Data classification and grading method, device and equipment and readable storage medium

Through large models, the problem of large-scale investment and low efficiency of data classification and grading resources in the existing technology is solved, and efficient and accurate data classification and grading is achieved, adapting to the needs of different industries, and reducing manual rules and labeling is dependency.

CN119004295APending Publication Date: 2024-11-22HANGZHOU DBAPPSECURITY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411263926.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing data classification and grading methods require a large number of manual rules and labels, with huge investment in resources, long cycles, high requirements for personnel, and highly dependent on data governance results, resulting in inefficiency.

Method used

Using large models for data classification and grading, and by obtaining the data classification framework of the target industry, using large models for hierarchical classification, gradually obtaining classification labels and splicing, reducing dependence on manual rules and labels, and improving the accuracy and efficiency of classification results. .

Benefits of technology

It significantly improves the accuracy and efficiency of data classification and grading, reduces cost investment, and adapts to the data classification and grading needs of different industries, without repeated training as the industry framework changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004295A_ABST
    Figure CN119004295A_ABST
Patent Text Reader

Abstract

The invention discloses a data classification and grading method, device and equipment and a readable storage medium. The data classification and grading method comprises the steps that all topmost-level classification labels contained in topmost-level subclasses in an industry data classification and grading framework corresponding to a to-be-classified field and topmost-level classification descriptions corresponding to the topmost-level classification labels are obtained; classifying the to-be-classified fields according to the to-be-classified fields, the topmost-level classification labels and the topmost-level classification descriptions by utilizing a large model; obtaining a next-level classification description of a next-level classification label corresponding to each next-level subclass corresponding to the current-level classification label; classifying the to-be-classified fields according to the to-be-classified fields, the classification labels of the current layer, the classification labels of the next layers and the classification descriptions of the next layers by utilizing a large model; and splicing the classification labels of each layer to obtain a new classification label of the current layer, and when the subclass of the next layer does not exist, determining the classification label of the current layer as a data classification result and grading to obtain a data grading result. The accuracy of the classification and grading result is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data classification and grading method, device, equipment and computer-readable storage medium. Background Art

[0002] Against the backdrop of the increasing importance of data security, a data classification, grading and protection system has been established, which explicitly requires all industries and departments to manage data in a classified and graded manner.

[0003] However, most of the current classification and grading methods use regular expressions, keyword matching methods, or training classification models based on expert-labeled data. These methods rely on experienced personnel to accumulate rules and keywords, or provide labeled training sets. Since the classification and grading frameworks of different industries are different, the rules and training data of different industries are also different, resulting in large investments and huge amounts of classification and grading data. If only relying on manual accumulation of rules or labeling of training sets, the resource investment is huge and the cycle is long. It has high requirements for personnel, requiring them to be familiar with the business and understand the standards and specifications for data classification and grading. It is highly dependent on data governance results. If the customer has not carried out metadata governance, data standard governance and other work, the classification and grading effect is not ideal.

[0004] To sum up, how to effectively solve the problems of current data classification and grading methods, such as huge resource investment, long cycle, high personnel requirements, and high dependence on data governance results, is an issue that technical personnel in this field urgently need to solve. Summary of the invention

[0005] The purpose of this application is to provide a data classification and grading method, which improves the accuracy of classification results, effectively reduces the reliance on manual rules and annotations, greatly saves costs, and significantly improves the efficiency of data classification and grading; another purpose of this application is to provide a data classification and grading device, equipment and computer-readable storage medium.

[0006] In order to solve the above technical problems, this application provides the following technical solutions:

[0007] A data classification and grading method, comprising:

[0008] Receive a field to be classified, and obtain a target industry data classification and grading framework corresponding to the field to be classified;

[0009] Obtain each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label;

[0010] Using the large model, the field to be classified is classified according to the field to be classified, each top-level classification label and each top-level classification description to obtain a current-level classification label;

[0011] Obtaining next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtaining next-level classification descriptions corresponding to each next-level classification label;

[0012] Using the large model, classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description to obtain a target next layer classification label;

[0013] The current layer classification label and the target next layer classification label are concatenated to obtain a new current layer classification label, and the step of obtaining the next layer classification labels corresponding to each next layer subclass corresponding to the current layer classification label is repeated until there is no next layer subclass, and the current layer classification label is determined as the data classification result corresponding to the field to be classified;

[0014] A hierarchical mapping is performed based on the target industry data classification and grading framework and the data classification result to obtain a data grading result.

[0015] In a specific implementation of the present application, it includes:

[0016] Receive the fields to be classified including data source, database name, table name, table comment, field name, field comment, and field sample content; wherein the table name and the field name are not empty.

[0017] In a specific implementation of the present application, the large model is used to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description, including:

[0018] Obtain the field in the same table corresponding to the field to be classified;

[0019] Determine the same table field as the background field of the field to be classified;

[0020] The large model is used to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label, each next layer classification description and the background field.

[0021] In a specific implementation of the present application, after obtaining the data classification result, the method further includes:

[0022] Determine the target permission protection measures corresponding to the to-be-classified fields according to the data classification results;

[0023] The target permission protection measure is used to perform data protection on the field to be classified.

[0024] In a specific implementation of the present application, determining the current layer classification label as the data classification result corresponding to the field to be classified includes:

[0025] Obtaining a classification description corresponding to a current-layer classification label, and concatenating the classification description with the current-layer classification label to obtain a new current-layer classification label;

[0026] The current layer classification label is determined as the data classification result corresponding to the field to be classified.

[0027] A data classification and grading device, comprising:

[0028] A classification and grading framework acquisition module is used to receive the field to be classified and acquire the target industry data classification and grading framework corresponding to the field to be classified;

[0029] A first label and description acquisition module is used to acquire each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label;

[0030] A current layer classification label obtaining module is used to classify the to-be-classified field according to the to-be-classified field, each top-level classification label and each top-level classification description using a large model to obtain a current layer classification label;

[0031] A second label and description acquisition module is used to acquire the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and acquire the next-level classification descriptions corresponding to each next-level classification label;

[0032] A next-level classification label obtaining module is used to classify the field to be classified according to the field to be classified, the current-level classification label, each next-level classification label and each next-level classification description using the large model to obtain a target next-level classification label;

[0033] A classification result acquisition module is used to concatenate the current layer classification label and the target next layer classification label to obtain a new current layer classification label, and repeatedly execute the step of obtaining the next layer classification labels corresponding to each next layer subclass corresponding to the current layer classification label until there is no next layer subclass, and then determine the current layer classification label as the data classification result corresponding to the field to be classified;

[0034] The grading result acquisition module is used to perform grading mapping according to the target industry data classification and grading framework and the data classification results to obtain the data grading results.

[0035] In a specific implementation of the present application, the next level classification label acquisition module includes:

[0036] The same table field acquisition submodule is used to acquire the same table field corresponding to the field to be classified;

[0037] A background field determination submodule, used for determining the same table field as the background field of the field to be classified;

[0038] The field classification submodule is used to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label, each next layer classification description and the background field by using the large model.

[0039] In a specific implementation of the present application, the classification result obtaining module includes:

[0040] A classification label concatenation submodule is used to obtain a classification description corresponding to a current-layer classification label, and concatenate the classification description with the current-layer classification label to obtain a new current-layer classification label;

[0041] The classification result obtaining submodule is used to determine the current layer classification label as the data classification result corresponding to the field to be classified.

[0042] A data classification and grading device, comprising:

[0043] Memory for storing computer programs;

[0044] A processor is used to implement the steps of the data classification and grading method as described above when executing the computer program.

[0045] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the data classification and grading method as described above.

[0046] The data classification and grading method provided in the present application receives a field to be classified, and obtains a target industry data classification and grading framework corresponding to the field to be classified; obtains each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and each top-level classification description corresponding to each top-level classification label; uses a large model to classify the field to be classified according to the field to be classified, each top-level classification label and each top-level classification description to obtain a current-level classification label; obtains the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtains the next-level classification description corresponding to each next-level classification label; Use the big model to classify the fields to be classified according to the fields to be classified, the current layer classification labels, each next-level classification labels and each next-level classification descriptions to obtain the target next-level classification labels; concatenate the current layer classification labels and the target next-level classification labels to obtain new current layer classification labels, and repeat the steps of obtaining the next-level classification labels corresponding to each next-level subclass corresponding to the current layer classification label until there are no next-level subclasses, then determine the current layer classification label as the data classification result corresponding to the field to be classified; perform hierarchical mapping according to the target industry data classification and grading framework and the data classification results to obtain the data grading results.

[0047] It can be seen from the above technical solution that by adopting the large model prompt engineering technology, the specific meaning of the field is first inferred, and then hierarchical multi-step classification is performed, and the final data classification result is obtained by splicing the classification labels corresponding to each layer. Through hierarchical classification, the accuracy of the classification results is improved by avoiding too many classification labels at one time and too long context of prompt words composed of classification descriptions at each layer, which leads to inaccurate classification results. Based on the powerful built-in knowledge, generalization ability and understanding ability of the large model, it can adapt to the data classification and grading needs of different industries, and there is no need to repeat training as the classification and grading framework of different industries changes. It effectively reduces the reliance on manual rules and annotations, greatly saves costs, and significantly improves the efficiency of data classification and grading.

[0048] Correspondingly, the present application also provides data classification and grading devices, equipment and computer-readable storage media corresponding to the above-mentioned data classification and grading methods, which have the above-mentioned technical effects and are not repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] FIG1 is a flowchart of an implementation of a data classification and grading method in an embodiment of the present application;

[0051] FIG2 is another implementation flow chart of the data classification and grading method in an embodiment of the present application;

[0052] FIG3 is a structural block diagram of a data classification and grading device in an embodiment of the present application;

[0053] FIG4 is a structural block diagram of a data classification and grading device in an embodiment of the present application;

[0054] FIG5 is a schematic diagram of the specific structure of a data classification and grading device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0056] Referring to FIG. 1 , FIG. 1 is a flowchart of an implementation method of data classification and grading in an embodiment of the present application. The method may include the following steps:

[0057] S101: Receive a field to be classified, and obtain a target industry data classification and grading framework corresponding to the field to be classified.

[0058] The industry data classification and grading framework is formulated in advance according to the data classification and grading standards of different industries or the classification and grading guidelines and methods made by the unit in accordance with the prescribed format. When the fields to be classified need to be classified, the fields to be classified are sent to the data classification and grading center. The data classification and grading center receives the fields to be classified and obtains the target industry data classification and grading framework corresponding to the fields to be classified. For example, the target industry to which the fields to be classified belong can be determined first, and then the target industry data classification and grading framework corresponding to the target industry can be obtained.

[0059] The data classification and grading center may be a controller, a processor, etc., which is deployed with a large model for field classification.

[0060] S102: Obtain each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification description corresponding to each top-level classification label.

[0061] Each industry data classification and grading framework contains subclasses at each level, each subclass contains one or more classification labels corresponding to the subclass, and the industry data classification and grading framework also contains classification descriptions corresponding to each classification label. After obtaining the target industry data classification and grading framework corresponding to the field to be classified, obtain each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label.

[0062] S103: Using the large model, classify the fields to be classified according to the fields to be classified, the top-level classification labels and the top-level classification descriptions to obtain the current-level classification labels.

[0063] Pre-deploy a large model for field classification. After obtaining the top-level classification labels contained in the top-level subclasses and the top-level classification descriptions corresponding to the top-level classification labels, use the large model to classify the fields to be classified according to the fields to be classified, the top-level classification labels and the top-level classification descriptions to obtain the current level classification labels.

[0064] S104: Obtain the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtain the next-level classification descriptions corresponding to each next-level classification label.

[0065] In the target industry data classification and grading framework, each top-level classification label is generally provided with corresponding sub-classes of the next level. After obtaining the current level classification label, obtain the next level classification labels corresponding to each sub-class of the next level corresponding to the current level classification label, and obtain the next level classification description corresponding to each next level classification label.

[0066] S105: Using the large model, classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description to obtain the target next layer classification label.

[0067] After obtaining each next-level classification label and the next-level classification description corresponding to each next-level classification label, the large model is used to classify the to-be-classified field according to the field to be classified, the current-level classification label, each next-level classification label and each next-level classification description to obtain the target next-level classification label. By further classifying the to-be-classified field in combination with the current-level classification label, the accuracy of the classification result is improved.

[0068] The full name of the large model is Large Language Model (Large Language Mpdel). Large models refer to deep learning models with huge parameter volumes and computing power. These models are usually composed of billions to hundreds of billions of parameters and can learn complex patterns and representations through large-scale data training. Compared with traditional small models, large models can better capture subtle differences and complex associations in the data, thereby providing more accurate and comprehensive analysis and predictions. The advantage of large models is that they can process massive amounts of data and produce excellent results in multiple fields and tasks. These models have achieved significant breakthroughs and applications in natural language processing, image recognition, speech recognition, recommendation systems and other fields.

[0069] S106: Concatenate the current layer classification label and the target next layer classification label to obtain a new current layer classification label.

[0070] After classifying and obtaining the target next-level classification label, the current-level classification label and the target next-level classification label are concatenated to obtain a new current-level classification label. By concatenating the classification labels of each layer, the classification labels of each layer can be mutually verified, further improving the credibility of the classification results.

[0071] S107: Determine whether there is a next-level subclass, if so, return to step S104, if not, go to step S108.

[0072] After concatenating the classification labels to obtain new current-layer classification labels, determine whether there is a next-level subcategory. If so, it means that further category subdivision is needed, and continue to obtain the next-level classification labels and corresponding classification descriptions corresponding to each next-level subcategory corresponding to the previous-layer classification label, and perform further classification label acquisition. If not, it means that the classification field has been clearly classified.

[0073] S108: Determine the current layer classification label as the data classification result corresponding to the field to be classified.

[0074] When it is determined that there is no subclass at the next level, it means that the field to be classified has been clearly classified, and the classification label of the current level is determined as the data classification result corresponding to the field to be classified, that is, the final classification label obtained by splicing the classification labels of each level is determined as the data classification result corresponding to the field to be classified. By using the big model to infer the specific meaning of the field to be classified and perform hierarchical multi-step classification, the big model prompts engineering technology in each level of classification step, gradually obtains the classification results of each level, and the last level result is the final classification result. The big model has built-in industry knowledge and the ability to understand industry classification and grading standards and specifications, which effectively reduces the dependence on personnel expertise and experience in the data classification and grading process, reduces manual input, can adapt to different industry data classification and grading needs, and significantly improves the efficiency and accuracy of data classification and grading.

[0075] Prompt Engineering is a method in the field of natural language processing (NLP) that cleverly designs and uses prompts to guide large language models (such as GPT-3, GPT-4, etc.) to generate specific content or complete specific tasks. These prompts can include questions, instructions, contextual information, etc., designed to help the model understand the task intent more accurately and generate output that meets the requirements. By cleverly constructing prompts, language models can be guided to complete various tasks such as question answering, text generation, data classification and grading annotation, and translation, thereby giving full play to the model's capabilities and improving its practicality. One of the core challenges of prompt engineering technology is to design the most effective prompts, fully understand the limitations of the model, and continuously adjust and optimize the prompts to obtain better performance and results.

[0076] S109: Perform grading mapping based on the target industry data classification and grading framework and the data classification results to obtain data grading results.

[0077] The industry data classification and grading framework also includes data levels corresponding to data categories. After determining the data classification results corresponding to the fields to be classified, hierarchical mapping is performed based on the target industry data classification and grading framework and the data classification results to obtain the data grading results.

[0078] It can be seen from the above technical solution that by adopting the large model prompt engineering technology, the specific meaning of the field is first inferred, and then hierarchical multi-step classification is performed, and the final data classification result is obtained by splicing the classification labels corresponding to each layer. Through hierarchical classification, the accuracy of the classification results is improved by avoiding too many classification labels at one time and too long context of prompt words composed of classification descriptions at each layer, which leads to inaccurate classification results. Based on the powerful built-in knowledge, generalization ability and understanding ability of the large model, it can adapt to the data classification and grading needs of different industries, and there is no need to repeat training as the classification and grading framework of different industries changes. It effectively reduces the reliance on manual rules and annotations, greatly saves costs, and significantly improves the efficiency of data classification and grading.

[0079] It should be noted that, based on the above embodiment, the embodiment of the present application also provides corresponding improved solutions. In the subsequent embodiments, the same steps or corresponding steps as those in the above embodiment can be referenced to each other, and the corresponding beneficial effects can also be referenced to each other, which will not be repeated one by one in the following improved embodiments.

[0080] Referring to FIG. 2 , FIG. 2 is another implementation flow chart of the data classification and grading method in an embodiment of the present application. The method may include the following steps:

[0081] S201: Receive the fields to be classified including data source, database name, table name, table comment, field name, field comment, and field sample content, and obtain the target industry data classification and grading framework corresponding to the fields to be classified.

[0082] Among them, the table name and field name are not empty.

[0083] The fields to be classified may contain data sources, database names, table names, table comments, field names, field comments, and field sample contents. The data classification and grading center receives the fields to be classified that contain data sources, database names, table names, table comments, field names, field comments, and field sample contents, and obtains the target industry data classification and grading framework corresponding to the fields to be classified. The table name and field name in the fields to be classified are non-empty items, and other items can be empty. By combining the field information such as the table name and field name in the fields to be classified, the specific meaning of the field is explained through the zero-sample thinking chain prompt technology, providing a reference basis for field classification and grading.

[0084] See Table 1, which is a table of fields to be classified in the education industry.

[0085] Table 1

[0086] Data Source Database Name Table name Table Notes Field Name Field comments Field content (sample) Education System test staff Employee Information Sheet position Administrative duties / Education System test staff Employee Information Sheet years Years of administrative management / Education System test staff Employee Information Sheet age age /

[0087] S202: Obtain each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification description corresponding to each top-level classification label.

[0088] See Table 2, which is a data classification and grading framework table for the education industry.

[0089] Table 2

[0090] industry First level subclass Secondary sub-category Third level subclass Four sub-classes Five sub-categories Category Description Field Examples Minimum reference level educate Profile data School basic data / / / / Identification code, name, address, unified social information code, postal code, school type code, school (kindergarten) license number, basic campus information, department and institute data, contact number, historical evolution data, etc. 1 educate Profile data Subject point data / / / / Discipline category data, master's degree data, doctoral degree data, subject point statistics, etc. 1 educate Student data Student basic data / / / / Sensitive personal information such as name, gender, ethnicity, date of birth, place of origin, political status, registered residence, personal identification information (ID number, passport number, etc.), personal health and physiological information (symptoms, past medical history, etc.), personal biometric information (fingerprints, facial recognition features, etc.), and personal information of minors under the age of 14 3 educate Student data Student management data / / / / New student data, enrollment data, student status data, academic degree data, practical activity data, funding data, graduate employment data, etc. 3 educate Faculty and Staff Data Basic data of faculty and staff / / / / Personal information such as name, gender, ethnicity, date of birth, place of origin, political status, registered residence, contact number, etc., personal identification information (ID number, passport number, etc.), personal health and physiological information (symptoms, medical history, etc.), personal biometric information (fingerprints, facial recognition features, etc.), personal property information (bank account, etc.), etc. 3 educate Faculty and Staff Data Faculty and staff management data / / / / Recruitment data, job position data, assessment data, expert management data, further education and training data, salary and benefits data 3

[0091] S203: Obtain the field in the same table corresponding to the field to be classified.

[0092] After receiving the field to be classified, the field in the same table corresponding to the field to be classified is obtained. The fields in the same table generally have similar category attributes.

[0093] S204: Determine the fields in the same table as background fields of the fields to be classified.

[0094] After obtaining the fields in the same table corresponding to the fields to be classified, the fields in the same table are determined as the background fields of the fields to be classified. By determining the fields in the same table as the background fields of the fields to be classified, the large model can be better prompted to infer the meaning of the fields. This method can more accurately infer the meaning of the fields to be classified that lack field annotations and have poor data governance.

[0095] S205: Using the large model, classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label, each next layer classification description and the background field to obtain the current layer classification label.

[0096] After the fields in the same table are determined as the background fields of the fields to be classified, the large model is used to classify the fields to be classified according to the fields to be classified, the current layer classification labels, the next layer classification labels, the next layer classification descriptions and the background fields to obtain the current layer classification labels.

[0097] S206: Obtain the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtain the next-level classification descriptions corresponding to each next-level classification label.

[0098] S207: Using the large model, classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description to obtain the target next layer classification label.

[0099] S208: Concatenate the current layer classification label and the target next layer classification label to obtain a new current layer classification label.

[0100] S209: Determine whether there is a next-level subclass, if so, return to step S206, if not, execute step S210.

[0101] S210: Obtain the classification description corresponding to the current layer classification label, and concatenate the classification description with the current layer classification label to obtain a new current layer classification label.

[0102] When it is determined that there is no subclass at the next level, it means that the field to be classified has been clearly classified, and the classification description corresponding to the current level classification label is obtained, and the classification description is concatenated with the current level classification label to obtain a new current level classification label.

[0103] S211: Determine the current layer classification label as the data classification result corresponding to the field to be classified.

[0104] The readability of the data classification and grading results is enhanced by concatenating the classification descriptions corresponding to the classification labels of the last layer with the classification labels of the current layer as the final data classification and grading results.

[0105] S212: Perform grading mapping based on the target industry data classification and grading framework and the data classification results to obtain data grading results.

[0106] S213: Determine the target permission protection measures corresponding to the fields to be classified according to the data classification results.

[0107] Each data classification is correspondingly provided with corresponding permission protection measures. After obtaining the data classification result, the target permission protection measures corresponding to the field to be classified are determined according to the data classification result.

[0108] S214: Perform data protection on the classification field using target permission protection measures.

[0109] After determining the target permission protection measure corresponding to the field to be classified, the target permission protection measure is used to perform data protection on the field to be classified.

[0110] Corresponding to the above method embodiment, the present application also provides a data classification and grading device. The data classification and grading device described below and the data classification and grading method described above can refer to each other.

[0111] Referring to FIG. 3 , FIG. 3 is a structural block diagram of a data classification and grading device in an embodiment of the present application, and the device may include:

[0112] The classification and grading framework acquisition module 31 is used to receive the field to be classified and acquire the target industry data classification and grading framework corresponding to the field to be classified;

[0113] The first label and description acquisition module 32 is used to acquire each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label;

[0114] The current layer classification label obtaining module 33 is used to classify the to-be-classified field according to the to-be-classified field, each top-level classification label and each top-level classification description using the large model to obtain the current layer classification label;

[0115] The second label and description acquisition module 34 is used to acquire the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and acquire the next-level classification description corresponding to each next-level classification label;

[0116] The next level classification label obtaining module 35 is used to classify the to-be-classified field according to the to-be-classified field, the current level classification label, each next level classification label and each next level classification description using the large model to obtain the target next level classification label;

[0117] The classification result obtaining module 36 is used to concatenate the current layer classification label and the target next layer classification label to obtain a new current layer classification label, and repeatedly execute the step of obtaining the next layer classification labels corresponding to each next layer subclass corresponding to the current layer classification label until there is no next layer subclass, and then determine the current layer classification label as the data classification result corresponding to the field to be classified;

[0118] The classification result obtaining module 37 is used to perform classification mapping according to the target industry data classification and grading framework and the data classification results to obtain the data classification results.

[0119] It can be seen from the above technical solution that by adopting the large model prompt engineering technology, the specific meaning of the field is first inferred, and then hierarchical multi-step classification is performed, and the final data classification result is obtained by splicing the classification labels corresponding to each layer. Through hierarchical classification, the accuracy of the classification results is improved by avoiding too many classification labels at one time and too long context of prompt words composed of classification descriptions at each layer, which leads to inaccurate classification results. Based on the powerful built-in knowledge, generalization ability and understanding ability of the large model, it can adapt to the data classification and grading needs of different industries, and there is no need to repeat training as the classification and grading framework of different industries changes. It effectively reduces the reliance on manual rules and annotations, greatly saves costs, and significantly improves the efficiency of data classification and grading.

[0120] In a specific implementation of the present application, the classification and grading framework acquisition module 31 is specifically a module for receiving fields to be classified including data source, database name, table name, table comment, field name, field comment, and field sample content; wherein the table name and field name are not empty.

[0121] In a specific implementation of the present application, the next level classification label obtaining module 35 includes:

[0122] The same table field acquisition submodule is used to obtain the same table field corresponding to the field to be classified;

[0123] A background field determination submodule, used to determine the same table field as the background field of the field to be classified;

[0124] The field classification submodule is used to use the large model to classify the fields to be classified according to the fields to be classified, the current layer classification labels, the classification labels of each next level, the classification descriptions of each next level and the background fields.

[0125] In a specific embodiment of the present application, the device may further include:

[0126] The permission protection measure determination module is used to determine the target permission protection measures corresponding to the to-be-classified fields according to the data classification results;

[0127] The data protection module is used to protect the data of the classification field by using the target permission protection measures.

[0128] In a specific implementation of the present application, the classification result obtaining module 36 includes:

[0129] The current layer classification label acquisition submodule is used to obtain the classification description corresponding to the current layer classification label, and concatenate the classification description with the current layer classification label to obtain a new current layer classification label;

[0130] The classification result obtaining submodule is used to determine the current layer classification label as the data classification result corresponding to the field to be classified.

[0131] Corresponding to the above method embodiment, referring to FIG. 4 , FIG. 4 is a schematic diagram of a data classification and grading device provided by the present application, and the device may include:

[0132] A memory 332, for storing computer programs;

[0133] The processor 322 is used to implement the steps of the data classification and grading method of the above method embodiment when executing a computer program.

[0134] Specifically, please refer to Figure 5, which is a schematic diagram of the specific structure of a data classification and grading device provided in this embodiment. The data classification and grading device may have relatively large differences due to different configurations or performances, and may include a processor (central processing units, CPU) 322 (for example, one or more processors) and a memory 332, and the memory 332 stores one or more computer programs 342 or data 344. Among them, the memory 332 can be a temporary storage or a permanent storage. The program stored in the memory 332 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the data processing device. Furthermore, the processor 322 can be configured to communicate with the memory 332 to execute a series of instruction operations in the memory 332 on the data classification and grading device 301.

[0135] The data classification and grading device 301 may further include one or more power supplies 326 , one or more wired or wireless network interfaces 350 , one or more input and output interfaces 358 , and / or one or more operating systems 341 .

[0136] The steps in the data classification and grading method described above can be implemented by the structure of the data classification and grading device.

[0137] Corresponding to the above method embodiment, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps can be implemented:

[0138] Receive the field to be classified, and obtain the target industry data classification and grading framework corresponding to the field to be classified; obtain the top-level classification labels contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to the top-level classification labels; use the big model to classify the field to be classified according to the field to be classified, the top-level classification labels and the top-level classification descriptions to obtain the current-level classification label; obtain the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtain the next-level classification descriptions corresponding to each next-level classification label; use the big model to classify the field to be classified according to the field to be classified, the top-level classification labels and the top-level classification descriptions to obtain the current-level classification label; The classification field to be classified is classified by classification fields, current-layer classification labels, each next-layer classification labels and each next-layer classification description to obtain the target next-layer classification label; the current-layer classification label and the target next-layer classification label are concatenated to obtain a new current-layer classification label, and the steps of obtaining the next-layer classification labels corresponding to each next-layer subclass corresponding to the current-layer classification label are repeated until there is no next-layer subclass, then the current-layer classification label is determined as the data classification result corresponding to the field to be classified; hierarchical mapping is performed according to the target industry data classification and grading framework and the data classification result to obtain the data grading result.

[0139] The computer-readable storage medium may include: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0140] For an introduction to the computer-readable storage medium provided in this application, please refer to the above method embodiment, and this application will not go into details here.

[0141] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the devices, equipment and computer-readable storage media disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0142] Specific examples are used herein to illustrate the principles and implementation methods of the present application, and the description of the above embodiments is only used to help understand the technical solution and core ideas of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the present application.

Claims

1. A data classification and grading method, characterized in that: include: Receive a field to be classified, and obtain a target industry data classification and grading framework corresponding to the field to be classified; Obtain each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label; Using the large model, the field to be classified is classified according to the field to be classified, each top-level classification label and each top-level classification description to obtain a current-level classification label; Obtaining next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and obtaining next-level classification descriptions corresponding to each next-level classification label; Using the large model, classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description to obtain a target next layer classification label; The current layer classification label and the target next layer classification label are concatenated to obtain a new current layer classification label, and the step of obtaining the next layer classification labels corresponding to each next layer subclass corresponding to the current layer classification label is repeated until there is no next layer subclass, and the current layer classification label is determined as the data classification result corresponding to the field to be classified; A hierarchical mapping is performed based on the target industry data classification and grading framework and the data classification result to obtain a data grading result.

2. The data classification and grading method according to claim 1, characterized in that: Receive the fields to be classified, including: Receive the fields to be classified including data source, database name, table name, table comment, field name, field comment, and field sample content; wherein the table name and the field name are not empty.

3. The data classification and grading method according to claim 1 or 2, characterized in that: Using the large model to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label and each next layer classification description, includes: Obtain the field in the same table corresponding to the field to be classified; Determine the same table field as the background field of the field to be classified; The large model is used to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label, each next layer classification description and the background field.

4. The data classification and grading method according to claim 1, characterized in that: After obtaining the data classification results, it also includes: Determine the target permission protection measures corresponding to the to-be-classified fields according to the data classification results; The target permission protection measure is used to perform data protection on the field to be classified.

5. The data classification and grading method according to claim 1, characterized in that: Determining the current layer classification label as the data classification result corresponding to the field to be classified includes: Obtaining a classification description corresponding to a current-layer classification label, and concatenating the classification description with the current-layer classification label to obtain a new current-layer classification label; The current layer classification label is determined as the data classification result corresponding to the field to be classified.

6. A data classification and grading device, characterized in that: include: A classification and grading framework acquisition module is used to receive the field to be classified and acquire the target industry data classification and grading framework corresponding to the field to be classified; A first label and description acquisition module is used to acquire each top-level classification label contained in the top-level subclass in the target industry data classification and grading framework and the top-level classification descriptions corresponding to each top-level classification label; A current layer classification label obtaining module is used to classify the to-be-classified field according to the to-be-classified field, each top-level classification label and each top-level classification description using a large model to obtain a current layer classification label; A second label and description acquisition module is used to acquire the next-level classification labels corresponding to each next-level subclass corresponding to the current-level classification label, and acquire the next-level classification descriptions corresponding to each next-level classification label; A next-level classification label obtaining module is used to classify the field to be classified according to the field to be classified, the current-level classification label, each next-level classification label and each next-level classification description using the large model to obtain a target next-level classification label; A classification result acquisition module is used to concatenate the current layer classification label and the target next layer classification label to obtain a new current layer classification label, and repeatedly execute the step of obtaining the next layer classification labels corresponding to each next layer subclass corresponding to the current layer classification label until there is no next layer subclass, and then determine the current layer classification label as the data classification result corresponding to the field to be classified; The grading result acquisition module is used to perform grading mapping according to the target industry data classification and grading framework and the data classification results to obtain the data grading results.

7. The data classification and grading device according to claim 6, characterized in that: The next level classification label acquisition module includes: The same table field acquisition submodule is used to acquire the same table field corresponding to the field to be classified; A background field determination submodule, used for determining the same table field as the background field of the field to be classified; The field classification submodule is used to classify the field to be classified according to the field to be classified, the current layer classification label, each next layer classification label, each next layer classification description and the background field by using the large model.

8. The data classification and grading device according to claim 6, characterized in that: The classification result acquisition module includes: A classification label concatenation submodule is used to obtain a classification description corresponding to a current-layer classification label, and concatenate the classification description with the current-layer classification label to obtain a new current-layer classification label; The classification result obtaining submodule is used to determine the current layer classification label as the data classification result corresponding to the field to be classified.

9. A data classification and grading device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the data classification and grading method as claimed in any one of claims 1 to 5 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the data classification and grading method according to any one of claims 1 to 5.