A data hierarchical classification model construction method, device and equipment and storage medium

Abstract

Embodiments of the present application disclose a data grading classification model construction method and device, equipment and a storage medium, which can be applied to the financial field or other fields. The method comprises: analyzing sensitive data of a business system to determine first associated information; training a first data grading classification model using a data grading classification policy document to obtain a second data grading classification model; and training the second data grading classification model using the first associated information to obtain a third data grading classification model. In this way, the third data grading classification model can be used to automatically determine the security level and category of the sensitive word features to be graded and classified, thereby improving the efficiency of data grading and classification and reducing errors in data grading and classification.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and more particularly, to a data hierarchical classification model construction method and device, equipment and a storage medium. BACKGROUND

[0002] In the big data era, the value and role of data are increasingly prominent, and data flow is in various business transactions and core processes, supporting the survival and development of enterprises. While data is growing explosively, data security problems also occur frequently.

[0003] At present, different levels of sensitive data are divided according to the sensitivity and the degree of influence, and are managed and protected in different levels. However, the existing data hierarchical classification relies on manual operation, which not only has the problem of low efficiency, but also is prone to hierarchical classification errors. SUMMARY

[0004] Therefore, the embodiments of the present application disclose a data hierarchical classification model construction method, device, equipment and storage medium to improve the efficiency of data hierarchical classification and reduce the errors of data hierarchical classification.

[0005] The technical scheme provided by the embodiments of the present application is as follows:

[0006] In a first aspect, the embodiments of the present application provide a data hierarchical classification model construction method, which comprises:

[0007] Analyzing sensitive data of a business system to determine first associated information; the first associated information includes sensitive words, sensitive word features and a first business scenario;

[0008] Training a first data hierarchical classification model using a data hierarchical classification policy document to obtain a second data hierarchical classification model; the data hierarchical classification policy document includes data content description information, a security level and a category, and the data content description information includes at least one of the following: sensitive words, a first business scenario;

[0009] Training the second data hierarchical classification model using the first associated information to obtain a third data hierarchical classification model; the third data hierarchical classification model is used to determine the security level and the category of a to-be-classified sensitive word feature.

[0010] In a possible implementation, the training of the first data hierarchical classification model using the data hierarchical classification policy document to obtain the second data hierarchical classification model comprises:

[0011] Taking the security level and the category as labels of the data content description information to construct first training data;

[0012] The first data hierarchical classification model is trained by using the first training data, to obtain a second hierarchical classification model.

[0013] In a possible implementation, the training of the second data hierarchical classification model by using the first association information comprises:

[0014] A common general word in the first business scenario is obtained.

[0015] Second training data is constructed by using the sensitive word, the sensitive word feature, the first business scenario and the common general word, and / or third training data is constructed by using the sensitive word, the sensitive word feature and the first business scenario.

[0016] The second data hierarchical classification model is trained by using the second training data, and / or the second data hierarchical classification model is trained by using the third training data, to obtain a third data hierarchical classification model.

[0017] In a possible implementation, the method further comprises:

[0018] A sensitive data identification model is trained by using first association information; the sensitive data model is used to locate a sensitive word to be classified and classified in the data to be classified and classified.

[0019] In a possible implementation, the method further comprises:

[0020] The data to be classified and classified is input into the sensitive data identification model, to locate the sensitive word to be classified and classified in the data to be classified and classified, and to obtain second association information corresponding to the data to be classified and classified; the second association information comprises a sensitive word to be classified and classified, a sensitive word feature to be classified and classified and a second business scenario.

[0021] The second association information is input into the third data hierarchical classification model, to obtain a security level and a category of the sensitive word feature to be classified and classified.

[0022] In a second aspect, an embodiment of the present application provides a data hierarchical classification model construction device, the device comprises:

[0023] An analysis module is configured to analyze sensitive data of a business system and determine first association information; the first association information comprises a sensitive word, a sensitive word feature and a first business scenario.

[0024] The training module is configured to train a first data classification model by using a data classification policy document to obtain a second data classification model; the data classification policy document comprises data content description information, a security level, and a category, and the data content description information comprises at least one of a sensitive word and a first business scenario.

[0025] The training module is further configured to train the second data classification model by using the first association information to obtain a third data classification model; and the third data classification model is configured to determine a security level and a category of a sensitive word feature to be classified.

[0026] In a possible implementation, the training module comprises:

[0027] The training data construction unit is configured to construct first training data by taking the security level and the category as labels of the data content description information.

[0028] The training subunit is configured to train the first data classification model by using the first training data to obtain a second classification model.

[0029] In a possible implementation, the training module comprises:

[0030] The acquisition unit is configured to acquire common general words in the first business scenario.

[0031] The training data construction unit is configured to construct second training data by using the sensitive word, the sensitive word feature, the first business scenario, and the common general words, and / or construct third training data by using the sensitive word, the sensitive word feature, and the first business scenario.

[0032] The training subunit is configured to train the second data classification model by using the second training data, and / or train the second data classification model by using the third training data to obtain a third data classification model.

[0033] In a third aspect, an embodiment of the present application provides an electronic device, comprising:

[0034] A memory is configured to store instructions.

[0035] A processor is configured to execute the instructions in the memory to perform the data classification model construction method in any one of the first aspect.

[0036] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium comprising instructions, which, when executed on a computer, cause the computer to perform the data classification model construction method in any one of the first aspect.

[0037] In a fifth aspect, an embodiment of the present application provides a computer program product, which, when running on a terminal device, causes the terminal device to perform the data hierarchical classification model construction method of any one of the above first aspect.

[0038] Based on the above technical solution, the present application has the following beneficial effects:

[0039] The embodiment of the present application discloses a data hierarchical classification model construction method, device, equipment and storage medium method. Among them, the method comprises: analyzing the sensitive data of the business system, and determining the first associated information; training the first data hierarchical classification model by using the data hierarchical classification policy document to obtain the second data hierarchical classification model; training the second data hierarchical classification model by using the first associated information to obtain the third data hierarchical classification model. As can be seen, in the embodiment of the present application, the data hierarchical classification policy document includes data content description information, security level and category, and the data content description information includes any one of sensitive words and the first business scene, so that the second data hierarchical classification model can learn the association relationship between the sensitive words and / or the first business scene and the security level and category. Since the first associated information includes sensitive words, sensitive word features and the first business scene, and the second data hierarchical classification model has learned the association relationship between the sensitive words and / or the first business scene and the security level and category, the third hierarchical model can associate the sensitive words, sensitive word features and the first business scene with the security level and category, so that the third hierarchical model can learn the association relationship between the sensitive words, sensitive word features and the first business scene and the security level and category, and further make the third data hierarchical classification model be used for determining the security level and category of the to-be-classified sensitive word feature. In this way, the security level and category of the to-be-classified sensitive word feature can be automatically determined by using the third data hierarchical classification model, so as to improve the efficiency of data hierarchical classification and reduce the error of data hierarchical classification. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the disclosed drawings without creative labor.

[0041] Figure 1 A flowchart of a data hierarchical classification model construction method disclosed by an embodiment of the present application;

[0042] Figure 2 A flowchart of a data hierarchical classification method disclosed by an embodiment of the present application;

[0043] Figure 3 FIG. 1 is a structural schematic diagram of a data hierarchical classification model construction device disclosed by an embodiment of the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0045] It should be noted that the data hierarchical classification model construction method, device, equipment and storage medium provided by the embodiments of the present application can be used in the financial field or other fields, for example, can be used in the application scenario of classifying sensitive data in financial data in the financial field. Other fields are any field except the financial field, for example, the field of data processing. The above are only examples, and do not limit the application field of the data hierarchical classification model construction method, device, equipment and storage medium provided by the embodiments of the present application.

[0046] Referring to FIG. 1, a flowchart of a data hierarchical classification model construction method disclosed by an embodiment of the present application is shown, and the method comprises the following steps. Figure 1

[0047] S101, analyzing sensitive data of a business system to determine first association information.

[0048] The first association information comprises a sensitive word, a sensitive word feature and a first business scenario.

[0049] It should be noted that the business system in the embodiments of the present application can be a core business system, a clearing and distribution system, a credit card system, etc., and the present application does not limit this, as long as it is a business system containing sensitive data.

[0050] Sensitive data refers to data that may cause serious harm to society or individuals after leakage. A sensitive word refers to a description word corresponding to a sensitive information, for example: name, gender, ID number, home address, etc. A sensitive word feature refers to specific information corresponding to a sensitive word, for example: the sensitive word feature corresponding to an ID number is 351XXXXXXXXXXXXXXX or 352XXXXXXXXXXXXXXX; the sensitive word feature corresponding to gender is female or male, etc. The first business scenario refers to, for example, insurance business, current account, collection and payment, salary payment, personal check, etc., and the present application does not limit this, which can be set according to actual needs.

[0051] ​For example, a piece of sensitive data is the data generated when a customer handles insurance business, and the first associated information is determined by analyzing the sensitive data, such as name-zhangsan-insurance business, identity account-351XXXXXXXXXXXXXXX-insurance business, and the like. It can be understood that the above is only an exemplary description and should not be construed as a limitation of the present application.

[0052] In the embodiments of the present application, the sensitive data of the business system can be analyzed by using big data technology to determine the first associated information, and then a sensitive word library is established based on the first associated information, and the first associated information is stored in the sensitive word library. After that, if new sensitive data is generated in the business system, the first associated information corresponding to the sensitive data can also be determined and stored in the sensitive word library. In a possible implementation manner, the third data classification model can be further trained by using new first associated information every preset time to continuously improve the performance of the data classification model. For example, the third data classification model is trained by using new first associated information every 1 month.

[0053] S102, training the first data classification model by using the data classification policy document to obtain a second data classification model;

[0054] The data classification policy document includes data content description information, security level, and category, and the data content description information includes at least one of the following: sensitive words and first business scenarios.

[0055] For example, the data classification policy document can be the Financial Data Security Data Security Classification Guide, which includes a data classification rule reference table. Part of the content of one table is shown in Table 1 below. It can be understood that the above is only an exemplary description and should not be construed as a limitation of the present application. In the embodiments of the present application, data classification models can also be trained by using data classification policy documents other than the Financial Data Security Data Security Classification Guide.

[0056] Table 1 Financial Industry Institution Typical Data Classification Rule Reference Table

[0057]

[0058]

[0059] It can be understood that the data content description information can include the definitions and contents in Table 1 described above, and the category can include at least one of the first subcategory, the second subcategory, the third subcategory, and the fourth subcategory described above. For example, the category can be represented as first subcategory-second subcategory-third subcategory-fourth subcategory, such as customer-personal-personal natural information-personal basic profile information or customer-personal-personal natural information-personal health physiological information; the category can also be represented as a first subcategory, such as customer; the category can also be represented as a fourth subcategory, such as personal basic profile information or personal health physiological information, and the like. The specific representation of the category is not limited in the present application, and can be set according to actual needs. The above is only an exemplary description, and should not be understood as a limitation on the present application.

[0060] S103, training the second data hierarchical classification model using the first association information to obtain a third data hierarchical classification model.

[0061] The third data hierarchical classification model is configured to determine the security level and the category of the sensitive word feature to be classified.

[0062] In the embodiments of the present application, the prediction accuracy of the third data hierarchical classification model can be verified using test samples, and the third data hierarchical classification model can be adjusted accordingly.

[0063] As can be seen, in the data hierarchical classification model construction method provided in the embodiments of the present application, since the data hierarchical classification policy document includes the data content description information, the security level, and the category, and the data content description information includes any one of the sensitive word and the first business scenario, the second data hierarchical classification model can learn the association relationship between the sensitive word and / or the first business scenario and the security level and the category. Since the first association information includes the sensitive word, the sensitive word feature, and the first business scenario, and the second data hierarchical classification model has learned the association relationship between the sensitive word and / or the first business scenario and the security level and the category, the third hierarchical model can associate the sensitive word, the sensitive word feature, and the first business scenario with the security level and the category. Thus, the third hierarchical model can learn the association relationship between the association relationship between the sensitive word, the sensitive word feature, and the first business scenario and the security level and the category. Further, the third data hierarchical classification model can be used to determine the security level and the category of the sensitive word feature to be classified. In this way, the third data hierarchical classification model can be used to automatically determine the security level and the category of the sensitive word feature to be classified, thereby improving the efficiency of data hierarchical classification and reducing errors in data hierarchical classification.

[0064] In a possible implementation, the step S102 in the data hierarchical classification model construction method provided in the embodiments of the present application includes:

[0065] S1021, constructing first training data with security levels and categories as labels of data content description information;

[0066] In the embodiments of the present application, different data content description information can be extracted from the data classification policy document, and the different data content description information corresponds to security levels and categories, and then a plurality of first training data are constructed.

[0067] For example, the data content description is the personal basic situation data, such as personal name, gender, nationality, ethnicity, marital status, certificate type, certificate number, certificate effective date, expiration date, home address, etc., and the corresponding label is customer-personal-personal natural information-personal basic profile information and 3. It can be understood that the above is only an example and should not be construed as a limitation of the present application.

[0068] S1022, training the first data classification model using the first training data to obtain a second classification model.

[0069] The first data classification model in the embodiments of the present application can be a support vector machine (SVM) or the like, and the present application does not limit this, which can be selected and set according to actual needs. The basic model of SVM is a linear classifier with the largest interval defined in the feature space, which is different from the perceptron; SVM also includes the kernel trick, which makes it a substantial nonlinear classifier.

[0070] It can be seen that in the embodiments of the present application, the first training data with labels can be quickly constructed based on the data classification policy document, and the first data classification model is supervised learning based on the first training data with labels, which avoids the problem of low efficiency caused by the need for manual construction of training data with labels, and improves the construction efficiency of the data classification model.

[0071] In a possible implementation, the step S103 of the data classification model construction method provided in the embodiments of the present application includes:

[0072] S1031, obtaining common general words in the first business scenario;

[0073] For example, the first business scenario is an insurance business scenario, and the common general words include insurance policy, property insurance, liability insurance, etc. It can be understood that the above is only an example and should not be construed as a limitation of the present application.

[0074] S1032, constructing second training data using sensitive words, sensitive word features, first business scenarios, common general words, and / or constructing third training data using sensitive words, sensitive word features, and first business scenarios;

[0075] It can be understood that the same sensitive word in the embodiments of the present application, and the sensitive word feature and the first business scenario corresponding to the sensitive word can construct the third training data, or can construct the second training data in combination with the corresponding common general word, and the present application does not make any limitation.

[0076] For example, the sensitive word is an insurer, the sensitive word feature is Zhang San, and the first business scenario is an insurance business, which can construct the third training data as insurer-Zhang San-insurance business, or can construct the third training data as insurer-Zhang San-insurance business-insurance policy. It can be understood that the above is only an exemplary description and should not be understood as a limitation of the present application.

[0077] In a possible implementation, the first feature matrix can be formed according to the sensitive word, the sensitive word feature, the first business scenario, and the common general word in the embodiments of the present application to construct the second training data; and the second feature matrix can be formed according to the sensitive word, the sensitive word feature, and the first business to construct the third training data. It can be understood that the second training data and / or the third training data can also be constructed by other ways in the embodiments of the present application, and the present application does not make any limitation.

[0078] S1033, training the second data classification model by using the second training data, and / or training the second data classification model by using the third training data to obtain a third data classification model.

[0079] It can be seen that the second training data is further constructed in combination with the common general word in the business scenario in the embodiments of the present application, and the third data classification model is unsupervised learned by using the second training data and / or the third training data, which can further associate the first association information with the data classification policy document, can further learn the association relationship between the sensitive word, the sensitive word feature, the first business scenario and the security level and the category, thereby improving the accuracy of the third data classification model.

[0080] Referring to Figure 2 , the flow chart of the data classification method disclosed in the embodiments of the present application, the method comprises:

[0081] S201, inputting the data to be classified into a sensitive data recognition model, positioning the sensitive word to be classified in the data to be classified, and obtaining the second association information corresponding to the data to be classified.

[0082] The second association information includes: the sensitive word to be classified, the sensitive word feature to be classified, and the second business scenario. It can be understood that the classified sensitive word is similar to the above-mentioned sensitive word, the sensitive word feature to be classified is similar to the above-mentioned sensitive word feature, and the second business scenario is similar to the above-mentioned first business scenario, and therefore it is not repeated here.

[0083] In a possible implementation, the sensitive data identification model is trained by using the first association information, and the sensitive data identification model is used to locate the sensitive word to be classified in the data to be classified. It can be seen that the sensitive data identification model for locating the sensitive word to be classified is trained in the embodiment of the application, so that the data input into the third data classification model for classification is more easily identified by the model, and thus the accuracy of data classification is improved.

[0084] S202, input the second association information into the third data classification model to obtain the security level and the category of the sensitive word feature to be classified.

[0085] In the embodiment of the application, the second association information can be input into the third data classification model to perform feature extraction, vector calculation, feature scoring, identification and classification on the second association information, and obtain the final classification result.

[0086] It can be seen that in the embodiment of the application, the sensitive data identification model is used to obtain the second association information corresponding to the data to be classified, and then the second association information is input into the third classification model for identification, so that the data input into the third data classification model for classification is more easily identified by the model, and thus the accuracy of data classification is improved.

[0087] Referring to Figure 3 The embodiment of the application discloses a structural diagram of a data classification model construction device, which comprises:

[0088] The analysis module 301 is configured to analyze the sensitive data of the business system and determine the first association information. The sensitive association information comprises a sensitive word, a sensitive word feature and a first business scenario.

[0089] The training module 302 is configured to train the first data classification model by using a data classification policy document to obtain a second data classification model. The data classification policy document comprises data content description information, a security level and a category. The data content description information comprises at least one of a sensitive word and a first business scenario.

[0090] The training module 302 is further configured to train the second data classification model by using the first association information to obtain a third data classification model. The third data classification model is used to determine the security level and the category of the sensitive word feature to be classified.

[0091] It can be seen that, in the data hierarchical classification model construction device provided in the embodiments of the present application, the data hierarchical classification policy document includes the data content description information, the security level, and the category, and the data content description information includes any one of the sensitive word and the first business scenario, so that the second data hierarchical classification model can learn the association relationship between the sensitive word and / or the first business scenario and the security level and the category. Since the first association information includes the sensitive word, the sensitive word feature, and the first business scenario, and the second data hierarchical classification model has learned the association relationship between the sensitive word and / or the first business scenario and the security level and the category, the third hierarchical model can associate the sensitive word, the sensitive word feature, and the first business scenario with the security level and the category, so that the third hierarchical model can learn the association relationship between the association relationship between the sensitive word, the sensitive word feature, and the first business scenario and the security level and the category, and further, the third data hierarchical classification model can be used to determine the security level and the category of the to-be-hierarchically-classified sensitive word feature. In this way, the security level and the category of the to-be-hierarchically-classified sensitive word feature can be automatically determined by using the third data hierarchical classification model, so that the efficiency of data hierarchical classification is improved, and the error of data hierarchical classification is reduced.

[0092] In a possible implementation, the training module 302 in the data hierarchical classification model construction device provided in the embodiments of the present application includes:

[0093] The training data construction unit is configured to construct first training data by taking the security level and the category as labels of the data content description information.

[0094] The training sub-unit is configured to train the first data hierarchical classification model by using the first training data to obtain a second hierarchical classification model.

[0095] In a possible implementation, the training module 302 in the data hierarchical classification model construction device provided in the embodiments of the present application includes:

[0096] The acquisition unit is configured to acquire common general words in the first business scenario.

[0097] The training data construction unit is configured to construct second training data by using the sensitive word, the sensitive word feature, the first business scenario, and the common general word, and / or construct third training data by using the sensitive word, the sensitive word feature, and the first business scenario.

[0098] The training sub-unit is configured to train the second data hierarchical classification model by using the second training data, and / or train the second data hierarchical classification model by using the third training data to obtain a third data hierarchical classification model.

[0099] In a possible implementation, the training module 302 in the data hierarchical classification model construction apparatus provided in the embodiments of the present application is further configured to:

[0100] The sensitive data recognition model is trained by using the first association information, and is used for locating sensitive words to be classified in the data to be classified.

[0101] In a possible implementation, the data hierarchical classification model construction apparatus provided in the embodiments of the present application further includes:

[0102] The recognition module is configured to input the data to be classified into the sensitive data recognition model, locate sensitive words to be classified in the data to be classified, and obtain second association information corresponding to the data to be classified, wherein the second association information includes the sensitive words to be classified, sensitive word features of the sensitive words to be classified, and a second business scenario.

[0103] The hierarchical classification module is configured to input the second association information into the third data hierarchical classification model, and obtain a security level and a category of the sensitive word features.

[0104] Further, the embodiments of the present application further provide an electronic device, which includes:

[0105] A memory is configured to store instructions.

[0106] A processor is configured to execute the instructions in the memory to perform any of the implementation methods of the data hierarchical classification model construction method.

[0107] Further, the embodiments of the present application further provide a computer readable storage medium, which stores instructions, and when the instructions run on a terminal device, the terminal device performs any of the implementation methods of the data hierarchical classification model construction method.

[0108] Further, the embodiments of the present application further provide a computer program product, which runs on a terminal device, and makes the terminal device perform any of the implementation methods of the data hierarchical classification model construction method.

[0109] Those skilled in the art can clearly understand that all or part of the steps of the above-mentioned method can be implemented by means of software and necessary universal hardware platforms through the description of the above embodiments. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product in essence or in the part that contributes to the prior art. The computer software product can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network communication device such as a media gateway, etc.) execute the method described in each embodiment or some part of the embodiments of the present application.

[0110] It should be noted that the embodiments in the specification are 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.

[0111] Finally, it should be noted that in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0112] The embodiments in the specification are 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.

[0113] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a data hierarchical classification model, characterized in that, The method comprises: analyzing sensitive data of a business system to determine first association information; the first association information comprises sensitive words, sensitive word features, and a first business scenario; training a first data classification model using a data classification policy document to obtain a second data classification model; the data classification policy document comprises data content description information, a security level, and a category; the data content description information comprises at least one of the sensitive words and the first business scenario; the security level and the category are labels of the data content description information, so that the second data classification model learns the association between the sensitive words and / or the first business scenario and the security level and the category; training the second data classification model using the first association information to obtain a third data classification model; the third data classification model is used to determine the security level and the category of a to-be-classified sensitive word feature.

2. The method of claim 1, wherein, The method comprises: using the security level and the category as labels of the data content description information to construct first training data; training the first data classification model using the first training data to obtain a second classification model.

3. The method of claim 1, wherein, The method comprises: obtaining common general words in the first business scenario; constructing second training data using the sensitive words, the sensitive word features, the first business scenario, and the common general words, and / or constructing third training data using the sensitive words, the sensitive word features, and the first business scenario; training the second data classification model using the second training data and / or training the second data classification model using the third training data to obtain a third data classification model.

4. The method of claim 1, wherein, The method further comprises: training a sensitive data recognition model using the first association information; the sensitive data model is used to locate a to-be-classified sensitive word in to-be-classified data.

5. The method of claim 4, wherein, The method further comprises: inputting the to-be-classified data into the sensitive data recognition model to locate a to-be-classified sensitive word in the to-be-classified data, and obtaining second association information corresponding to the to-be-classified data; the second association information comprises a to-be-classified sensitive word, a to-be-classified sensitive word feature, and a second business scenario; inputting the second association information into the third data classification model to obtain the security level and the category of the to-be-classified sensitive word feature. 6.A data hierarchical classification model construction apparatus, characterized by comprising: The apparatus comprises: an analysis module configured to analyze sensitive data of a business system to determine first association information; the first association information comprises sensitive words, sensitive word features, and a first business scenario; The training module is configured to train a first data classification model by using a data classification policy document to obtain a second data classification model; the data classification policy document comprises data content description information, a security level, and a category; the data content description information comprises at least one of a sensitive word and a first business scenario; the security level and the category are labels of the data content description information; and the second data classification model is configured to learn an association between the sensitive word and / or the first business scenario and the security level and the category. The training module is further configured to train the second data classification model by using the first association information to obtain a third data classification model; and the third data classification model is configured to determine a security level and a category of a to-be-classified sensitive word feature.

7. The apparatus of claim 6, wherein, The training module comprises: a training data construction unit configured to construct first training data by taking the security level and the category as labels of the data content description information; a training subunit configured to train the first data classification model by using the first training data to obtain a second classification model.

8. The apparatus of claim 6, wherein, The training module comprises: an acquisition unit configured to acquire common general words in the first business scenario; a training data construction unit configured to construct second training data by using the sensitive word, the sensitive word feature, the first business scenario, and the common general words, and / or to construct third training data by using the sensitive word, the sensitive word feature, and the first business scenario; a training subunit configured to train the second data classification model by using the second training data, and / or to train the second data classification model by using the third training data to obtain a third data classification model.

9. An electronic device, comprising: The training module comprises: a memory configured to store instructions; a processor configured to execute the instructions in the memory to perform the data classification model construction method in any one of claims 1 to 5.

10. A computer-readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the data classification model construction method in any one of claims 1 to 5.

Images