Data labeling method and device, equipment, medium and program product

Abstract

The invention provides a data labeling method and device, equipment, a medium and a program product, can be applied to the fields of financial science and technology and artificial intelligence, and relates to application of a large model in an intelligent labeling scene. The method comprises the following steps: extracting field information of a target database table, and generating a field structured table based on the field information; performing field correlation analysis on the field information to obtain a field analysis result; based on a field analysis result and a grading and classification standard, labeling a security grading label corresponding to the field structured table through an intelligent agent; and verifying the security grading label corresponding to the field structured table, and correcting the security grading label corresponding to the field structured table under the condition that the verification is not passed.

Description

Technical Field

[0001] This application relates to the fields of fintech and artificial intelligence, and to the application of large models in intelligent annotation scenarios. More specifically, it relates to a data annotation method, device, equipment, medium, and program product. Background Technology

[0002] Financial industry databases carry diverse data, including massive amounts of data resources such as system configuration data, customer data, business data, and operational management data. The definition and identification of the security level of data in the database are the core prerequisites for ensuring accurate control of data access permissions, compliance with regulatory requirements for privacy protection, and effective operation of risk prevention and control mechanisms.

[0003] Traditional database field security classification and grading often rely on manual labeling based on human experience or matching using static rule bases. This results in problems such as low efficiency, insufficient rule coverage, and poor dynamic adaptability, making it difficult to meet the increasingly complex data security needs of the financial industry. Summary of the Invention

[0004] In view of the above problems, this application provides data annotation methods, apparatus, equipment, media and program products to improve the level of intelligence.

[0005] According to a first aspect of this application, a data annotation method is provided, comprising: extracting field information from a target database table and generating a field structured table based on the field information; performing field association analysis on the field information to obtain field analysis results; annotating the field structured table with security classification labels corresponding to the field structured table using an intelligent agent based on the field analysis results and classification standards; and verifying the security classification labels corresponding to the field structured table, and correcting the security classification labels corresponding to the field structured table if the verification fails.

[0006] According to an embodiment of this application, the field analysis results include field heatmap values, field keywords, and field relationship graphs; the step of performing field association analysis on the field information to obtain field analysis results includes: extracting field metadata using a metadata analysis tool based on the field information; calculating the field heatmap value based on the field metadata; performing natural language processing on the field metadata to identify the field keywords; and constructing the field relationship graph based on the field metadata and the field keywords.

[0007] According to an embodiment of this application, calculating the field heat value based on the field metadata includes: obtaining the number of calls based on the access logs in the field metadata; and calculating the field heat value based on the number of calls, weight coefficient, and time decay factor.

[0008] According to an embodiment of this application, the method further includes: constructing the hierarchical classification standard based on an analysis model; wherein the hierarchical classification standard includes: hierarchical subclasses with hierarchical dependencies, security levels, sensitivity levels, and field definition descriptions; and there is a mapping relationship between the hierarchical subclasses, the security levels, the sensitivity levels, and the field definition descriptions.

[0009] According to an embodiment of this application, the security classification label includes classification subclass results, security level results, sensitivity level results, and field definition description results. The step of labeling the security classification label corresponding to the field structured table using an intelligent agent based on the field analysis results and classification standards includes: constructing task labeling rules based on constraint rules and task objectives; mapping the field information of the field structured table to the classification subclasses corresponding to the classification standards using the intelligent agent based on the field analysis results, and determining the classification subclass results; matching the security level results, sensitivity level results, and field definition description results corresponding to the classification subclass results through mapping; and labeling the field structured table using the intelligent agent based on the classification subclass results, security level results, sensitivity level results, and field definition description results.

[0010] According to an embodiment of this application, the verification of the security classification label corresponding to the field structured table includes: verifying the consistency between the security classification label and the classification standard one by one based on a preset verification dimension.

[0011] A second aspect of this application provides a data annotation apparatus, comprising: a structuring module for extracting field information from a target database table and generating a field structured table based on the field information; an association analysis module for performing field association analysis on the field information to obtain field analysis results; an annotation module for annotating the security classification labels corresponding to the field structured table using an intelligent agent based on the field analysis results and classification standards; and a verification and correction module for verifying the security classification labels corresponding to the field structured table and correcting the security classification labels corresponding to the field structured table if the verification fails.

[0012] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0013] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0014] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method.

[0015] In the embodiments of this application, field information is extracted to generate a structured table. Field association analysis is used to uncover potential relationships between fields. An intelligent agent, combined with hierarchical standard annotation, standardizes and organizes data and field relationships to provide a comprehensive basis for hierarchical classification. This overcomes the limitations of manual and static rules, improving the intelligence and accuracy of hierarchical classification. It achieves intelligent, efficient, dynamic, and accurate security classification and annotation of database table fields in the financial industry. The verification and correction process supports dynamic optimization of results, forming a closed-loop management system that ensures efficient batch annotation while adapting to the complex and dynamic data security needs of the financial industry. Attached Figure Description

[0016] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0017] Figure 1 The illustrations depict application scenarios of data annotation methods, apparatus, devices, media, and program products according to embodiments of this application.

[0018] Figure 2 A flowchart illustrating a data annotation method according to an embodiment of this application is shown schematically;

[0019] Figure 3 This illustration schematically shows a flowchart of field association analysis in the data annotation method according to an embodiment of this application;

[0020] Figure 4 This illustration schematically shows another flowchart of a data annotation method according to an embodiment of this application;

[0021] Figure 5 A schematic diagram illustrating the structure of a data annotation apparatus according to an embodiment of this application is shown; and

[0022] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a data annotation method according to an embodiment of this application. Detailed Implementation

[0023] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0025] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0026] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0027] As the digital transformation of the fintech industry continues to deepen, the market's emphasis on data security is also constantly increasing. Faced with the industry's rapid development, the standardized requirements for data security classification and protection have gradually become clear. Financial industry databases store massive amounts of data, including system configuration data, customer data, business data, and operational management data. The security level classification and labeling of this data directly relates to data access control, privacy protection compliance, and risk prevention capabilities. However, traditional database field security classification and labeling often relies on manual labeling based on experience or matching using static rule bases, which has some problems and is difficult to meet the increasingly complex data security needs of the financial industry.

[0028] The classification and labeling of database field security levels mainly rely on the following two methods: (1) manual labeling: security levels are labeled manually; (2) matching using a static rule base: fields are classified by predefined static rule bases (such as field name keyword matching, data type matching). For example, when the English name of a field contains "mobile phone" or "phone", it is labeled as "customer data-personal customer-natural information-basic information-contact information", with a security level of "level three" and a sensitivity level of "high sensitivity". Some systems combine regular expressions or pattern matching to identify sensitive information (such as ID number, bank card number) in the field content.

[0029] In practical applications, existing data security level labeling methods still have the following drawbacks:

[0030] (1) Inefficient: Manual annotation requires field-by-field analysis and annotation, which is time-consuming and difficult to adapt to the security level annotation requirements of massive financial data.

[0031] (2) Insufficient rule coverage: Static rule bases are difficult to adapt to complex and ever-changing field naming conventions and business scenarios.

[0032] (3) Poor dynamic adaptability: unable to respond in real time to regulatory updates or changes in business needs.

[0033] This application provides a data annotation method that extracts field information from a target database table and generates a structured table based on this information. It then performs field association analysis on the field information to obtain analysis results. Based on the analysis results and classification standards, an intelligent agent annotates the structured table with corresponding security classification labels. Finally, it verifies the security classification labels, correcting them if the verification fails. In this application, the method extracts field information to generate a structured table, utilizes field association analysis to uncover potential field relationships, and uses an intelligent agent combined with classification standards for annotation. This standardized approach to data and field relationships provides a comprehensive basis for classification, overcoming the limitations of manual and static rules, improving the intelligence and accuracy of classification, and achieving intelligent, efficient, dynamic, and accurate security classification and annotation of database table fields in the financial industry. The verification and correction process supports dynamic optimization of results, forming a closed-loop management system that ensures efficient batch annotation while adapting to the complex and dynamic data security needs of the financial industry.

[0034] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0035] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this application all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0036] It should be noted that the data annotation method and apparatus of this application can be used in the fields of fintech and artificial intelligence, involving the application of large models in intelligent annotation scenarios, and can also be used in any field other than fintech and artificial intelligence. The application fields of the data annotation method and apparatus of this application are not limited.

[0037] Figure 1 The illustrations depict application scenarios of data annotation methods, apparatus, devices, media, and program products according to embodiments of this application.

[0038] like Figure 1 As shown, application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0039] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0040] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0041] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0042] It should be noted that the data annotation method provided in this application embodiment can generally be executed by server 105. Correspondingly, the data annotation device provided in this application embodiment can generally be located in server 105. The data annotation method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the data annotation device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0043] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0044] The following will be based on Figure 1 The described scene, through Figures 2-4 The data annotation method according to the embodiments of this application will be described in detail.

[0045] Figure 2 A flowchart illustrating a data annotation method according to an embodiment of this application is shown schematically.

[0046] like Figure 2 As shown, the data annotation method in this embodiment includes operations S210 to S240. This data annotation method is not limited to a specific executing entity. The executing entity can be any electronic device, such as a terminal device or a server device. The executing entity can also be any software application or client.

[0047] In operation S210, the field information of the target database table is extracted, and a field structured table is generated based on the field information.

[0048] Based on the target database type, extract the target table field information using database management tools or query tools. The field information includes core attributes such as the Chinese name of the field, the English name of the field, the data type, the field description, and the default value. Remove duplicates from the extracted raw information and fill in any missing items (such as adding unspecified field descriptions). Organize the field information into a structured table according to a unified format, with each field attribute corresponding to a table column, ensuring that the structured table is consistent with the field information of the source table.

[0049] A target database table is a structured collection that stores specific business data. It consists of several fields, each of which defines the data storage rules and is the basic unit for storing data in the database.

[0050] A structured table is a standardized organization of field information from a target database table. It clearly presents the core attributes of each field in tabular form, enabling visualized and standardized management of field information. This allows developers, testers, and operations personnel to quickly access field details and supports data integration, system development, and other tasks.

[0051] In operation S220, field association analysis is performed on the field information to obtain the field analysis results.

[0052] The field analysis results include field heatmap values, field keywords, and field relationship graphs. When performing field association analysis, the field metadata is first extracted using a metadata analysis tool. Then, the heatmap value is calculated by combining the number of calls, weight coefficients, and time decay factors. Next, natural language processing is performed on the metadata to extract keywords. Finally, based on the metadata and keywords, the business associations of the fields are mined, and a field relationship graph containing nodes and edges is constructed.

[0053] When operating S230, based on the field analysis results and hierarchical classification standards, the security classification labels corresponding to the field structured table are labeled by the intelligent agent.

[0054] The hierarchical classification standard represents the criteria for classifying and categorizing field security levels. It is constructed by the analysis model and can be provided in tabular form. The hierarchical classification standard includes field security subcategories from level one to level five, security level, sensitivity level, and field definition description. Note that there are strict dependencies between the level one to level five subcategories, which can also be represented by a tree diagram. The level five subcategories and sensitivity level correspond to a unique security level and field definition description.

[0055] The field analysis results and classification standards from the preparation phase are provided to the agent. The agent can then select a suitable large model for intelligent judgment and output the labeled database table fields and their corresponding security classification labels, such as five-level subclasses, security level, sensitivity level, and field definition descriptions.

[0056] An intelligent agent is an artificial intelligence agent program with natural language understanding and task execution capabilities. It can analyze and process documents and natural language, dynamically adapt to complex scenarios and continuously optimize its own performance. It has autonomous perception, decision-making and execution capabilities. Based on preset logic and model capabilities, it completes the automated hierarchical classification and labeling tasks of data fields.

[0057] The intelligent agent comprises an input perception layer, a policy decision layer, and an execution output layer. The input perception layer receives structured / unstructured information from the preparation phase, such as field analysis results, classification standards, and task labeling rules, and parses and stores it. The policy decision layer autonomously matches or selects a suitable large model (such as a text classification model, rule matching model, or general language model) based on the characteristics of the input content. The execution output layer calls the selected large model to intelligently analyze the database table fields, ultimately outputting results with security classification labels (five-level subclass, security level, sensitivity level, field definition description, etc.), and can iteratively optimize the decision logic based on labeling feedback.

[0058] In operation S240, the security classification label corresponding to the field structure table is verified. If the verification fails, the security classification label corresponding to the field structure table is corrected.

[0059] Based on the preset verification dimensions, the security classification labels of the structured table are compared one by one with the classification standards. The consistency of core information such as classification subclasses and security levels is checked. Items that fail the verification (such as mismatched subclass names, incorrect level mapping, or missing field annotations) are marked, and an anomaly details report is generated simultaneously to clarify the anomaly type and inconsistency information.

[0060] If the verification passes, the security classification label corresponding to the field structure table output by the agent will be used as the final field annotation information.

[0061] If the verification fails, the label deviation information is located, and the erroneous labels are adjusted according to the needs of the financial business scenario and business rules, and missing labels are supplemented. At the same time, the rule defects exposed during the labeling process are reviewed, the natural language description of the task labeling rules is updated, and the hierarchical mapping relationship, sensitivity judgment threshold, and other contents in the classification standard are supplemented or adjusted.

[0062] The corrected labels are re-verified, archived after passing the verification, and the optimized rules and standards are then distributed to the intelligent agent for subsequent field classification labeling, forming a dynamic closed loop of verification, correction, optimization, and reuse, continuously improving the accuracy and adaptability of financial database field security classification.

[0063] It is worth noting that during the design phase of some application systems, designers can follow this application to classify and label the database table fields according to their security levels. This allows for differentiated security protection and management measures to be taken for fields with different security levels based on their security classification labels.

[0064] The existing database table fields are classified and labeled with security levels and sensitivity levels. After the labeling is completed, they are imported into the metadata resource information management system in batches. Subsequently, different access permissions can be set according to the field security level in the system to restrict the operation scope of the access subject.

[0065] In the embodiments of this application, field information is extracted to generate a structured table. Field association analysis is used to uncover potential relationships between fields. An intelligent agent, combined with hierarchical standard annotation, standardizes and organizes data and field relationships to provide a comprehensive basis for hierarchical classification. This overcomes the limitations of manual and static rules, improving the intelligence and accuracy of hierarchical classification. It achieves intelligent, efficient, dynamic, and accurate security classification and annotation of database table fields in the financial industry. The verification and correction process supports dynamic optimization of results, forming a closed-loop management system that ensures efficient batch annotation while adapting to the complex and dynamic data security needs of the financial industry.

[0066] Figure 3 The flowchart illustrating the field association analysis of the data annotation method according to an embodiment of this application is shown in the illustration.

[0067] like Figure 3 As shown, the field analysis results include field heat values, field keywords, and field relationship graphs. Operation S220 performs field association analysis on the field information, and the field analysis results include operations S310 to S340.

[0068] When operating S310, based on field information, the field metadata is extracted using a metadata analysis tool.

[0069] Extract field metadata from the target table using metadata analysis tools, such as field name (hkname), Chinese name (customer name), data type (string), encryption method (national cryptographic SM4), associated systems (core business system, transaction system), and access logs (application call count, call method).

[0070] Based on the field information of the target database table, identify the metadata dimensions to be extracted (field name, Chinese name, data type, etc.), configure the database connection parameters (address, port, account password) of the metadata analysis tool, and establish a communication link between the tool and the target database.

[0071] Metadata analysis tools extract basic metadata by parsing database system tables, data dictionaries, and field comments. They then correlate this metadata with data encryption configuration files and system integration registry entries to complete extended attributes such as encryption methods and system associations. Finally, they connect to the database access log system to statistically analyze access characteristics such as application call counts and call methods. After deduplication and format standardization, a structured set of field metadata is output.

[0072] When operating S320, calculate the field heat value based on the field metadata.

[0073] According to an embodiment of this application, in operation S320, calculating the field heat value based on field metadata, the steps include: obtaining the number of calls based on the access log in the field metadata; and calculating the field heat value based on the number of calls, weight coefficient, and time decay factor.

[0074] The field heat value indicates the frequency of field access, helping to determine the usage scenario and sensitivity. The formula for calculating the field heat value is as follows:

[0075]

[0076] Where h is the heat value and c is the number of calls (representing the number of times the application programming interface is called). t is the weighting coefficient and t is the time decay factor.

[0077] In the embodiments of this application, the number of calls is extracted by combining the access logs of field metadata, and weight coefficients and time decay factors are introduced to calculate the heat value. This not only quantifies the actual activity level of the field, but also takes into account the weight and timeliness of different scenarios, providing a dynamic and objective heat dimension basis for the intelligent agent to accurately label the security level.

[0078] When operating S330, natural language processing is performed on the field metadata to identify field keywords.

[0079] The natural language processing process is as follows: First, preprocess the Chinese name / description text in the field metadata, remove redundant punctuation and stop words (such as "of", "and"), and complete word segmentation (using a word segmentation tool); then screen nouns and noun phrases through词性标注 (such as "customer name" being segmented into "customer", "name"); subsequently, use a named entity recognition model to identify core entities (such as "customer", "name", "mobile phone number"); finally, combine with a business dictionary for keyword filtering and weight ranking, and extract keywords that can represent the core meaning of the field (such as "customer name" corresponding to keywords "customer", "name") to ensure that the keywords accurately match the business attributes of the field.

[0080] In operation S340, based on the field metadata and field keywords, construct a field relationship graph.

[0081] In the field relationship graph, use the field keywords as the core nodes, associate metadata attributes (such as data type, encryption method, etc.) and keywords as attribute nodes, and business associations (such as the attribution association between customer name and customer identifier) as edges.

[0082] First, store the field metadata and the extracted keywords in a structured manner and establish a unique field identifier. Second, use an association rule algorithm to mine business associations between fields (such as associations between fields in the same business domain, foreign key dependencies), and combine keyword similarity matching (such as clustering of fields related to "customer") to determine the association type. Finally, define the node attributes (field name, keyword, security level, etc.) and the relationship labels of the edges (attribution, dependency, association), and use a graph tool for visual modeling. After verifying and removing invalid associations, generate a complete field relationship graph.

[0083] In the embodiments of this application, through hierarchical operations of metadata analysis, heat value calculation, natural language processing, and graph construction, quantify the field heat, refine core keywords, and clarify field associations, providing multi-dimensional and refined analysis support for accurate annotation by the intelligent agent, and enhancing the scientificity and rationality of field classification and grading.

[0084] According to the embodiments of this application, the security classification label includes classification subclass results, security level results, sensitivity level results, and field definition description results; in operation S230, based on the field analysis results and classification and grading criteria, when the intelligent agent annotates the security classification label corresponding to the field structured table, it includes: constructing task annotation rules based on constraint rules and task objectives; based on the field analysis results, mapping the field information of the field structured table to the classification subclass corresponding to the classification and grading criteria through the intelligent agent to determine the classification subclass results; mapping and matching the security level results, sensitivity level results, and field definition description results corresponding to the classification subclass results; and annotating the field structured table through the intelligent agent according to the classification subclass results, security level results, sensitivity level results, and field definition description results.

[0085] Task annotation rules are standardized descriptions, providing unified task instructions for intelligent agents to clearly define analysis requirements, output formats, and constraint rules. Task annotation rules include constraint rules, task objectives, and output requirements.

[0086] Task objective: Based on the structured table of fields and hierarchical classification standards, analyze the five-level subclasses, security level, sensitivity level, and field definition descriptions of all-numeric fields.

[0087] Constraint rules: Strictly match category names in the standard (no self-created names allowed), analyze all fields (no omissions allowed), analyze fields independently (no associations allowed), and refer to other metadata in the table structure when there are no keywords (no creation of keywords allowed).

[0088] The role of the intelligent agent can be defined as a data security level analyst or a tag annotation analyst, specifying the dimensions to be output (first to fifth level subclasses, security level, sensitivity level, and field definition description).

[0089] For example, the prompt input by the agent is: You are a data security level analyst. Task: Based on all information in the table structure and the field security classification standards, analyze the data security classification and security level of each field in this table. Output: Data security level subclasses (Level 1, Level 2, Level 3, Level 4, Level 5), data security level, and data sensitivity level. Requirements: Data security level classification must strictly adhere to the defined names in the data security classification rules; creating new classification names is not allowed. All table fields must be analyzed; abbreviations or omissions are not allowed. There should be no correlation between fields; each field needs to be analyzed independently. If no matching keyword is found, refer to information from other columns in the table structure; creating keywords out of thin air is not allowed. Based on the field security classification standard, the agent infers from the analysis results that each table field is mapped to the five-level subclass system of the security classification standard, and matches the corresponding security level and sensitivity level. Finally, it outputs standardized results according to the task labeling rules (e.g., the "Customer Name" field: Level 1 subclass "Party Member", Level 2 subclass "Customer", Level 3 subclass "Natural Information", Level 4 subclass "Basic Information", Level 5 subclass "Name", Security Level "Level 3", Sensitivity Level "Medium Sensitive"). The field definition description is "Refers to personal basic information data, name".

[0090] In the embodiments of this application, by constructing task annotation rules, the intelligent agent maps field information to the hierarchical subclasses of the hierarchical classification standard based on the field analysis results, and then matches the corresponding security and sensitivity levels and field definition descriptions to complete the annotation, forming a standardized mapping link, eliminating the subjective limitations of human experience, getting rid of the limitations of the annotation mode dominated by human experience, and improving the accuracy and standardization level of security classification annotation of financial database fields.

[0091] According to an embodiment of this application, the operation S240 verifies the security classification labels corresponding to the structured table, including: verifying the consistency between the security classification labels and the classification standards one by one based on a preset verification dimension.

[0092] Rule-based backtracking verification is used to check the consistency between the field annotation results and the hierarchical classification standards one by one. The verification dimensions include: whether the names of the five-level subclasses match the standards, whether the security level / sensitivity level is uniquely mapped to the five-level subclasses, and whether any fields are missing.

[0093] Based on the validation dimensions, the process first verifies whether the names of the five-level subclasses completely match the hierarchical classification standards. Then, it verifies whether the security level and sensitivity level form a unique mapping relationship with their corresponding subclasses. Finally, it checks all fields to confirm that there are no missing annotations or invalid tags. Fields that fail validation are marked as abnormal, and a validation report is generated for review and correction.

[0094] In the embodiments of this application, verifying the consistency between the safety classification label and the classification standard can promptly detect deviations between the label and the standard, ensure the compliance and accuracy of the labeling results, facilitate timely adjustment of labels that do not conform to the classification standard, and improve the labeling quality.

[0095] Figure 4 Another flowchart illustrating a data annotation method according to an embodiment of this application is shown schematically.

[0096] like Figure 4 As shown, the data annotation method in this embodiment includes operations S410 to S450.

[0097] When operating S410, a hierarchical classification standard is constructed based on the analysis model. The hierarchical classification standard includes: hierarchical subclasses with hierarchical dependencies, security levels, sensitivity levels, and field definition descriptions. There is a mapping relationship between hierarchical subclasses, security levels, sensitivity levels, and field definition descriptions.

[0098] We collect massive amounts of historical business data, and through analytical models, we mine data association rules, business scenario characteristics, and leakage risk cases. Based on core dimensions such as data subjects and business domains, we automatically cluster and generate hierarchical subclasses with hierarchical dependencies. Then, combined with the scope and severity of the impact of data leakage, we construct a security level and sensitivity level assessment system. Finally, we establish a mapping relationship between hierarchical subclasses, security levels, sensitivity levels, and field definition descriptions. Through iterative optimization and verification of the mapping rationality through analytical models, we form a hierarchical classification standard that is adapted to actual business needs and can be implemented, ensuring coverage of all business data scenarios.

[0099] The analytical model integrates data mining, machine learning, and rule engine capabilities. Inputs are raw business data and field information, and outputs standardized hierarchical classification criteria. The analytical model includes, but is not limited to, clustering models (used to automatically divide data subjects, business domains, and other hierarchical subcategories), risk assessment models (determining security / sensitivity levels based on data leakage risks and usage scenarios), and association rule models (establishing mapping relationships between hierarchical subcategories and field definitions).

[0100] For example, by analyzing the business databases of various transaction systems using an analytical model, a hierarchical classification standard is derived, including five levels of subcategories, four levels of security, three levels of sensitivity, and field definition descriptions. The first level subcategories are data subject types (e.g., parties, institutions, transaction entities); the second level subcategories are business domain affiliations (e.g., customers, accounts, transactions, risk control); the third level subcategories are data nature classifications (e.g., identifiers, attributes, behaviors, derivatives); the fourth level subcategories are usage scenario dimensions (e.g., authentication, verification, analysis, disclosure); and the fifth level subcategories are specific field types (e.g., direct identifiers, indirect identifiers, sensitive attributes). The first level of security is for publicly available data (e.g., industry benchmark values); the second level is for internally shared data (e.g., product codes); the third level is for sensitive data (e.g., contact information, transaction records); and the fourth level is for core confidential data (e.g., biometrics, key materials). Sensitivity levels include low sensitivity, medium sensitivity, and high sensitivity.

[0101] The hierarchical classification standard is a hierarchical and mappable security classification system. It follows a hierarchical dependency relationship: Data Subject Type (Level 1) → Business Domain Affiliation (Level 2) → Data Nature Classification (Level 3) → Usage Scenario Dimension (Level 4) → Specific Field Type (Level 5). Each level subclass has clearly defined classification items; for example, Level 1 subclasses include "Party / Institution / Transaction Entity," and Level 5 subclasses include "Direct Identifier / Indirect Identifier / Sensitive Attribute." Security levels (Levels 1 to 4, e.g., Level 1 = Public Data, Level 4 = Core Confidentiality) are assigned, and each Level 5 subclass is matched with a unique security level, sensitivity level (e.g., High Sensitivity), and field definition description, forming a unique mapping relationship: Level 5 subclass → Level 4 security level → Level 3 sensitivity level → field definition description.

[0102] In the embodiments of this application, a hierarchical classification standard containing hierarchical subclasses and security levels is constructed based on the analysis model, and a mapping relationship between each element is established, providing a standardized and systematic basis for judgment and improving the standardization and accuracy of the hierarchical classification of financial database fields.

[0103] In operation S420, field information from the target database table is extracted, and a structured table is generated based on this field information. It should be noted that operation S420 is the same as operation S210 described earlier, and will not be repeated here.

[0104] In operation S430, field association analysis is performed on the field information to obtain the field analysis results. It should be noted that the method for operation S430 is the same as that for operation S220 described earlier, and will not be repeated here.

[0105] In operation S440, based on the field analysis results and hierarchical classification standards, the security classification labels corresponding to the field structured table are labeled by the intelligent agent. It should be noted that the method for operating S440 is the same as that for operation S230 described earlier, and will not be repeated here.

[0106] In operation S450, the security classification label corresponding to the structured table of the field is validated. If the validation fails, the security classification label corresponding to the structured table of the field is corrected. It should be noted that operation S450 is the same as operation S240 described above, and will not be repeated here.

[0107] Based on the above data annotation method, this application also provides a data annotation apparatus. The following will combine... Figure 5 The device is described in detail.

[0108] Figure 5 A schematic block diagram of a data annotation apparatus according to an embodiment of this application is shown.

[0109] like Figure 5 As shown, the data annotation device 800 of this embodiment includes a structuring module 810, an association analysis module 820, an annotation module 830, and a verification and correction module 840.

[0110] The structuring module 810 is used to extract field information from the target database table and generate a field structured table based on the field information. In one embodiment, the structuring module 810 can be used to perform the operation S210 described above, which will not be repeated here.

[0111] The association analysis module 820 is used to perform field association analysis on the field information to obtain field analysis results. In one embodiment, the association analysis module 820 can be used to perform the operation S220 described above, which will not be repeated here.

[0112] The annotation module 830 is used to annotate the security classification labels corresponding to the field structured table through an intelligent agent based on the field analysis results and classification criteria. In one embodiment, the annotation module 830 can be used to perform the operation S230 described above, which will not be repeated here.

[0113] The verification and correction module 840 is used to verify the security classification label corresponding to the field structured table. If the verification fails, the security classification label corresponding to the field structured table is corrected. In one embodiment, the verification and correction module 840 can be used to perform the operation S240 described above, which will not be repeated here.

[0114] According to an embodiment of this application, the field analysis results include field heatmap values, field keywords, and field relationship graphs; the association analysis module 820 includes: a metadata extraction unit, used to extract field metadata using a metadata analysis tool based on the field information; a heatmap value calculation unit, used to calculate the field heatmap value based on the field metadata; a natural language processing unit, used to perform natural language processing on the field metadata to identify the field keywords; and a graph construction unit, used to construct the field relationship graph based on the field metadata and the field keywords.

[0115] According to an embodiment of this application, the heat value calculation unit includes: a call count acquisition subunit, used to acquire the call count based on the access log in the field metadata; and a calculation subunit, used to calculate the field heat value based on the call count, weight coefficient, and time decay factor.

[0116] According to an embodiment of this application, the device 800 further includes: a standard construction module, used to construct the hierarchical classification standard based on the analysis model; wherein, the hierarchical classification standard includes: hierarchical subclasses with hierarchical dependencies, security levels, sensitivity levels, and field definition descriptions; there is a mapping relationship between the hierarchical subclasses, the security levels, the sensitivity levels, and the field definition descriptions.

[0117] According to an embodiment of this application, the security classification label includes classification subclass results, security level results, sensitivity level results, and field definition description results; the annotation module 830 includes: a rule construction unit, used to construct task annotation rules based on constraint rules and task objectives; a field mapping unit, used to map the field information of the field structured table to the classification subclass corresponding to the classification standard through the intelligent agent based on the field analysis results, and determine the classification subclass results; a mapping matching unit, used to match the security level results, sensitivity level results, and field definition description results corresponding to the classification subclass results through mapping; and an annotation unit, used to annotate the field structured table through the intelligent agent according to the classification subclass results, the security level results, the sensitivity level results, and the field definition description results.

[0118] According to an embodiment of this application, the verification and correction module 840 includes: a step-by-step verification unit, used to verify the consistency between the security classification label and the classification standard one by one based on a preset verification dimension.

[0119] According to embodiments of this application, any multiple modules among the structured module 810, correlation analysis module 820, annotation module 830, verification and correction module 840, and standard construction module can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the structured module 810, correlation analysis module 820, annotation module 830, verification and correction module 840, and standard construction module can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the structure module 810, the association analysis module 820, the annotation module 830, the verification and correction module 840, and the standard construction module can be implemented at least partially as a computer program module, which can perform corresponding functions when the computer program module is run.

[0120] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a data annotation method according to an embodiment of this application.

[0121] like Figure 6 As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0122] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that the programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0123] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0124] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0125] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0126] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the data annotation method provided in the embodiments of this application.

[0127] When the computer program is executed by the processor 901, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0128] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0129] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of this application embodiment. According to the embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0130] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0131] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0132] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A data labeling method, characterized by, The method comprises: extracting field information of a target database table, and generating a field structured table based on the field information; performing field correlation analysis on the field information to obtain a field analysis result; based on the field analysis result and a hierarchical classification standard, labeling a security hierarchical label corresponding to the field structured table by an intelligent agent; and verifying the security hierarchical label corresponding to the field structured table, and correcting the security hierarchical label corresponding to the field structured table if the verification fails.

2. The method of claim 1, wherein, The field analysis result comprises a field heat value, a field keyword, and a field relationship graph; the field correlation analysis on the field information to obtain a field analysis result comprises: extracting field metadata based on the field information by a metadata analysis tool; calculating the field heat value based on the field metadata; performing natural language processing on the field metadata to identify the field keyword; and constructing the field relationship graph based on the field metadata and the field keyword.

3. The method of claim 2, wherein, The calculation of the field heat value based on the field metadata comprises: obtaining a calling frequency according to access logs in the field metadata; and calculating the field heat value based on the calling frequency, a weight coefficient, and a time decay factor.

4. The method of claim 1, wherein, The method further comprises: constructing the hierarchical classification standard based on an analysis model; wherein the hierarchical classification standard comprises a hierarchical subclass of a hierarchical dependency relationship, a security level, a sensitive level, and a field definition description; the hierarchical subclass, the security level, the sensitive level, and the field definition description have a mapping relationship.

5. The method of claim 1, wherein, The security hierarchical label comprises a hierarchical subclass result, a security level result, a sensitive level result, and a field definition description result; the labeling of the security hierarchical label corresponding to the field structured table by the intelligent agent based on the field analysis result and the hierarchical classification standard comprises: constructing a task labeling rule based on a constraint rule and a task target; mapping field information of the field structured table to a hierarchical subclass corresponding to the hierarchical classification standard by the intelligent agent based on the field analysis result, to determine the hierarchical subclass result; mapping matching the security level result, the sensitive level result, and the field definition description result corresponding to the hierarchical subclass result; and labeling the field structured table by the intelligent agent according to the hierarchical subclass result, the security level result, the sensitive level result, and the field definition description result.

6. The method of claim 1, wherein, The verification of the security hierarchical label corresponding to the field structured table comprises: verifying the consistency of the security hierarchical label and the hierarchical classification standard one by one based on a preset verification dimension.

7. A data labeling apparatus, comprising: The device comprises: a structured module configured to extract field information of a target database table, and generate a field structured table based on the field information; an association analysis module configured to perform field correlation analysis on the field information to obtain a field analysis result; a labeling module configured to label a security hierarchical label corresponding to the field structured table by an intelligent agent based on the field analysis result and a hierarchical classification standard; and The check and correction module is configured to check the security classification label corresponding to the field structured table, and correct the security classification label corresponding to the field structured table if the check fails. 8.An electronic device comprising: one or more processors; memory storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-6.

10. A computer program product comprising computer programs or instructions, characterized in that, The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-6. The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-6.

Images