A method, apparatus, device and medium for generating pseudo-column data

Abstract

This application discloses a method, apparatus, device, and medium for generating pseudo-column data. Based on a machine learning classification task, classification labels are determined; the classification label can be considered as the classification result corresponding to the classification task. To reduce the impact of pseudo-column data on machine learning, data types with weak or no correlation to the classification labels can be selected. Furthermore, to prevent the pseudo-column data from being directly detected, it is necessary to ensure a certain correlation between the type of pseudo-column data and the types of data in each column of the original data. Therefore, according to a set number of pseudo-column data columns, pseudo-column types that meet the requirement of being uncorrelated with the classification labels and correlated with each type of data in the original data can be selected. Pseudo-column data can be constructed according to the data generation rules corresponding to the pseudo-column types. The construction of pseudo-column data fully considers the classification labels related to machine learning, reducing the impact of pseudo-column data on machine learning classification.

Description

Technical Field

[0001] This application relates to the field of data security technology, and in particular to a method, apparatus, device and computer-readable storage medium for generating pseudo-column data. Background Technology

[0002] In the era of big data, data has become an increasingly valuable resource, containing immense social and economic value. With the ever-increasing demand for network data sharing and exchange, data breaches have occurred frequently in recent years. Source tracing is key to eradicating data breaches at their source. On the one hand, source tracing helps companies understand weaknesses in their internal security management and technical measures; on the other hand, it serves as a psychological deterrent to perpetrators of data breaches, thereby effectively reducing the occurrence of similar incidents.

[0003] Data watermarking uses algorithms to embed watermarks into raw data, ensuring the normal use of distributed data. Watermarked data is highly available, highly transparent, and highly concealed, making it difficult for external parties to detect or crack. Upon discovering an information leak, the watermark identifier can be extracted from the leaked data immediately. By reading the watermark identifier code, the leaking entity and responsible person can be accurately located, enabling precise accountability for data leaks and improving the security and traceability of data transmission.

[0004] For structured data stored in databases, common data watermarking algorithms include adding pseudo-rows, adding pseudo-columns, de-identifying watermarks, and adding invisible characters. Among these, adding pseudo-columns refers to the technique of artificially constructing one or more columns and embedding watermarks within them when adding watermarks to some output data.

[0005] Currently, there are two main ways to implement pseudo-column watermarking algorithms. The first method involves manually selecting a pseudo-column type from the supported data types and then constructing data of that pseudo-column type. The second method involves randomly selecting from the supported data types. Random selection can be based on the type of each column, using the contents of the corresponding built-in dictionary to construct the entire column, or it can be based on the composition rules of each column's data type, using internal generation logic to randomly construct the forged data for each column. Both of these implementations have strong randomness and do not consider subsequent data analysis scenarios, introducing noise that affects machine learning classification tasks, resulting in poor model classification performance or even unusable models.

[0006] It is evident that how to reduce the impact of pseudo-column data on machine learning classification is a problem that needs to be solved by those skilled in the art. Summary of the Invention

[0007] The purpose of this application is to provide a method, apparatus, device, and computer-readable storage medium for generating pseudo-column data, which can reduce the impact of pseudo-column data on machine learning classification.

[0008] To address the aforementioned technical problems, embodiments of this application provide a method for generating pseudo-column data, comprising:

[0009] The classification task is based on machine learning to determine the classification labels;

[0010] Based on the set number of pseudo-column data columns, select pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data;

[0011] Based on the data generation rules corresponding to the pseudo-column type, construct pseudo-column data.

[0012] Optionally, the step of selecting pseudo-column types that meet the non-relevance requirement with the classification label and the correlation requirement with each type of data in the original data, according to the set number of pseudo-column data columns, includes:

[0013] The association algorithm is used to determine the candidate data types that meet the correlation requirements with various types of data in the original data;

[0014] Based on the number of pseudo-column data columns, pseudo-column types that meet the non-relevance requirement with the classification label are selected from the candidate data types.

[0015] Optionally, constructing pseudo-column data according to the data generation rules corresponding to the pseudo-column type includes:

[0016] When the data corresponding to the pseudo-column type is numerical attribute data, numerical data is selected from the numerical range corresponding to the numerical attribute data under each category label in a uniformly distributed sampling manner within the same range to generate pseudo-column data.

[0017] Optionally, constructing pseudo-column data according to the data generation rules corresponding to the pseudo-column type includes:

[0018] When the data corresponding to the pseudo-column type belongs to category attribute data, category data is selected from the category set corresponding to the category attribute data using equal probability sampling under each category label to generate pseudo-column data.

[0019] Optionally, constructing pseudo-column data according to the data generation rules corresponding to the pseudo-column type includes:

[0020] When the data corresponding to the pseudo-column type belongs to business type data, the business type data is divided into numerical attribute data and category attribute data;

[0021] Generate numerical data according to the data generation rules corresponding to the numerical attribute data;

[0022] Generate category data according to the data generation rules corresponding to the category attribute data;

[0023] The numerical data and the category data are combined as pseudo-column data.

[0024] Optionally, after constructing the pseudo-column data according to the data generation rules corresponding to the pseudo-column type, the method further includes:

[0025] The pseudo-column data is embedded with a watermark and then inserted into the original data.

[0026] Optionally, the step of embedding the pseudo-column data with a watermark and then inserting it into the original data includes:

[0027] Embed a watermark in the pseudo-column data;

[0028] The pseudo-column data with embedded watermarks is added to the original data in a uniform distribution.

[0029] This application also provides a pseudo-column data generation apparatus, including a determining unit, a selecting unit, and a constructing unit;

[0030] The determining unit is used to determine the classification label based on a machine learning-based classification task;

[0031] The selection unit is used to select pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data, according to the set number of pseudo-column data columns.

[0032] The construction unit is used to construct pseudo-column data according to the data generation rules corresponding to the pseudo-column type.

[0033] Optionally, the selection unit includes a type determination subunit and a filtering subunit;

[0034] The type determination subunit is used to determine the candidate data types that meet the correlation requirements with various types of data in the original data using an association algorithm;

[0035] The filtering subunit is used to filter out pseudo-column types from the candidate data types that meet the requirement of non-relevance to the classification label based on the number of pseudo-column data columns.

[0036] Optionally, the construction unit is used to select numerical data from the numerical range corresponding to the numerical attribute data under each of the classification labels in a uniformly distributed sampling manner according to the same range when the data corresponding to the pseudo-column type belongs to numerical attribute data, so as to generate pseudo-column data.

[0037] Optionally, the construction unit is used to select category data from the category set corresponding to the category attribute data under each category label by using equal probability sampling when the data corresponding to the pseudo-column type belongs to category attribute data, so as to generate pseudo-column data.

[0038] Optionally, the construction unit is used to, when the data corresponding to the pseudo-column type belongs to business type data, divide the business type data into numerical attribute data and category attribute data; generate numerical data according to the data generation rules corresponding to the numerical attribute data; generate category data according to the data generation rules corresponding to the category attribute data; and combine the numerical data and the category data as pseudo-column data.

[0039] Optionally, it also includes an insertion unit;

[0040] The insertion unit is used to embed the pseudo-column data with a watermark and then insert it into the original data.

[0041] Optionally, the insertion unit includes an embedding subunit and an adding subunit;

[0042] The embedding subunit is used to embed a watermark into the pseudo-column data;

[0043] The added sub-unit is used to add the pseudo-column data with embedded watermark to the original data in a uniform distribution manner.

[0044] This application also provides an electronic device, including:

[0045] Memory, used to store computer programs;

[0046] A processor for executing the computer program to implement the steps of the method for generating pseudo-column data as described above.

[0047] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the pseudo-column data generation method described above.

[0048] As can be seen from the above technical solution, a classification label is determined based on the machine learning classification task; each row of data in the original data has a corresponding classification label. The classification label can be regarded as the classification result corresponding to the classification task. In order to reduce the impact of pseudo-column data on machine learning, when constructing pseudo-column data, data types with weak or no correlation with the classification label can be selected. In this way, even if pseudo-column data is added to the original data, its weak correlation with the classification label can greatly reduce the impact on machine learning classification. Furthermore, in order to avoid the direct detection of pseudo-column data, it is also necessary to ensure that the type of pseudo-column data has a certain correlation with the types of data in each column of the original data. Therefore, in this application, according to the set number of pseudo-column data columns, pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data can be selected. Different types of column data have their own corresponding data generation rules. Pseudo-column data can be constructed according to the data generation rules corresponding to the pseudo-column type. In this technical solution, the classification label related to machine learning is fully considered when constructing pseudo-column data, reducing the impact of pseudo-column data on machine learning classification. Furthermore, it comprehensively considers the types of data in each column of the original data, avoiding the direct detection of pseudo-column data due to excessive deviation from the data in the original data, thus improving the concealment and usability of pseudo-column data. Attached Figure Description

[0049] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a method for generating pseudo-column data provided in this application embodiment;

[0051] Figure 2 A schematic diagram of a pseudo-column data generation device provided in an embodiment of this application;

[0052] Figure 3 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0054] The terms “comprising” and “having” in the specification, claims, and accompanying drawings of this application, and any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may include steps or units not listed.

[0055] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] Next, a method for generating pseudo-column data provided in the embodiments of this application will be described in detail. Figure 1 A flowchart illustrating a method for generating pseudo-column data provided in this application embodiment, the method comprising:

[0057] S101: Classification task based on machine learning, determining classification labels.

[0058] Each row of data in the original data has a corresponding category label.

[0059] Category labels can be viewed as the classification results corresponding to a classification task. Category labels can contain at least two types.

[0060] For example, a classification task might involve categorizing data based on whether a loan has been taken out. In credit data, "loan status" could be specified as a classification label. Classification labels could include two types: "loan" and "no loan."

[0061] S102: Select pseudo-column types that meet the requirements of being unrelated to the category labels and are related to each category of data in the original data, according to the set number of pseudo-column data columns.

[0062] If there are too many pseudo-column data, it will greatly affect the distribution of the original data and make it easy to detect that the data is pseudo-column data. Therefore, in this embodiment, the number of pseudo-column data columns can be preset. When configuring the number of pseudo-column data columns, in order to reduce the introduction of noise and ensure the concealment of the watermark data, the number of pseudo-column data columns can be set to 1 or 2 columns.

[0063] In this embodiment, an association algorithm can be used to determine candidate data types that meet the correlation requirements with various types of data in the original data. Based on the number of pseudo-column data columns, pseudo-column types that meet the non-correlation requirements with the classification labels are selected from the candidate data types.

[0064] The non-relevance requirement can be achieved by ranking the candidate data types and category labels from highest to lowest relevance, thus selecting the pseudo-column data type with the lowest relevance. In practical applications, association rules can be generated using association algorithms such as April based on historical experience or industry data, and then candidate data types can be recommended based on the types of each column in the original data.

[0065] For example, if the columns of the original data are "Name", "Gender", "Age", "Address", "Annual Income", etc., then the association algorithm recommends "Contact Information" as a candidate data type. Since the correlation between "Contact Information" and the category label "Loan Status" is weak, "Contact Information" can be used as a pseudo column type. The contact information is generally a mobile phone number. Based on the way the mobile phone number is generated, a mobile phone number with the same number of rows as the original data can be forged as pseudo column data.

[0066] S103: Construct pseudo-column data according to the data generation rules corresponding to the pseudo-column type.

[0067] There can be various pseudo-column types, and different types of column data have different data generation rules. Based on type, common types of column data can include numeric attribute data, categorical attribute data, and business type data.

[0068] Numerical attribute data refers to data that is in numerical form and whose values ​​do not exhibit regularity, such as annual income and height.

[0069] Category attribute data is a limited number of fixed types, such as province, city, district, etc.

[0070] Business type data is generated according to arrangement rules, such as ID card number, mobile phone number, etc.

[0071] Taking numerical attribute data as an example, when the data corresponding to the pseudo-column type belongs to numerical attribute data, numerical data is selected from the numerical range corresponding to the numerical attribute data in a uniformly distributed sampling method under each category label to generate pseudo-column data.

[0072] Taking the category label "Loan or Not" as an example, if the original data contains 100 rows, then the "Age" column needs to be fabricated to have 100 age values. Assuming that the age for the "Loan" category label is set to "40 years old", then the age for the "No Loan" category label also needs to be set to "40 years old" to reduce the impact of the "Age" column data on the "Loan or Not" classification task.

[0073] Taking category attribute data as an example, when the data corresponding to the pseudo-column type belongs to category attribute data, the category data is selected from the category set corresponding to the category attribute data by using equal probability sampling under each category label to generate pseudo-column data.

[0074] Taking the category label "Loan Status" as an example, the target column data of the category attribute is "Province". Assuming that the "Provinces" belonging to "Loan" are set to Province A, Province B and Province C, with a ratio of 1:1:2, then the "Provinces" belonging to "No Loan" also need to be set to Province A, Province B and Province C, with a ratio of 1:1:2, so as to reduce the impact of the "Province" column data on the "Loan Status" classification task.

[0075] Taking business type data as an example, when the data corresponding to the pseudo-column type belongs to business type data, the business type data is divided into numerical attribute data and category attribute data; numerical data is generated according to the data generation rules corresponding to the numerical attribute data; category data is generated according to the data generation rules corresponding to the category attribute data; and the numerical data and category data are combined as pseudo-column data.

[0076] Taking "Mobile Number" as an example, the first 3 digits are the network identification number, digits 4-7 are the area code (HLR), and digits 8-11 are the user number (randomly assigned). The first 3 digits are category attribute data, digits 4-7 (area code) are category attribute data, and digits 8-11 are numerical attribute data. For mobile numbers, digits 8-11 can be generated using a random assignment method.

[0077] The composition of a single column of business type data can be divided into numerical attribute data and category attribute data. For specific forgery methods, please refer to the forgery methods of numerical attribute data and category attribute data introduced above, which will not be repeated here.

[0078] As can be seen from the above technical solution, a classification label is determined based on the machine learning classification task; each row of data in the original data has a corresponding classification label. The classification label can be regarded as the classification result corresponding to the classification task. In order to reduce the impact of pseudo-column data on machine learning, when constructing pseudo-column data, data types with weak or no correlation with the classification label can be selected. In this way, even if pseudo-column data is added to the original data, its weak correlation with the classification label can greatly reduce the impact on machine learning classification. Furthermore, in order to avoid the direct detection of pseudo-column data, it is also necessary to ensure that the type of pseudo-column data has a certain correlation with the types of data in each column of the original data. Therefore, in this application, according to the set number of pseudo-column data columns, pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data can be selected. Different types of column data have their own corresponding data generation rules. Pseudo-column data can be constructed according to the data generation rules corresponding to the pseudo-column type. In this technical solution, the classification label related to machine learning is fully considered when constructing pseudo-column data, reducing the impact of pseudo-column data on machine learning classification. Furthermore, it comprehensively considers the types of data in each column of the original data, avoiding the direct detection of pseudo-column data due to excessive deviation from the data in the original data, thus improving the concealment and usability of pseudo-column data.

[0079] After constructing the pseudo-column data, the pseudo-column data can be embedded with a watermark and then inserted into the original data.

[0080] In practice, watermarks can be embedded into pseudo-column data first; then, the watermarked pseudo-column data can be added to the original data in a uniform distribution.

[0081] For example, if the original data contains 10 columns and the pseudo-column data contains 2 columns, then one pseudo-column data can be inserted every 5 columns of the original data.

[0082] In addition to adding pseudo-column data to the original data in a uniformly distributed manner, pseudo-column data can also be directly inserted into the end of the original data column. In this embodiment, the method of inserting pseudo-column data is not limited.

[0083] Figure 2 A schematic diagram of a pseudo-column data generation device provided in an embodiment of this application includes a determining unit 21, a selecting unit 22, and a constructing unit 23;

[0084] Unit 21 is used for machine learning-based classification tasks to determine classification labels;

[0085] Selection unit 22 is used to select pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data, according to the set number of pseudo-column data columns;

[0086] Construction unit 23 is used to construct pseudo-column data according to the data generation rules corresponding to the pseudo-column type.

[0087] Optionally, the selection unit includes a type determination sub-unit and a filtering sub-unit;

[0088] The type determination subunit is used to determine the candidate data types that meet the correlation requirements with various types of data in the original data using an association algorithm;

[0089] The filtering sub-unit is used to filter out pseudo-column types from the candidate data types that meet the requirement of non-relevance to the category label, based on the number of pseudo-column data columns.

[0090] Optionally, the construction unit is used to select numerical data from the numerical range corresponding to the numerical attribute data under each category label in a uniformly distributed sampling manner according to the same range, so as to generate pseudo-column data when the data corresponding to the pseudo-column type belongs to numerical attribute data.

[0091] Optionally, the construction unit is used to select category data from the category set corresponding to the category attribute data under each category label by using equal probability sampling when the data corresponding to the pseudo-column type belongs to category attribute data, so as to generate pseudo-column data.

[0092] Optionally, the construction unit is used to divide the business type data into numerical attribute data and category attribute data when the data corresponding to the pseudo-column type belongs to the business type data; generate numerical data according to the data generation rules corresponding to the numerical attribute data; generate category data according to the data generation rules corresponding to the category attribute data; and combine the numerical data and category data as pseudo-column data.

[0093] Optionally, it also includes an insertion unit;

[0094] The insertion unit is used to embed pseudo-column data with a watermark and insert it into the original data.

[0095] Optionally, the insertion unit includes embedded sub-units and added sub-units;

[0096] Embedded sub-units are used to embed watermarks into pseudo-column data;

[0097] Add a sub-cell to add the pseudo-column data with embedded watermark to the original data in a uniform distribution.

[0098] Figure 2 For a description of the features in the corresponding embodiments, please refer to Figure 1The relevant descriptions of the corresponding embodiments will not be repeated here.

[0099] As can be seen from the above technical solution, a classification label is determined based on the machine learning classification task; each row of data in the original data has a corresponding classification label. The classification label can be regarded as the classification result corresponding to the classification task. In order to reduce the impact of pseudo-column data on machine learning, when constructing pseudo-column data, data types with weak or no correlation with the classification label can be selected. In this way, even if pseudo-column data is added to the original data, its weak correlation with the classification label can greatly reduce the impact on machine learning classification. Furthermore, in order to avoid the direct detection of pseudo-column data, it is also necessary to ensure that the type of pseudo-column data has a certain correlation with the types of data in each column of the original data. Therefore, in this application, according to the set number of pseudo-column data columns, pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data can be selected. Different types of column data have their own corresponding data generation rules. Pseudo-column data can be constructed according to the data generation rules corresponding to the pseudo-column type. In this technical solution, the classification label related to machine learning is fully considered when constructing pseudo-column data, reducing the impact of pseudo-column data on machine learning classification. Furthermore, it comprehensively considers the types of data in each column of the original data, avoiding the direct detection of pseudo-column data due to excessive deviation from the data in the original data, thus improving the concealment and usability of pseudo-column data.

[0100] Figure 3 A structural diagram of an electronic device provided in an embodiment of this application, such as... Figure 3 As shown, the electronic device includes: a memory 20 for storing computer programs;

[0101] The processor 31 is used to execute a computer program to implement the steps of the method for generating pseudo-column data as described in the above embodiment.

[0102] The electronic devices provided in this embodiment may include, but are not limited to, smartphones, tablets, laptops, or desktop computers.

[0103] The processor 31 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 31 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 31 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 31 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 31 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0104] The memory 20 may include one or more computer-readable storage media, which may be non-transitory. The memory 20 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 20 is used to store at least the following computer program 201, which, after being loaded and executed by the processor 31, is capable of implementing the relevant steps of the pseudo-column data generation method disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 20 may also include an operating system 202 and data 203, and the storage method may be temporary or permanent storage. The operating system 202 may include Windows, Unix, Linux, etc. The data 203 may include, but is not limited to, the number of pseudo-column data columns and the data generation rules corresponding to the pseudo-column types.

[0105] In some embodiments, the electronic device may further include a display screen 32, an input / output interface 33, a communication interface 34, a power supply 35, and a communication bus 36.

[0106] Those skilled in the art will understand that Figure 3 The structures shown do not constitute a limitation on electronic devices and may include more or fewer components than those shown.

[0107] It is understood that if the method for generating pseudo-column data in the above embodiments is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and executes all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes: USB flash drive, mobile hard disk, read-only memory (ROM), random access memory (RAM), electrically erasable programmable ROM, register, hard disk, removable disk, CD-ROM, magnetic disk, or optical disk, and other media capable of storing program code.

[0108] Based on this, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the pseudo-column data generation method described above.

[0109] The foregoing has provided a detailed description of a method, apparatus, device, and computer-readable storage medium for generating pseudo-column data according to embodiments of this application. The various embodiments are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0110] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] The foregoing has provided a detailed description of a method, apparatus, device, and computer-readable storage medium for generating pseudo-column data provided in this application. Specific examples have been used to illustrate the principles and implementation methods of the invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of the invention. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from the principles of the invention, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A method for generating pseudo-column data, characterized in that, include: The classification task is based on machine learning to determine the classification labels; Based on the set number of pseudo-column data columns, select pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data; Based on the data generation rules corresponding to the pseudo-column type, construct pseudo-column data; The pseudo-column data is embedded with a watermark and then inserted into the original data; The process of constructing pseudo-column data based on the data generation rules corresponding to the pseudo-column type includes: When the data corresponding to the pseudo-column type is numerical attribute data, numerical data is selected from the numerical range corresponding to the numerical attribute data under each category label in a uniformly distributed sampling manner within the same range to generate pseudo-column data.

2. The method for generating pseudo-column data according to claim 1, characterized in that, The step of selecting pseudo-column types that meet the non-relevance requirement to the classification label and the relevance requirement to each type of data in the original data, according to the set number of pseudo-column data columns, includes: The association algorithm is used to determine the candidate data types that meet the correlation requirements with various types of data in the original data; Based on the number of pseudo-column data columns, pseudo-column types that meet the non-relevance requirement with the classification label are selected from the candidate data types.

3. The method for generating pseudo-column data according to claim 1, characterized in that, The process of constructing pseudo-column data based on the data generation rules corresponding to the pseudo-column type includes: When the data corresponding to the pseudo-column type belongs to category attribute data, category data is selected from the category set corresponding to the category attribute data using equal probability sampling under each category label to generate pseudo-column data.

4. The method for generating pseudo-column data according to claim 1, characterized in that, The process of constructing pseudo-column data based on the data generation rules corresponding to the pseudo-column type includes: When the data corresponding to the pseudo-column type belongs to business type data, the business type data is divided into numerical attribute data and category attribute data; Generate numerical data according to the data generation rules corresponding to the numerical attribute data; Generate category data according to the data generation rules corresponding to the category attribute data; The numerical data and the category data are combined as pseudo-column data.

5. The method for generating pseudo-column data according to claim 1, characterized in that, The step of embedding the pseudo-column data with a watermark and then inserting it into the original data includes: Embed a watermark in the pseudo-column data; The pseudo-column data with embedded watermarks is added to the original data in a uniform distribution.

6. A device for generating pseudo-column data, characterized in that, This includes determining the element, selecting the element, and constructing the element; The determining unit is used to determine the classification label based on a machine learning-based classification task; The selection unit is used to select pseudo-column types that meet the non-correlation requirement with the classification label and the correlation requirement with each type of data in the original data, according to the set number of pseudo-column data columns. The construction unit is used to construct pseudo-column data according to the data generation rules corresponding to the pseudo-column type; The construction unit is used to select numerical data from the numerical range corresponding to the numerical attribute data under each category label in a uniformly distributed sampling method according to the same range, so as to generate pseudo-column data when the data corresponding to the pseudo-column type belongs to numerical attribute data. It also includes an insertion unit; the insertion unit is used to embed pseudo-column data with a watermark and insert it into the original data.

7. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the method for generating pseudo-column data as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method for generating pseudo-column data as described in any one of claims 1 to 5.

Images