios obtained through research to form a basic data set of dynamic portraits of cross-border data business scenarios.

[0106] (2) Form a basic data set for a cross-border data privacy accurate grading model. Import the basic data on cross-border data privacy grading obtained from the survey to form a basic data set for a cross-border data privacy accurate grading model.

[0107] (3) Form a basic matching dataset of cross-border business scenarios and privacy levels. Import the basic dataset of the relationship between cross-border business scenarios and privacy levels obtained through research and analysis, such as Figure 8 As shown, a basic matching data set is formed.

[0108] (4) Form a cross-border scene label dataset. Figure 8 As shown, the matched dynamic portraits of cross-border business scenarios are labeled to form a labeled data set.

[0109] (5) Cross-border scenario-data association model based on label matching. Fig. 9 As shown, the present invention creates a model of adaptive scene matching labels. Through the reverse embedding of data hierarchical classification labels, an enhanced dataset with labels is obtained, which can be used to train the model and use the distance metric learning method to calculate the similarity between samples in different scenarios. Through the optimized distance metric function, we can more accurately reflect the similarity and difference of samples in different scenarios. Apply association rule mining technology to identify the association pattern between privacy data labels and scene features.

[0110] In this embodiment, 7. Dynamically associated adaptive model

[0111] like Fig. 9 As shown, the present invention designs a comprehensive set of automated processes to effectively realize the automated association between privacy data and usage scenarios. The present invention creates an adaptive scenario matching label model, forms a new cross-border scenario dynamic portrait based on the new cross-border business scenario, imports the labels of the new cross-border scenario based on the cross-border scenario dynamic portrait into the scenario-data association model based on label matching, and uses the rough set hierarchical classification method of the inter-domain distance metric school to predict and calculate the new cross-border data hierarchical classification, and constructs a dynamic association adaptive model.

[0112] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data classification method for adaptive cross-border scenarios, characterized in that: The following steps are involved: S1. Perform hierarchical classification based on the correlation depth of data in the privacy association model to form a detailed privacy data classification system; S2. Use graph chains to build privacy association models and establish relationships between data items, regulations, and corporate policies to achieve dynamic updates of cross-border privacy classification; S3. Design a data attribute privacy management model, including classification and grading dimensions, to achieve accurate and intelligent grading of cross-border data privacy; S4. Build a scalable cross-border scenario portrait model and use data modeling to depict the full picture of cross-border business scenarios; S5. Provide cross-border scenario dynamic expansion technology and cross-border scenario dynamic mapping technology to ensure the dynamic and accurate cross-border scenario portraits between jurisdictions; S6. Form a dynamic association method for privacy scenarios, and form a dynamic association model between cross-border business scenarios and privacy grading models by training the association relationship between the label model and the dynamic association model.

2. According to claim 1, the data classification method for adaptive cross-border scenarios is characterized by: The hierarchical classification includes: Level 1 privacy data: information directly related to privacy, which is the most sensitive; Secondary privacy data: including ID number, social security number, driver's license number, ID photo, etc. Although this type of data does not directly identify the identity, it is closely related to the personal identity; Subsequent privacy levels: arranged in descending order based on how closely they are linked to a person's identity.

3. The data classification method for adaptive cross-border scenarios according to claim 2 is characterized by: The data privacy association classification model includes: Class A privacy, Class B privacy, Class C privacy, and Class D privacy.

4. The data classification method for adaptive cross-border scenarios according to claim 3 is characterized by: The Class A privacy refers to data items related to laws and regulations; the Class B privacy refers to data items related to corporate policies; the Class C privacy refers to data items related to personal privacy; and the Class D privacy refers to data items related to privacy data.

5. The data classification method for adaptive cross-border scenarios according to claim 4 is characterized by: The data attribute privacy management model includes classification dimension and grading dimension. It divides data attributes into four categories according to their associated objects - laws and regulations, enterprises, individuals, and privacy data. It is further subdivided into N levels according to the closeness of data-related privacy, forming a clearly hierarchical privacy level.

6. The data classification method for adaptive cross-border scenarios according to claim 5 is characterized by: The cross-border scene static portrait model includes: Core data model: covers key entities such as individuals, legal terms and organizations; Business domain data model: involving business details such as social services and international trade; Cross-border business code set: contains codes for specific matters such as network and vehicle identification.

7. The data classification method for adaptive cross-border scenarios according to claim 6 is characterized by: The cross-border scene dynamic portrait technology includes: Cross-border scenario dynamic expansion technology: supports the addition, deletion and modification of dynamic elements in cross-border scenarios; Cross-border scenario dynamic mapping technology: supports the mapping of dynamic elements of cross-border scenarios between legal jurisdictions.

8. The data classification method for adaptive cross-border scenarios according to claim 7 is characterized by: The method for dynamically associating privacy scenarios also includes: (1) forming a basic data set of dynamic portraits of cross-border data business scenarios; (2) forming a basic data set of a cross-border data privacy precise classification model; (3) forming a basic matching data set of cross-border business scenarios and privacy classifications; (4) forming a cross-border scenario label data set; and (5) a cross-border scenario-data association model based on label matching.

9. The data classification method for adaptive cross-border scenarios according to claim 8 is characterized by: The dynamic association adaptive model includes: creating a model of adaptive scene matching labels and using a rough set hierarchical classification method based on inter-domain distance metric learning to predict and calculate new cross-border data hierarchical classifications.

10. The data classification method for adaptive cross-border scenarios according to claim 9 is characterized by: The data classification and categorization method is used to: (1) support the privacy computing processing of cross-border personal data; (2) assist enterprises and institutions in cross-border data classification practices; and (3) provide a reference for regulatory agencies to formulate policies and standards.

Images