Data classification and grading methods and systems based on industry knowledge structure and relationships

Abstract

This application relates to the field of data processing technology, specifically disclosing a data classification and grading method and system based on industry knowledge structure and relationships. The method includes: first, capturing and preprocessing raw data fragments of enterprise business traffic to obtain structured data metadata; then, performing type inference and name semantic parsing on data items to obtain association information, and extracting entity relationships to generate a heterogeneous relationship graph; next, constructing a multi-layered dynamic flow map; then, obtaining a first classification and grading result based on an industry regulatory knowledge base to ensure compliance; obtaining a second classification and grading result based on the multi-layered dynamic flow map to reflect the actual business situation; finally, combining the two to obtain the target classification and grading result, comprehensively considering regulatory requirements and actual business scenarios to make the result more accurate and reasonable. This helps enterprises fully leverage the value of data business while ensuring data security and compliance, flexibly adjust data management strategies, and improve data utilization efficiency, achieving high efficiency and accuracy.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to a data classification and grading method and system based on industry knowledge structure and relationships. Background Technology

[0002] In the field of data security, classification and grading are crucial and primary steps in data security governance. With decades of information technology development, users have accumulated massive amounts of data, far exceeding human processing capabilities. To further process this data and extract business value, it is necessary to clearly define the quantity and distribution of existing data, understand its data types and sensitivity levels, and formulate corresponding governance strategies for different types and levels of data sensitivity. Currently, classification and grading technologies include classification based on metadata type and classification based on actual application scenarios. In practice, a top-down classification approach is primarily used, combined with scanning and identification functions to discover data, with final manual verification and classification.

[0003] However, current classification and grading technologies have significant shortcomings. Their level of information technology assistance is lagging behind, relying mainly on human resources and static analysis. The mainstream approach combines manual annotation, machine scanning, and rule discovery to identify data structure relationships, and then uses big data models, AI, and other technologies to associate and identify more data. However, this approach requires a large amount of human intervention and organizational support and coordination, resulting in high costs and error rates, with actual results far below expectations. In addition, classification and grading work is time-consuming, requires a lot of manpower, and has poor results. Furthermore, the existing practices are not sustainable and cannot meet the requirements for continuous classification and grading. Summary of the Invention

[0004] In view of this, the present disclosure provides a data classification and grading method and system based on industry knowledge structure and relationships, which can solve the problems of poor accuracy of data classification and grading results, large manpower input, high cost, long time consumption and difficulty in sustainability in the prior art.

[0005] In a first aspect, embodiments of this disclosure provide a data classification and grading method based on industry knowledge structure and relationships, including:

[0006] Extract raw data fragments from the enterprise's business traffic and preprocess the raw data fragments to obtain structured data metadata;

[0007] Type inference and name semantic parsing are performed on each data item in the structured data metadata to obtain the association information of each data item;

[0008] Based on the association information, entity relationships are extracted from all the data items to identify the dynamic association relationships between entities, entities and data, and data and data, and a weighted heterogeneous relationship graph is generated.

[0009] Based on the heterogeneous relationship graph, a multi-layered dynamic flow map covering entities, data, users, systems, accounts, and roles is constructed.

[0010] The first classification and grading results of the data to be analyzed are obtained from the industry regulatory knowledge base;

[0011] The second classification and grading results of the data to be analyzed are obtained based on the multi-layer dynamic flow map.

[0012] Based on the first classification and grading results and the second classification and grading results, the target classification and grading results of the data to be analyzed are obtained.

[0013] Secondly, this disclosure also provides a data classification and grading system based on industry knowledge structure and relationships, including:

[0014] The preprocessing module is used to capture raw data fragments from the enterprise's business traffic and preprocess the raw data fragments to obtain structured data metadata.

[0015] The association information acquisition module is used to perform type inference and name semantic parsing on each data item in the structured data metadata to obtain the association information of each data item;

[0016] The heterogeneous relationship graph acquisition module is used to extract entity relationships from all data items based on association information, identify dynamic association relationships between entities, entities and data, and data, and generate a weighted heterogeneous relationship graph.

[0017] The multi-layer dynamic flow map construction module is used to build a multi-layer dynamic flow map covering entities, data, users, systems, accounts, and roles based on heterogeneous relationship graphs.

[0018] The first classification and grading result acquisition module is used to obtain the first classification and grading result of the data to be analyzed based on the industry regulatory knowledge base.

[0019] The second classification and grading result acquisition module is used to obtain the second classification and grading results of the data to be analyzed based on the multi-layer dynamic flow map.

[0020] The target classification and grading result acquisition module is used to obtain the target classification and grading results of the data to be analyzed based on the first classification and grading results and the second classification and grading results.

[0021] Thirdly, this disclosure also provides a computer device, which adopts the following technical solution:

[0022] The computer device includes:

[0023] At least one processor; and,

[0024] A memory communicatively connected to the at least one processor; wherein,

[0025] The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform any of the above-described data classification and grading methods based on industry knowledge structure and relationships.

[0026] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium storing computer instructions for causing a computer to execute any of the above-described data classification and grading methods based on industry knowledge structures and relationships.

[0027] Fifthly, embodiments of this disclosure also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0028] The data classification and grading method disclosed in this application, based on industry knowledge structure and relationships, firstly captures raw data fragments from an enterprise's business traffic and preprocesses these fragments to obtain structured data metadata. Secondly, it performs type inference and name semantic parsing on each data item in the structured data metadata to obtain the association information for each data item. Then, based on the association information, it extracts entity relationships from all data items, identifies dynamic relationships between entities, entities, and data, and generates a weighted heterogeneous relationship graph. Next, based on the heterogeneous relationship graph, it constructs a multi-layered dynamic flow map covering entities, data, users, systems, accounts, and roles. Finally, it obtains the first classification and grading results of the data to be analyzed based on an industry regulatory knowledge base, and obtains the results based on the multi-layered dynamic flow map. The second classification and grading result of the data to be analyzed is obtained. Based on the first and second classification and grading results, the target classification and grading result of the data to be analyzed is obtained. The first classification and grading result is derived from the industry regulatory knowledge base, which ensures that the data classification and grading comply with the requirements of relevant laws and regulations. The second classification and grading result is obtained based on a multi-layer dynamic flow map, reflecting the use and correlation of data in actual business. Combining the two allows for a comprehensive consideration of regulatory requirements and actual business scenarios, making the data classification and grading results more accurate and reasonable. This comprehensive approach helps enterprises to fully leverage the business value of data while ensuring data security and compliance. It satisfies regulatory constraints and allows for flexible adjustment of data management strategies according to actual business needs, thereby improving the efficiency of data utilization for enterprises.

[0029] The above description is merely an overview of the technical solution disclosed herein. In order to better understand the technical means of this disclosure and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 A flowchart illustrating the data classification and grading method based on industry knowledge structure and relationships provided in this embodiment of the disclosure.

[0032] Figure 2 This is a flowchart illustrating a method for obtaining structured data metadata provided in an embodiment of this disclosure.

[0033] Figure 3 This is a flowchart illustrating a method for obtaining associated information for each data item provided in an embodiment of this disclosure.

[0034] Figure 4 This is a flowchart illustrating the method for constructing a multi-layer dynamic flow map provided in an embodiment of this disclosure.

[0035] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. Detailed Implementation

[0036] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0037] It should be understood that the following specific examples illustrate the implementation of this disclosure, and those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0038] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0039] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this disclosure. The drawings only show the components related to this disclosure and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0040] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0041] Reference Figure 1 This application discloses a data classification and grading method based on industry knowledge structure and relationships, including:

[0042] S100 captures raw data fragments from the enterprise's business traffic and preprocesses these raw data fragments to obtain structured data metadata.

[0043] This step captures raw data fragments from enterprise business traffic and preprocesses them, transforming the messy raw data into structured data metadata. This effectively avoids the tedious process of manually processing massive amounts of data, greatly improves data usability, lays a good foundation for subsequent analysis and processing, and makes the data processing workflow more efficient and smooth.

[0044] S200 performs type inference and name semantic parsing on each data item in the structured data metadata to obtain the association information of each data item.

[0045] This step performs type inference and name semantic parsing on the data items in the structured data metadata, automatically obtaining the association information of each data item. This intelligent processing method reduces the time and workload of manual judgment, and can quickly and accurately identify the potential relationships between data items, thus improving the efficiency of data association analysis.

[0046] S300 extracts entity relationships from all data items based on association information, identifies dynamic relationships between entities, entities and data, and data, and generates a weighted heterogeneous relationship graph.

[0047] This step extracts entity relationships based on the associated information, identifies the dynamic relationships between entities, entities and data, and data and data, and generates a weighted heterogeneous relationship graph. Compared with traditional static analysis methods, it can more comprehensively and accurately reflect the complex relationships between data, effectively avoid information loss and erroneous judgment caused by static analysis, and provide a more reliable basis for subsequent data classification and grading.

[0048] S400, based on heterogeneous relationship graphs, constructs a multi-layered dynamic flow map covering entities, data, users, systems, accounts, and roles.

[0049] This step involves constructing a multi-layered dynamic flow map based on a heterogeneous relationship graph, covering entities, data, users, systems, accounts, and roles. This map comprehensively displays the flow of data at different levels and stages, enabling enterprises to clearly understand the flow path and usage of data. This helps to classify and grade data more accurately, improving the accuracy and refinement of data management.

[0050] S500 obtains the first classification and grading results of the data to be analyzed based on the industry regulatory knowledge base.

[0051] This step obtains the first classification and grading results of the data to be analyzed based on the industry regulatory knowledge base, and integrates industry regulatory requirements into the data classification and grading process. This ensures that the company's data management activities comply with relevant laws and regulations, avoids legal risks caused by non-compliance with regulations in data classification and grading, and provides compliance protection for the company's data security governance.

[0052] S600 obtains the second classification and grading results of the data to be analyzed based on the multi-layer dynamic flow map.

[0053] The multi-layered dynamic flow map covers multiple levels, including entities, data, users, systems, accounts, and roles. It details the flow path of data between different stages and systems, which allows the classification and grading of data to be closely integrated with actual business usage scenarios. In other words, through the multi-layered dynamic flow map, changes in data usage scenarios can be accurately captured, thereby enabling more precise classification and grading of data.

[0054] S700 obtains the target classification and grading results of the data to be analyzed based on the first classification and grading results and the second classification and grading results.

[0055] The first classification and grading result is derived from an industry regulatory knowledge base, ensuring that the data classification and grading comply with relevant laws and regulations. The second classification and grading result is obtained based on a multi-layered dynamic flow map, reflecting the data's usage and relationships in actual business operations. Combining the two allows for a comprehensive consideration of regulatory requirements and actual business scenarios, making the data classification and grading results more accurate and reasonable. This comprehensive approach helps enterprises fully leverage the business value of data while ensuring data security and compliance. It satisfies regulatory constraints while allowing for flexible adjustments to data management strategies based on actual business needs, thereby improving the efficiency of data utilization for enterprises.

[0056] Traditional data classification and grading is often a large-scale, one-time investment project, requiring companies to concentrate significant human, material, and financial resources within a specific timeframe. This model is not only costly, but also quickly becomes outdated as business operations and data evolve, failing to meet new business needs. The method disclosed in this application innovatively transforms classification and grading into a continuous, ongoing process. Through automated data capture, preprocessing, and analysis workflows, it enables continuous processing and classification of newly generated data. This means companies no longer need large-scale, one-time investments, but can gradually and continuously advance the classification and grading process based on actual needs, closely integrating it with daily operations and making it more flexible and adaptable.

[0057] The method disclosed in this application constructs a multi-layered dynamic flow map covering entities, data, users, systems, accounts, and roles, which enables continuous monitoring of the data flow process within an organization. This allows enterprises to understand the flow of data between different stages and systems in real time, including information such as the source, destination, and frequency of use of the data.

[0058] Furthermore, this application can also help enterprises continuously improve the classification and grading of data based on the results of ongoing monitoring. As business develops and data flow changes, the sensitivity and importance of certain data may change. Through dynamic flow maps, enterprises can promptly identify these changes and adjust the classification and grading of data accordingly, thereby achieving a near real-time correspondence between the classification and grading results and the actual situation. This real-time adjustment capability ensures that the classification and grading results always accurately reflect the actual state of the data, improving the effectiveness of data management.

[0059] Traditional methods rely primarily on static analysis, which struggles to accurately reflect the complex relationships and dynamic changes between data, leading to inaccurate classification and grading results. This solution identifies dynamic relationships between entities, entities and data, and data, generating weighted heterogeneous relationship graphs and constructing multi-layered dynamic flow maps. This comprehensively and accurately reflects the actual situation of the data. Furthermore, by combining the target classification and grading results derived from the industry regulatory knowledge base and the dynamic flow map, and by comprehensively considering regulatory requirements and actual business scenarios, the accuracy of classification and grading is further improved.

[0060] Existing classification and grading methods lack sustainability and cannot adapt to the continuous growth and changes in enterprise data. In contrast, the solution proposed in this application can continuously process and classify new data, and can adjust the classification and grading results in real time according to changes in data flow. This continuous improvement mechanism ensures that the classification and grading work can be carried out continuously and effectively, meeting the long-term data management needs of enterprises.

[0061] Reference Figure 2 The method S100, which "captures raw data fragments from an enterprise's business traffic and preprocesses these fragments to obtain structured data metadata," specifically includes the following:

[0062] S110 uses bypass monitoring technology to capture raw data fragments from business traffic.

[0063] Specifically, port mirroring / ERSPAN can be enabled at core switching nodes or cloud-native traffic probes to replicate all TCP / UDP traffic.

[0064] In this step, bypass monitoring technology does not interfere with the normal business traffic of the enterprise network because it monitors by copying traffic rather than directly intervening in the transmission path of business data. Therefore, it does not affect the performance and stability of the business system. At the same time, it can acquire all business traffic in the enterprise network, including communication between internal systems and interaction with external systems. This allows the enterprise to have a comprehensive understanding of the flow of its business data, providing a rich data source for subsequent data processing and analysis.

[0065] S120, a session reassembly algorithm based on a sliding window, reassembles all original data fragments into a complete session corresponding to the business interaction process.

[0066] Reassembling fragmented data into complete sessions allows businesses to clearly understand the entire process of business interactions. This helps them analyze whether business processes are smooth and whether user behavior meets expectations, providing strong support for business optimization. Complete session data is also easier to analyze and mine later. For example, when conducting user behavior analysis, complete session data can provide more comprehensive information on user actions, helping businesses better understand user needs and preferences, thereby developing more precise marketing strategies.

[0067] S130 appends a globally unified session ID, client certificate fingerprint, server certificate fingerprint, user identifier, and nanosecond-level timestamp to each data fragment in the complete session to form structured data metadata.

[0068] A globally unified session ID enables enterprises to easily manage and query different sessions. Through the session ID, enterprises can quickly locate and retrieve all data fragments of a specific session, improving data retrieval efficiency. Meanwhile, client and server certificate fingerprints can be used to verify the identities of both communicating parties, ensuring the security of data transmission. User identifiers and timestamps provide traceability for the data; enterprises can use this information to track the source of data, the time of user operations, etc., which helps in investigations and audits in the event of security issues or business disputes.

[0069] The method for obtaining structured data metadata disclosed in this embodiment ensures the integrity and accuracy of business data through capture, reorganization, and structuring. Enterprises can obtain complete business interaction information, avoiding data loss and errors, and providing a reliable foundation for subsequent data processing and analysis. Complete and structured business data enables enterprises to deeply analyze business processes and user behavior, and discover potential problems and opportunities. Additional information such as certificate fingerprints, user identifiers, and timestamps helps enterprises conduct security monitoring and compliance checks. Enterprises can monitor abnormal session activity in real time, promptly detect security threats, and meet the requirements of relevant regulations and industry standards for data traceability and security.

[0070] Reference Figure 3 The method S200, which involves "performing type inference and name semantic parsing for each data item in structured data metadata to obtain the association information of each data item," specifically includes the following methods for obtaining the association information of each data item:

[0071] S210 constructs a multi-layered knowledge graph covering industry terms, synonyms, and abbreviations.

[0072] The multi-layered knowledge graph includes a core layer knowledge graph, an extended layer knowledge graph, and a customized layer knowledge graph.

[0073] Furthermore, the core layer of the knowledge graph contains the most basic and core industry terms in the field. These terms form the cornerstone of the entire knowledge graph and are highly professional and authoritative.

[0074] The extended layer knowledge graph covers synonyms and common abbreviations related to the core layer terms, enabling the knowledge graph to handle more diverse forms of expression by expanding the semantic scope of the core layer.

[0075] The customized knowledge graph is a set of terms that are customized according to specific business needs or specific scenarios. It can be flexibly added or modified to adapt to different application environments.

[0076] Multi-layered knowledge graphs can comprehensively and deeply represent industry knowledge. The core layer provides the basic framework, the extension layer enriches the knowledge details, and the customization layer meets specific business needs. Through knowledge graphs, the semantics of data items can be better understood, providing strong support for subsequent entity linking and disambiguation.

[0077] S220 uses a BERT-BiLSTM-CRF cascaded network to sequentially label the names of data items in structured data metadata to obtain entity boundaries.

[0078] BERT can learn powerful semantic representations, BiLSTM can process the contextual information of sequences, and CRF can perform accurate sequence labeling. Through this cascaded network, entity boundaries in data item names can be accurately identified, providing accurate entity information for subsequent entity links.

[0079] S230 links entities in the entity boundary with nodes in the multi-layer knowledge graph, and uses the PageRank algorithm to disambiguate the linking results to obtain the semantic label of each data item.

[0080] Entity links connect data items to the knowledge graph, giving the data items clear semantics; the PageRank algorithm can effectively resolve ambiguity in entity links, improve the accuracy of semantic labels, and thus better understand the meaning of data items.

[0081] Suppose that the entity "Apple phone" is obtained from the data item "product_name" via S220. Matching "Apple phone" with nodes in a multi-layered knowledge graph reveals that the knowledge graph contains a node for "Apple phone." However, there may be multiple pieces of information related to "Apple phone," such as different models of Apple phones. In this case, the PageRank algorithm is used to disambiguate these link results. The PageRank algorithm assigns a weight to each link result based on the importance of nodes and their relationships in the knowledge graph, selecting the link result with the highest weight as the final semantic label, such as "iPhone 15."

[0082] S240, Extract the data content of each data item in the structured data metadata.

[0083] This step provides actual data samples for subsequent data type analysis. Only after obtaining the data content can its type be determined and analyzed.

[0084] For example, in e-commerce order data, for the data item "price", its corresponding data content is extracted from the data record, such as "9999"; for "product_name", the specific product name is extracted, such as "Apple iPhone 15".

[0085] S250: The target model is trained using data samples of known types. The trained model is then used to analyze the data content of each data item to obtain the content type label for each data item.

[0086] The associated information includes semantic tags and content type tags for data items.

[0087] Through model training and analysis, the data content type of data items can be automatically and accurately determined, providing important type information for data processing and analysis, which is helpful for subsequent data cleaning, transformation and mining operations.

[0088] Specifically, a large number of data samples of known types are collected, such as numeric types (100, 20.5), string types ("Apple", "phone"), date types ("2025-08-14"), etc., and these samples are used to train the target model (such as a deep learning model). Then, the extracted data content "9999" is input into the trained model, and the model judges it to be a numeric type, thus obtaining the content type label "numeric type" for the "price" data item.

[0089] By performing type inference and name semantic parsing on data items, combined with multi-layered knowledge graphs and entity link disambiguation, the meaning and type of data can be understood more accurately, providing a more reliable basis for data analysis and decision-making. Clear data types and semantic labels facilitate automated data processing workflows, reduce manual intervention, and improve the speed and accuracy of data processing. The customized layers and type inference mechanisms of multi-layered knowledge graphs can be adjusted and optimized according to different business needs to meet diverse business scenarios. Unified data types and semantic labels facilitate data fusion and sharing between different data sources, breaking down data silos and improving the utilization value of data.

[0090] The S300 method, which "extracts entity relationships from all data items based on association information, identifies dynamic relationships between entities, entities, and data, and generates a weighted heterogeneous relationship graph," specifically includes the following steps:

[0091] S310: Data Preprocessing and Integration of Association Information. Specifically, 1) Clean the association information of all data items obtained from step S200, removing noisy data such as incorrect semantic labels and incomplete type information. For example, if a semantic label in the association information of a data item contains a spelling error, it is corrected; if type information is missing, it is supplemented or deleted according to the context or common rules. Duplicate data items are handled; for data items with completely identical association information, only one copy is retained to reduce the complexity of subsequent processing.

[0092] 2) Standardize the format and expression of related information. For example, map all semantic tags according to a predefined vocabulary to ensure that semantic tags with the same meaning have a unified representation; standardize data types, such as unifying date types with different representations into a standard format.

[0093] S320: Entity Recognition and Classification. Specifically, 1) Develop a set of rules to identify entities in data items. For example, based on semantic tags in the associated information, if the semantic tag of a data item contains keywords such as "customer," "product," or "order," then it is identified as the corresponding entity.

[0094] 2) Use the type information of data items to assist in entity identification. For example, data items with the type "object" are more likely to be entities.

[0095] 3) Classify the identified entities according to their business attributes. For example, classify entities into customer entities, product entities, order entities, etc. Simultaneously, define a set of features for each entity category for subsequent relation extraction.

[0096] S330: Entity Relationship Extraction. Specifically, this includes: 1) Using machine learning-based relationship extraction, specifically: selecting appropriate machine learning algorithms, such as Support Vector Machine (SVM) and decision trees, to extract relationships between entities, between entities and data, and between data. First, extract features from the association information, such as semantic label similarity and data type matching, as input to the machine learning model.

[0097] The model is trained using labeled training data, which contains known entity relationship pairs and their corresponding relationship types. After training, the association information of all data items is input into the model, and the relationship types between entities are output.

[0098] 2) Rule-based relation extraction can be used, which specifically includes: formulating a series of rules to supplement the relations that the machine learning model has not identified. For example, if two data items have semantic labels of "order" and "product" respectively, and they often appear together in the business process, then a "containment" relationship can be inferred between them.

[0099] Considering the temporal order and business logic between data items, such as in a business process, if data item A always appears before data item B and there is a data transfer relationship between them, then it can be inferred that there is a "precedence" relationship between them.

[0100] S340: Relationship Weight Calculation. Specifically, this includes: 1) Weight calculation based on association strength can be used, specifically: for the identified relationships between entities, entities-data, and data-data, the weight of the relationship is calculated based on their association strength. Association strength can be measured in various ways, such as semantic similarity and data interaction frequency. For example, if two entities frequently interact in a business process and their semantic tags have high similarity, then the relationship weight between them is high.

[0101] 2) Weighting based on business importance can be used, specifically including: adjusting the weight of relationships according to their importance in the business. For example, appropriately increase the weight of relationships involving core business processes; decrease the weight of auxiliary relationships.

[0102] S350: Heterogeneous Relationship Graph Construction. This specifically includes: using identified entities and data items as nodes in the graph, and extracting relationships as edges. Each node and edge has corresponding attributes; node attributes include association information and entity category, while edge attributes include relationship type and weight. A graph database (such as Neo4j) is used to construct and store the heterogeneous relationship graph. The node and edge information is inserted into the graph database to form a complete heterogeneous relationship graph. Simultaneously, an index is created for the graph database to improve query and analysis efficiency.

[0103] S360: Dynamic Relationship Updates. Specifically, this includes: 1) Continuously monitoring raw data fragments in business traffic and promptly updating structured data metadata and relationship information when new data is generated or existing data changes. 2) Based on the updated relationship information, re-performing entity identification, relationship extraction, and weight calculation to update the heterogeneous relationship graph, reflecting the dynamic relationships between entities, entity-data, and data-data.

[0104] Reference Figure 4The method for S400 to "build a multi-layered dynamic flow map covering entities, data, users, systems, accounts, and roles based on heterogeneous relationship graphs" specifically includes the following:

[0105] S410, determine the hierarchical structure of the initial multi-layer map and the relationship between each layer.

[0106] The hierarchical structure includes the entity layer, data layer, user layer, system layer, account layer, and role layer.

[0107] The role layer represents different roles within the enterprise, such as administrators, regular employees, and auditors. Each role has a specific set of responsibilities and permissions, forming an abstract layer for permission management within the multi-layered map.

[0108] The account layer contains all account information within the enterprise. Each account corresponds to a unique identifier and is associated with a specific role. The account is the specific identity identifier for users to operate in the system.

[0109] The user layer represents the people who actually use the enterprise system. A user can have multiple accounts, and each account may be associated with different roles. The user layer embodies the main body of interaction between people and the system.

[0110] The system layer encompasses various application systems, database systems, business systems, etc. within an enterprise. These systems serve as carriers for data storage and processing, and there may be data interactions and dependencies between different systems.

[0111] The entity layer contains various entities in the enterprise's business, such as customers, products, and orders. Entities are abstract representations of business data and have clear business meanings.

[0112] The data layer stores the enterprise's specific data, including structured and unstructured data. The data is closely related to entities and is the specific object of business operations.

[0113] The association methods are defined as follows: 1) Role-Account Association: One role can be associated with multiple accounts, implemented through a role-account mapping table. This table records the role information corresponding to each account, used for permission allocation and management. 2) Account-User Association: A user can have multiple accounts, established through a user-account mapping table. This table records the correspondence between users and accounts, facilitating user identity management and operation auditing. 3) User-System Association: Users can access different systems through their accounts, reflected through a user-system access record table. This table records information such as the time and operation of users accessing systems using their accounts, reflecting the interaction between users and systems. 4) System-Entity Association: Systems are responsible for managing and operating entities, established through a system-entity mapping table. This table records the types and scope of entities managed by each system, reflecting the business logic relationship between systems and business entities. 5) Entity-Data Association: An entity is an abstraction of data; one entity can correspond to multiple data records. This is established through an entity-data index table, which records the correspondence between entities and specific data, facilitating data querying and management.

[0114] A clear hierarchical structure helps organize and manage complex information, making the location and classification of each element clear; a clear association method provides a basic framework for subsequent data flow and permission configuration, facilitating system construction and maintenance.

[0115] S420 maps entities and data nodes in the heterogeneous relationship graph to the corresponding levels of the initial multi-layer map.

[0116] The entity mapping specifically involves: traversing the entity nodes in the heterogeneous relationship graph and mapping them to the entity layer of the multi-level map based on the entity's type and business meaning. For example, if the entity is "customer," it is mapped to the "customer" category in the entity layer; if it is "product," it is mapped to the "product" category. A unique identifier is assigned to each entity node mapped to the entity layer, and an association is established with the corresponding node in the heterogeneous relationship graph for subsequent querying and tracing.

[0117] The data mapping specifically includes: for data nodes in a heterogeneous graph, mapping them to data layers of a multi-layered map based on their associated entity and data type. For example, data related to the "customer" entity, such as customer name and contact information, is mapped to the data set corresponding to the "customer" entity in the data layer. Establishing associations between data nodes and entity nodes ensures that the data in the data layer accurately reflects the attributes and states of the entities in the entity layer.

[0118] This step enables the connection between heterogeneous relationship diagrams and multi-layer maps, integrating data from different sources and formats, making the information more unified and orderly, facilitating further processing and analysis of entities and data in multi-layer maps, and avoiding data chaos and duplication.

[0119] S430 constructs data flow relationships in the initial multi-layer map based on the data flow paths and methods between different entities, users, and systems.

[0120] The data flow path analysis specifically includes: extracting data flow information between different entities, users, and systems from heterogeneous relationship graphs, including the data's starting node, intermediate nodes, and ending node, as well as the time sequence and frequency of data flow. It also involves analyzing the business rules and logic of data flow to determine the triggering conditions and constraints for data flow, such as the approval processes that data needs to go through when flowing between different systems.

[0121] The construction of data flow relationships specifically includes: in a multi-layered map, representing data flow relationships by establishing directed edges between nodes at different levels. For example, establishing a directed edge from a user node in the user layer to a system node in the system layer indicates that the user has submitted data to the system; establishing a directed edge from a system node in the system layer to an entity node in the entity layer indicates that the system has performed a data update operation on that entity. Attributes are added to each directed edge, including information such as the time of data flow, the amount of data, and the flow method (e.g., real-time transmission, batch transmission), to describe the data flow process in more detail.

[0122] This step visually demonstrates the data flow process, helping to identify bottlenecks and problems in data flow, optimize business processes, and provide a basis for subsequent access control, as different data flow stages may require different permissions.

[0123] S440 determines the access permissions of different roles to entities and data based on business rules and security policies.

[0124] The business rules review specifically includes: collecting the company's business rules and process documents, and analyzing the needs and restrictions of different roles on entities and data in business operations. For example, an administrator role may have permissions to create, modify, and delete all entities and data, while a regular employee role may only have read-only permissions for some entities and data. In conjunction with the company's business development strategy and compliance requirements, the operational permission boundaries for different roles in different business scenarios are determined.

[0125] The specific aspects of security policy formulation include: developing data access security policies based on the company's security management system and regulatory requirements. For example, for sensitive data, strict access control policies should be set up, allowing only authorized roles to access the data. Considering the confidentiality, integrity, and availability of data, the operational permissions for different roles should be refined and graded, such as setting different access levels and operation types (e.g., read, write, modify, delete).

[0126] The permission determination process includes: determining the specific operation permissions for each role on entities and data based on business rules and security policies. The operation permissions of each role on each entity and data are recorded using a permission matrix. Rows in the matrix represent roles, columns represent entities and data, and matrix elements represent the operation permissions of that role on that entity or data.

[0127] This step ensures the security and integrity of data, preventing unauthorized access and operations; at the same time, it complies with business rules, allowing different roles to perform operations related to their own responsibilities, thus improving the standardization and efficiency of the business.

[0128] S450 configures operation permissions in the initial multi-layer map to obtain the constructed multi-layer dynamic flow map.

[0129] The permission configuration implementation includes: in the multi-layered map's role layer, account layer, user layer, system layer, entity layer, and data layer, configuring corresponding permission information for each node and edge based on a determined operation permission matrix. For example, in the role layer, adding the corresponding permission set to each role node; in the account layer, associating the permissions of its assigned role with each account node. In the system layer, configuring permission control for the system's access interfaces and functional modules ensures that only roles and accounts with the corresponding permissions can access and operate.

[0130] The dynamic update mechanism includes: establishing a multi-layered dynamic update mechanism for the map, which can promptly update the permission configuration and data flow relationships in the multi-layered map when changes occur in the enterprise's business rules, security policies, role definitions, account information, etc. Regular audits and evaluations of the multi-layered map are conducted to check the rationality of permission configurations and the compliance of data flow, and any problems are promptly adjusted and optimized to obtain a multi-layered dynamic flow map that can reflect the enterprise's business and security status in real time.

[0131] This step enables precise control over the operations of different roles, ensuring the security and stability of the system; the multi-layered dynamic flow map can reflect the flow of data and changes in permissions in real time, providing strong support for enterprise decision-making and management.

[0132] The method for constructing a multi-layered dynamic flow map disclosed in this embodiment enables enterprises to better manage and utilize data, reducing data redundancy and errors through a clear hierarchical structure and data flow relationships. Role-based permission configuration can effectively prevent data leakage and illegal operations, protecting the enterprise's core information. The intuitive display of the data flow process helps to identify problems in business processes, enabling optimization and improvement. The multi-layered dynamic flow map provides enterprises with a comprehensive information view, helping management to make more informed decisions.

[0133] The S700 method for "obtaining the target classification and grading results of the data to be analyzed based on the first and second classification and grading results" specifically includes:

[0134] S710, if the first classification and grading result is consistent with the second classification and grading result, then the corresponding classification and grading result shall be taken as the target classification and grading result.

[0135] S720, If the first classification and grading results are inconsistent with the second classification and grading results, obtain the data type of the data to be analyzed;

[0136] S730, determine the corresponding industry knowledge structure weight and correlation weight based on the data type;

[0137] S740 retrieves the classification and grading result corresponding to the largest weight among the industry knowledge structure weight and the correlation weight, and uses it as the target classification and grading result for the data to be analyzed.

[0138] This embodiment, by comprehensively considering the results of different classification and grading systems, as well as data type, industry knowledge, and relationships, can classify and grade data more comprehensively and accurately, reducing the errors that may exist in a single classification and grading system. This solution can handle situations where the results of different classification and grading systems are inconsistent, has strong adaptability, and is applicable to a variety of different industries and data scenarios. By introducing industry knowledge structure weights and relationship weights, the classification and grading process becomes more scientific and reasonable, avoiding subjective arbitrariness and improving the credibility and authority of the classification and grading results.

[0139] Furthermore, the data classification and grading method based on industry knowledge structure and relationships disclosed in this application also includes: triggering version management of target classification and grading results according to a preset period; and using a graph difference algorithm based on edit distance and a content difference algorithm based on semantic embedding to automatically identify data items added, deleted, or modified between adjacent versions and their changes in sensitivity level, and generate a difference report.

[0140] Version control allows for a clear record of changes to data classification and grading results within each preset period. This is akin to taking photos of every "transformation" of the data and arranging them chronologically. When querying the classification and grading status of specific data at a particular point in time, one can directly trace back to the corresponding version, clearly identifying the data's state at different stages. For example, in medical data management, if an error is subsequently discovered in the classification and grading of a patient's medical records, version control can quickly pinpoint the period in which the change occurred, as well as the circumstances before and after the change, facilitating timely error correction and accountability.

[0141] In many industries, such as finance and healthcare, there are strict regulations and compliance requirements. Versioning makes the process and results of data classification and grading auditable. Regulatory agencies or internal auditors can review the classification and grading of different versions of data to ensure that data processing and protection comply with relevant regulations and policies. For example, financial institutions are required to regularly report on the security management of data to regulatory authorities, and the results of versioned classification and grading can serve as strong evidence that the institution's data management is standardized and traceable.

[0142] When business operations change, organizational structures are adjusted, or data processing processes are updated, the classification and grading results of data may also need to be adjusted accordingly. Versioning allows updates to be made without destroying historical data records, ensuring business continuity. For example, after a business restructuring, an enterprise may need to reassess the importance and sensitivity of data. Versioning can retain the classification and grading results of the old version as a reference, while recording the changes in the new version, ensuring the stability of data management during the transition period.

[0143] By leveraging graph difference algorithms based on edit distance and content difference algorithms based on semantic embedding, it is possible to automatically and quickly identify the addition, deletion, and modification of data items between adjacent versions, as well as changes in sensitivity levels. This significantly improves the efficiency of data monitoring, eliminating the need for manual comparison of large amounts of data. For example, in large e-commerce platforms, a large amount of product and user data is updated daily. These algorithms can quickly detect changes in data classification and grading, allowing for timely implementation of appropriate security measures.

[0144] Difference reports clearly identify which data items have changed and how their sensitivity levels have changed, helping businesses pinpoint potential risks. For example, if a data item that was previously considered low-sensitivity is upgraded to high-sensitivity in a new version, the business can immediately strengthen its protection measures to prevent data breaches.

[0145] Based on the variance report, enterprises can allocate resources appropriately for data protection. For newly added or upgraded data items, increase corresponding security investment; for deleted or downgraded data items, appropriately reduce protection resources. For example, in a cloud computing environment, for data with upgraded sensitivity, increase storage encryption levels and access control permissions, while for data with downgraded sensitivity, appropriately reduce storage costs.

[0146] Difference reports provide clear information for communication between different departments. The data management team can share reports with the security team, business teams, and others, allowing all parties to understand the changes in data. For example, the security team can strengthen security measures based on the report, and the business team can adjust business strategies based on changes in data sensitivity levels, thereby promoting collaboration between teams and overall business development.

[0147] Furthermore, the data classification and grading method based on industry knowledge structure and relationships disclosed in this application also includes: submitting the difference report to user approval through a multi-role collaborative approval workflow; wherein, the workflow has a built-in interpretability presentation module based on role-attribute access control, and the approval process records the complete decision trajectory and reason chain; the updated difference content after approval is merged into the production environment through a zero-downtime canary release mechanism, triggering online incremental updates.

[0148] Specifically, this includes: a) adopting a version switching strategy based on a dual-temporal database to ensure historical traceability; b) adopting an audit log based on event tracing to record each upgrade action of the classification and grading results; and c) adopting a distributed cache invalidation strategy based on consistent hashing to ensure global consistency.

[0149] In this embodiment, regarding "submitting the difference report to user approval through a multi-role collaborative approval workflow," assuming a financial enterprise adopts this data classification and grading method, after the difference report is generated, the system will send the report to different personnel according to preset roles. For example, it is first sent to the data administrator, who reviews the differences in data classification and grading in the report, assesses their impact on the existing system and business, and records their approval opinion in the system, such as "Some differences may affect the risk assessment model and require further review." Then, the report flows to the business expert, who analyzes the differences from a business perspective, determines whether they comply with business rules and regulatory requirements, and provides an approval opinion, such as "Some differences comply with the new business specifications and can be approved." Finally, it flows to the security expert, who assesses the impact of the differences on data security. If the security risk is confirmed to be controllable, the entire approval process is completed.

[0150] The interpretability presentation module in the workflow displays the approval basis and rationale for each role in an intuitive way. For example, it lists each role's approval comments, approval time, and relevant reference document links in tabular form. The decision-making trajectory and rationale chain during the approval process are fully recorded in the system, facilitating subsequent auditing and traceability.

[0151] Multi-role collaborative approval allows for the evaluation of discrepancy reports from different professional perspectives, avoiding the limitations of a single role and enabling more comprehensive and accurate decisions. A fully documented decision-making process and rationale chain ensure transparency and traceability, facilitating subsequent audits and compliance checks, and aiding in the rapid identification of responsibility when issues arise. Many industries, such as finance and healthcare, have stringent compliance requirements for data changes. Multi-role collaborative approval and complete documentation can meet these compliance requirements.

[0152] For the process of "integrating approved and confirmed changes into the production environment through a zero-downtime canary release mechanism to trigger online incremental updates," a version switching strategy based on a dual-temporal database is adopted to ensure historical traceability. Taking an e-commerce enterprise's data classification and grading system as an example, a dual-temporal database is used. When approved changes need to be integrated into the production environment, the system creates a new database version. In the dual-temporal database, each data record has two time dimensions: validity time and transaction time. Valid time represents the valid time period of the data in the real world, and transaction time represents the time the data was recorded in the database. When new changes are integrated, new validity times and transaction times are set for these data. For example, if a product's data category changes from "regular product" to "popular product," the new record will take effect in the new version, while the old version's record is still retained. The validity time and transaction time clearly trace the data change history.

[0153] This solution allows for easy querying of data status at different points in time, which is helpful for data analysis, compliance auditing, and troubleshooting. The dual-temporal database version switching strategy enables clear management of different versions of data, avoiding data loss or confusion. If problems arise with newly merged differences, the system can be quickly switched back to the old version, ensuring system stability.

[0154] The phrase "using event-based audit logs to record every upgrade action of classification and grading results" refers to, for example, a logistics company's data classification and grading system. Whenever an upgrade action occurs, the system generates an event log. For instance, when a batch of goods is upgraded from "ordinary goods" to "high-value goods," the event log contains detailed information about the event, such as the time of the event, the user or system component that triggered it, and the classification and grading results before and after the upgrade. These event logs are stored in the audit log in chronological order, forming a complete event flow.

[0155] Audit logs can record in detail every upgrade action of the classification and grading results, providing complete evidence for subsequent audits and compliance checks; when data anomalies or business problems occur, audit logs can be used to quickly locate the time and cause of the problem, which helps to resolve the problem in a timely manner; by analyzing the events in the audit logs, we can understand the changing trends of data classification and grading, and provide a reference for business decisions.

[0156] Regarding the strategy of "adopting a distributed cache invalidation strategy based on consistent hashing to ensure global consistency," consider a scenario where a large internet company has multiple distributed cache nodes storing data classification and grading results. When new discrepancies are merged into the production environment, the data in the cache needs to be updated to ensure global consistency. A consistent hashing algorithm maps data keys to a hash ring, with each cache node having a corresponding position on the hash ring. When data changes, it's only necessary to recalculate the position of the affected data key on the hash ring using the consistent hashing algorithm, and then update the corresponding cache nodes. For example, if the classification and grading results for a certain type of data change, the system will use the consistent hashing algorithm to determine which cache nodes store that type of data, and then only update the caches of those nodes, without affecting other unrelated cache nodes.

[0157] Consistent hashing's distributed cache invalidation strategy ensures that data remains consistent across all cache nodes in a distributed environment, preventing business errors caused by data inconsistency. It only updates affected cache nodes, avoiding a full update of all cache nodes and reducing system overhead and performance degradation. When adding or removing cache nodes, the consistent hashing algorithm can smoothly adjust the data distribution, ensuring system scalability.

[0158] Through multi-role collaborative approval and complete logging, the system ensures that data classification and grading updates comply with business rules and compliance requirements. Dual-temporal databases and event tracing audit logs further enhance the traceability and transparency of data management, helping to meet various regulatory requirements. A zero-downtime canary release mechanism allows for the integration of new differences without affecting normal system operation, minimizing business impact. Version switching strategies based on dual-temporal databases and distributed cache invalidation strategies using consistent hashing guarantee data consistency and rollbackability, improving system stability and reliability. Timely integration of approved differences into the production environment, triggering online incremental updates, enables rapid response to changes in business needs, enhancing business flexibility and competitiveness. Simultaneously, the traceability of historical data and event audit logs facilitate business analysis and decision-making, promoting continuous business optimization.

[0159] Furthermore, traditional methods require significant manual intervention, including manual labeling, screening, and classification, which is costly. The solution proposed in this application, however, reduces reliance on manual labor through automated and intelligent processes. For example, in the data acquisition and preprocessing stages, the system can automatically complete most of the work, reducing labor costs. Simultaneously, the continuous investment model avoids the waste of resources from large one-time investments, enabling enterprises to allocate resources more rationally and further reducing costs. Automated processes and real-time monitoring mechanisms make data element organization more efficient. The system can quickly process and classify newly generated data, reducing data processing time. Moreover, the comprehensive data correlation information provided by the multi-layered dynamic flow map helps enterprises more quickly identify the type and sensitivity of data, improving the speed and accuracy of classification and grading, thereby enhancing the overall efficiency of data element organization.

[0160] A single classification and grading method may have limitations and be susceptible to factors such as incomplete data and inaccurate analysis methods. The method disclosed in this application combines two different classification and grading results, allowing for mutual verification and supplementation, thus improving the reliability and credibility of the classification and grading. If the first and second classification and grading results are consistent, the accuracy of the classification and grading can be further confirmed; if differences exist, the reasons for the differences can be further analyzed, leading to a more in-depth evaluation and judgment, thereby deriving a more reliable target classification and grading result. Reliable classification and grading results provide a solid foundation for enterprise data security governance. Enterprises can formulate more effective data protection strategies, access control rules, and emergency response plans based on accurate classification and grading results, reducing data security risks and ensuring the normal operation of the enterprise.

[0161] As businesses grow and the external regulatory environment changes, data classification and grading require continuous adjustment and optimization. The target classification and grading result, obtained based on the first and second classification and grading results, can serve as a dynamic reference. When regulations change, the first classification and grading result can be adjusted according to new regulatory requirements. When business processes or data flow changes, the second classification and grading result can be updated. By continuously updating and optimizing these two classification and grading results and recalculating the target classification and grading result, enterprises can achieve continuous improvement in data governance, ensuring that data classification and grading always adapt to the enterprise's actual needs and regulatory requirements. This continuous optimization mechanism helps enterprises maintain the advancement and effectiveness of data management, improving their data security level and competitiveness. Simultaneously, it provides strong support for enterprises' digital transformation and innovative development, enabling them to better cope with increasingly complex data security challenges.

[0162] Secondly, this application discloses a data classification and grading system based on industry knowledge structure and relationships, used to execute the data classification and grading method based on industry knowledge structure and relationships disclosed in the first aspect of this application. The system specifically includes:

[0163] The preprocessing module is used to capture raw data fragments from the enterprise's business traffic and preprocess the raw data fragments to obtain structured data metadata.

[0164] The association information acquisition module is used to perform type inference and name semantic parsing on each data item in the structured data metadata to obtain the association information of each data item;

[0165] The heterogeneous relationship graph acquisition module is used to extract entity relationships from all data items based on association information, identify dynamic association relationships between entities, entities and data, and data, and generate a weighted heterogeneous relationship graph.

[0166] The multi-layer dynamic flow map construction module is used to build a multi-layer dynamic flow map covering entities, data, users, systems, accounts, and roles based on heterogeneous relationship graphs.

[0167] The first classification and grading result acquisition module is used to obtain the first classification and grading result of the data to be analyzed based on the industry regulatory knowledge base.

[0168] The second classification and grading result acquisition module is used to obtain the second classification and grading results of the data to be analyzed based on the multi-layer dynamic flow map.

[0169] The target classification and grading result acquisition module is used to obtain the target classification and grading results of the data to be analyzed based on the first classification and grading results and the second classification and grading results.

[0170] A computer device according to embodiments of the present disclosure includes a memory and a processor. The memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0171] The processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the computer device to perform desired functions. In one embodiment of this disclosure, the processor is used to execute computer-readable instructions stored in the memory, causing the computer device to perform all or part of the steps of the data classification and grading method based on industry knowledge structure and relationships described in the foregoing embodiments of this disclosure.

[0172] Those skilled in the art will understand that, in order to solve the technical problem of how to achieve a good user experience, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included within the protection scope of this disclosure.

[0173] like Figure 5 This is a schematic diagram of a computer device provided for an embodiment of the present disclosure. It illustrates a structural schematic diagram suitable for implementing the computer device in the embodiments of the present disclosure. Figure 5 The computer device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0174] like Figure 5 As shown, a computer device may include a processor (such as a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) or programs loaded from storage devices into random access memory (RAM). The RAM also stores various programs and data required for the operation of the computer device. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0175] Typically, the following devices can be connected to the I / O interface: input devices, such as sensors or visual information acquisition devices; output devices, such as displays; storage devices, such as magnetic tapes or hard drives; and communication devices. Communication devices allow the computer device to communicate wirelessly or wiredly with other devices (such as edge computing devices) to exchange data. Although Figure 5A computer apparatus with various devices is shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or included alternatively.

[0176] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processor, all or part of the steps of the data classification and grading method based on industry knowledge structure and relationships according to embodiments of this disclosure are performed.

[0177] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0178] A computer-readable storage medium according to embodiments of the present disclosure stores non-transitory computer-readable instructions. When these non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the data classification and grading methods based on industry knowledge structures and relationships described in the foregoing embodiments of the present disclosure are performed.

[0179] The aforementioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or portable hard drive), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).

[0180] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0181] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0182] In this disclosure, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The block diagrams of devices, apparatuses, devices, and systems involved in this disclosure are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as "comprising," "including," "having," etc., are open-ended terms meaning "including but not limited to," and are used interchangeably with them. The terms "or" and "and" as used herein refer to the terms "and / or," and are used interchangeably with them unless the context clearly indicates otherwise. The term "such as" as used herein refers to the phrase "such as but not limited to," and is used interchangeably with it.

[0183] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0184] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0185] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0186] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0187] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A data classification and grading method based on industry knowledge structure and association relationship, characterized in that, The method comprises the following steps: grabbing raw data fragments in the business traffic of an enterprise and preprocessing the raw data fragments to obtain structured data element information; performing type inference and name semantic analysis on each data item in the structured data element information to obtain associated information of each data item; performing entity relationship extraction on all the data items according to the associated information to identify dynamic association relationships between entities-entities, entities-data, and data-data, and generate a heterogeneous relationship graph; based on the heterogeneous relationship graph, constructing a multi-layer dynamic flow transfer map covering entities, data, users, systems, accounts, and roles; obtaining a first classification and grading result of the data to be analyzed according to an industry regulation knowledge base; obtaining a second classification and grading result of the data to be analyzed according to the multi-layer dynamic flow transfer map; obtaining a target classification and grading result of the data to be analyzed based on the first classification and grading result and the second classification and grading result; the method of grabbing raw data fragments in the business traffic of an enterprise and preprocessing the raw data fragments to obtain structured data element information comprises the following steps: grabbing raw data fragments in the business traffic through bypass monitoring technology; based on a sliding window session reorganization algorithm, reorganizing all the raw data fragments into complete sessions corresponding to the business interaction process; for each data fragment in the complete session, adding a globally unified session ID, a client certificate fingerprint, a server certificate fingerprint, a user identifier, and a nanosecond-level timestamp to form structured data element information; the method of performing type inference and name semantic analysis on each data item in the structured data element information to obtain associated information of each data item comprises the following steps: constructing a multi-layer knowledge graph covering industry terms, synonyms, and abbreviations; the multi-layer knowledge graph comprises a core layer knowledge graph, an expansion layer knowledge graph, and a customized layer knowledge graph; using a BERT-BiLSTM-CRF cascade network to perform sequence labeling on the names of data items in the structured data element information to obtain entity boundaries; linking the entities in the entity boundaries with the nodes in the multi-layer knowledge graph, and using a PageRank algorithm to disambiguate the linking results to obtain semantic labels of each data item; extracting the data content of each data item in the structured data element information; training a target model using known type data samples, analyzing the data content of each data item through the trained model to obtain content type labels of each data item; the associated information comprises semantic labels and content type labels of data items.

2. The data classification and grading method based on industry knowledge structure and association relationship according to claim 1, characterized in that, the method of constructing a multi-layer dynamic flow transfer map covering entities, data, users, systems, accounts, and roles based on the heterogeneous relationship graph comprises the following steps: determining the hierarchical structure of the initial multi-layer map and the association mode between the levels; wherein the hierarchical structure comprises an entity layer, a data layer, a user layer, a system layer, an account layer, and a role layer; mapping the entity and data nodes in the heterogeneous relationship graph to the corresponding levels of the initial multi-layer map; constructing data flow transfer relationships in the initial multi-layer map according to the flow transfer path and mode of data between different entities, users, and systems; According to the business rules and the security policy, the operation permissions of different roles on entities and data are determined; The operation permissions are configured in the initial multi-layer map, and a constructed multi-layer dynamic flow transfer map is obtained. 3.The method of claim 1, wherein, The target classification grading result of the data to be analyzed is obtained based on the first classification grading result and the second classification grading result, including: If the first classification grading result is consistent with the second classification grading result, the corresponding classification grading result is taken as the target classification grading result; If the first classification grading result is inconsistent with the second classification grading result, the data type of the data to be analyzed is obtained; According to the data type, the corresponding industry knowledge structure weight and the correlation weight are determined; The classification grading result corresponding to the maximum weight in the industry knowledge structure weight and the correlation weight is obtained, and is taken as the target classification grading result of the data to be analyzed. 4.The method of claim 1, wherein, Further comprising: According to a preset period, the version management of the target classification grading result is triggered; Using an edit distance-based graph difference algorithm and a semantic embedding-based content difference algorithm, the data items and their sensitive level changes added, deleted or modified between adjacent versions are automatically identified, and a difference report is generated. 5.The method of claim 4, wherein, Further comprising: The difference report is submitted to user approval through a multi-role collaborative approval workflow; wherein the workflow is built-in an explainability presentation module based on role-attribute access control, and the approval process records complete decision trajectory and reason chain; The difference content confirmed by the approval is merged into the production environment through a zero-downtime gray release mechanism, triggering online incremental update.

6. A data classification and grading system based on industry knowledge structure and association relationship, characterized in that, Comprising: A preprocessing module configured to capture raw data segments in business traffic of an enterprise and preprocess the raw data segments to obtain structured data element information; An associated information acquisition module configured to perform type inference and name semantic analysis on each data item in the structured data element information to obtain associated information of each data item; A heterogeneous relationship graph acquisition module configured to perform entity relationship extraction on all the data items according to the associated information, identify dynamic correlation relationships between entity-entity, entity-data and data-data, and generate a weighted heterogeneous relationship graph; A multi-layer dynamic flow transfer map construction module configured to construct a multi-layer dynamic flow transfer map covering entities, data, users, systems, accounts and roles based on the heterogeneous relationship graph; A first classification grading result acquisition module configured to obtain a first classification grading result of data to be analyzed according to an industry regulation knowledge base; A second classification grading result acquisition module configured to obtain a second classification grading result of data to be analyzed according to the multi-layer dynamic flow transfer map; A target classification grading result acquisition module configured to obtain a target classification grading result of data to be analyzed based on the first classification grading result and the second classification grading result; The method comprises the following steps: capturing original data segments in business traffic of the enterprise and preprocessing the original data segments to obtain structured data element information, including capturing original data segments in business traffic through bypass monitoring technology; reorganizing all the original data segments into complete sessions corresponding to business interaction processes based on a sliding window session reorganization algorithm; attaching a global uniform session ID, a client certificate fingerprint, a server certificate fingerprint, a user identifier and a nanosecond-level timestamp to each data segment in the complete session to form structured data element information; The method comprises the following steps: performing type inference and name semantic analysis on each data item in the structured data element information to obtain associated information of each data item, including constructing a multi-layer knowledge graph covering industry terms, synonyms and abbreviations; the multi-layer knowledge graph comprises a core layer knowledge graph, an expansion layer knowledge graph and a customized layer knowledge graph; performing sequence labeling on the name of the data item in the structured data element information by using a BERT-BiLSTM-CRF cascade network to obtain an entity boundary; linking the entity in the entity boundary with a node in the multi-layer knowledge graph and performing disambiguation on the linking result by using a PageRank algorithm to obtain a semantic label of each data item; extracting data content of each data item in the structured data element information; training a target model by using a known type of data sample, analyzing the data content of each data item by using the trained model and obtaining a content type label of each data item; the associated information comprises the semantic label and the content type label of the data item.

7. A computer apparatus, comprising: The computer device comprises: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method for classifying and grading data based on industry knowledge structure and associated relationship according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing a computer to perform the method for classifying and grading data based on industry knowledge structure and associated relationship according to any one of claims 1-5.

Images