Trusted environment sample data publishing method, device, equipment and readable storage medium

Abstract

A method, apparatus, device, and readable storage medium for publishing sample data in a trusted environment relate to the field of data processing technology, and include traversing a preset real data set to obtain data attributes; adding data attributes to real data in a real data subset to obtain a real data subset to be processed; perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set; and publishing the target sample data set. This application generates sample data by introducing differential privacy technology, that is, adding data attribute information to real data, and perturbing the real data with added data attribute information based on the privacy budget and perturbation mechanism before publishing the data, thereby obtaining noisy data to automatically form sample data and publish it, thereby achieving the goal of obtaining approximate data without leaking sensitive information therein, effectively reducing costs and improving the efficiency of sample data publishing.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, apparatus, device, and readable storage medium for publishing sample data in a trusted environment. Background Art

[0002] The trusted environment in privacy computing provides a secure isolation mechanism to protect sensitive data and prevent unauthorized access and data leakage. Figure 1 As shown, data consumers can use sample data provided by the trusted environment to initialize and debug their models. The initialized model is then sent to the trusted environment and trained with real data to improve performance. After training, the final model is released to the data consumer for performance verification, but only the model parameters are included, without the original data. This process ensures that sensitive data is protected while allowing for model construction and performance verification.

[0003] In related technologies, the most basic method for generating sample data is to randomly select a small amount of data from the full data set and publish it. However, this method will obviously lead to the leakage of sensitive information in this data. Therefore, this data must be desensitized manually or through automated scripts before being published. However, in actual production, due to the differences in data attributes, the amount of sensitive information contained, the degree of data sensitivity, and even the quality of data among different data sets, manual data desensitization will increase significant labor costs. Using automated scripts for desensitization requires setting different desensitization strategies and writing different scripts to address these issues, which is not only costly but also inefficient. Summary of the Invention

[0004] The present application provides a method, apparatus, device and readable storage medium for publishing sample data in a trusted environment, which can solve the technical problems in the prior art of high cost and low efficiency caused by publishing sample data through manual or automatic scripts.

[0005] In a first aspect, an embodiment of the present application provides a method for publishing sample data in a trusted environment, the method comprising:

[0006] Traversing a preset real data set to obtain at least one data attribute corresponding to the real data set;

[0007] Adding the data attributes to the real data in the real data subset to obtain a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set;

[0008] Perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set;

[0009] The target sample dataset is published.

[0010] In conjunction with the first aspect, in one embodiment, adding the data attribute to the real data in the real data subset includes:

[0011] Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

[0012] In combination with the first aspect, in one implementation, the data attributes include numerical attributes and category attributes.

[0013] In combination with the first aspect, in one implementation, the perturbation mechanism corresponding to the numerical attribute is a Laplace mechanism, and the perturbation mechanism corresponding to the categorical attribute is an exponential mechanism.

[0014] In conjunction with the first aspect, in one embodiment, perturbing the real data in the to-be-processed real data subset according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample dataset includes:

[0015] perturbing the real data containing the numerical attribute in the real data subset to be processed based on the privacy budget, the sensitivity corresponding to the numerical attribute, and the Laplace mechanism to obtain first sample data;

[0016] perturbing the real data containing the category attribute in the real data subset to be processed based on the privacy budget and the exponential mechanism to obtain second sample data;

[0017] The first sample data and the second sample data constitute a target sample data set.

[0018] In a second aspect, an embodiment of the present application provides a trusted environment sample data publishing device, the trusted environment sample data publishing device comprising:

[0019] An attribute acquisition module, configured to traverse a preset real data set to acquire at least one data attribute corresponding to the real data set;

[0020] an attribute adding module, configured to add the data attributes to the real data in the real data subset to obtain a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set;

[0021] A data perturbation module, configured to perturb the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set;

[0022] A data publishing module is used to publish the target sample data set.

[0023] In conjunction with the second aspect, in one embodiment, the attribute adding module is specifically configured to:

[0024] Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

[0025] In conjunction with the second aspect, in one implementation, the data attributes include numerical attributes and category attributes.

[0026] In combination with the second aspect, in one implementation, the perturbation mechanism corresponding to the numerical attribute is a Laplace mechanism, and the perturbation mechanism corresponding to the categorical attribute is an exponential mechanism.

[0027] In conjunction with the second aspect, in one embodiment, the data perturbation module is specifically configured to:

[0028] perturbing the real data containing the numerical attribute in the real data subset to be processed based on the privacy budget, the sensitivity corresponding to the numerical attribute, and the Laplace mechanism to obtain first sample data;

[0029] perturbing the real data containing the category attribute in the real data subset to be processed based on the privacy budget and the exponential mechanism to obtain second sample data;

[0030] The first sample data and the second sample data constitute a target sample data set.

[0031] In a third aspect, an embodiment of the present application provides a trusted environment sample data publishing device, which includes a processor, a memory, and a trusted environment sample data publishing program stored on the memory and executable by the processor, wherein when the trusted environment sample data publishing program is executed by the processor, the steps of the trusted environment sample data publishing method as described above are implemented.

[0032] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a trusted environment sample data publishing program is stored, wherein when the trusted environment sample data publishing program is executed by a processor, the steps of the trusted environment sample data publishing method as described above are implemented.

[0033] The beneficial effects of the technical solutions provided in the embodiments of the present application include:

[0034] By traversing the preset real data set to obtain at least one data attribute corresponding to the real data set; adding the data attribute to the real data in the real data subset to obtain the real data subset to be processed, the real data subset includes real data randomly selected from the real data set; perturbing the real data in the real data subset to be processed according to the preset privacy budget and the preset perturbation mechanism to obtain the target sample data set; and publishing the target sample data set. This application generates sample data by introducing differential privacy technology, that is, adding data attribute information to the randomly selected real data, and perturbing the real data with added data attribute information based on the privacy budget and the perturbation mechanism before publishing the data, to obtain noisy data, so as to automatically form sample data and publish it without building automatic scripts for different data sets, thereby achieving the goal of obtaining approximate data without leaking sensitive information therein, and effectively reducing the human and material costs required in the process of data desensitization or data generation, and improving the efficiency of sample data publishing. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the trusted environment model training process;

[0036] Figure 2 This is a flow chart of an embodiment of a method for publishing sample data in a trusted environment of this application;

[0037] Figure 3 A schematic diagram of the data attribute acquisition process involved in the embodiment of this application;

[0038] Figure 4 This is a schematic diagram of the sample data synthesis process involved in the embodiment of this application;

[0039] Figure 5 This is a schematic diagram of the hardware structure of the trusted environment sample data publishing device involved in the embodiment of the present application. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0041] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0042] In a first aspect, an embodiment of the present application provides a method for publishing sample data in a trusted environment.

[0043] In one embodiment, referring to Figure 2 , Figure 2 This is a flow chart of an embodiment of the method for publishing sample data in a trusted environment of this application. Figure 2 As shown, the trusted environment sample data publishing method includes:

[0044] Step S10: traverse a preset real data set to obtain at least one data attribute corresponding to the real data set.

[0045] For example, it can be understood that privacy computing is an interdisciplinary technology system that integrates cryptography, data science, artificial intelligence, security hardware, and other fields. Based on protecting the privacy and security of data throughout its life cycle, it enables computing and analysis of data in an encrypted or non-transparent state, thereby promoting the circulation and integration of data elements and effectively extracting their value. While ensuring the security and privacy of the original data of all parties, it completes the fusion computing of multi-party data, making the data of multiple parties "available but invisible."

[0046] The trusted environment in privacy computing provides a secure isolation mechanism for protecting sensitive data from unauthorized access and data leakage. However, when generating sample data, it is necessary to avoid the leakage of sensitive information. Data desensitization is often achieved through manual or automated scripts. This involves deleting or replacing sensitive data attributes, such as replacing certain digits of a phone number with "X" or deleting the user's real name, to generate sample data. Machine learning methods can also be used to generate sample data. This involves training a data generator model with real data and then synthesizing data based on the data generator model. However, in actual production, manual data desensitization increases significant labor costs due to the varying data attributes, amount of sensitive information contained, data sensitivity, and even data quality of different datasets. Automated desensitization, on the other hand, requires setting different desensitization strategies and writing different scripts to address these issues.

[0047] Therefore, as datasets continue to grow, how to write and update desensitization strategies also needs to be considered. For example, using some additional data obtained externally to construct sample data may also leak sensitive information. For example, by comparing certain publicly available data with data containing a person's attributes (birthday, gender, etc.) obtained from other channels, it is possible to identify the corresponding individual in the original table and obtain sensitive information. Although the method of generating sample data through machine learning models can alleviate the above problems, this method requires the use of a large amount of data to train the data generation model, otherwise it is easy to cause model underfitting; and the model needs to be trained once for each dataset, which will also consume a large amount of computing resources in the trusted environment.

[0048] To address the aforementioned issues, this embodiment constructs sample data using differential privacy. Differential privacy, as a privacy protection method, provides a mathematically provable, strong privacy protection method. In this embodiment, before publishing the sample data, data attribute information is obtained from the real dataset corresponding to the sample data to be published. This involves traversing the real dataset to obtain the attributes contained in each piece of real data. These attributes include salary, height, gender, and work location, among others. The specific data attributes can be determined based on actual needs and are not limited here.

[0049] Step S20: adding the data attributes to the real data in the real data subset to obtain a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set.

[0050] Exemplarily, in this embodiment, multiple real data are first randomly selected from the real data set to form a real data subset; then the data attributes are added to the corresponding real data in the real data subset respectively, so that the real data includes the data attributes, and then the real data subset to be processed is obtained, and the real data after use will no longer participate in the synthesis of subsequent sample data.

[0051] Step S30: perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set.

[0052] For example, in this embodiment, when generating sample data through differential privacy, the degree of differential privacy protection can be controlled by adjusting the value of the privacy budget ε. Specifically, the smaller ε, the stronger the privacy protection, but the lower the availability of the calculation results, and vice versa. Therefore, in this embodiment, the data owner can set the global privacy budget variable ε based on the strength of privacy requirements to form a preset privacy budget. Simultaneously, this embodiment also protects privacy by introducing a perturbation mechanism into the data, making it impossible to accurately infer the true data from the perturbed data. Therefore, by perturbing the real data in the real data subset to be processed using the privacy budget and perturbation mechanism, noisy data can be obtained, and the target sample dataset can be generated. It should be noted that the perturbation mechanism can preferably be a Laplace mechanism, a Gaussian mechanism, an exponential mechanism, etc.

[0053] Step S40: publishing the target sample data set.

[0054] For example, in this embodiment, after the target sample data set is generated, the target sample data set can be published to the data user, so that the data user can initialize the model according to the target sample data set, and then transmit the initialized model back to the data owner, and the data owner uses real data to train and publish the model.

[0055] It can be seen that this embodiment generates sample data by introducing differential privacy technology, that is, adding data attribute information to real data, and perturbing the real data with added data attribute information based on the privacy budget and perturbation mechanism before publishing the data, so as to obtain noisy data, so as to automatically form sample data and publish it without building automatic scripts for different data sets, thereby achieving the goal of obtaining approximate data without leaking sensitive information therein, and effectively reducing the human and material costs required in the process of data desensitization or data generation, and improving the efficiency of sample data publishing.

[0056] Furthermore, in one embodiment, the data attributes include numerical attributes and category attributes.

[0057] For example, in this embodiment, as shown in Table 1, data attributes can generally be divided into numerical attributes and categorical attributes. Numerical attributes represent continuous data, such as salary, height, sales, etc., while categorical attributes represent non-continuous data, such as gender, work location, number of properties, etc.

[0058] Table 1 Data attributes

[0059]

[0060]

[0061] See also Figure 3 As shown, this embodiment will traverse the real data set to obtain the data set of each attribute, and traverse all attributes of the data set to determine whether each attribute is a numerical attribute or a categorical attribute, and then calculate the data range corresponding to the numerical attribute and the data set corresponding to the categorical attribute.

[0062] Furthermore, in one embodiment, adding the data attribute to the real data in the real data subset includes:

[0063] Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

[0064] For example, in this embodiment, each data attribute is assigned to a piece of real data in the real data subset, so that each piece of real data in the real data subset contains a data attribute, thereby ensuring that all attributes are covered, and then the single piece of data is perturbed. It can be understood that, assuming that a piece of data has d attributes, d pieces of real data will be used for each synthesized sample data. Therefore, this embodiment can ensure that the availability of the sample data is improved by sampling and perturbing each attribute using a single piece of data.

[0065] Furthermore, in one embodiment, the perturbation mechanism corresponding to the numerical attribute is a Laplace mechanism, and the perturbation mechanism corresponding to the categorical attribute is an exponential mechanism.

[0066] For example, in this embodiment, different perturbation methods are used for different attributes, that is, the Laplace mechanism is used as the perturbation mechanism for numerical attributes, and the exponential mechanism is used as the perturbation mechanism for categorical attributes, so as to further improve the privacy of the sample data.

[0067] Furthermore, in one embodiment, perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set includes:

[0068] perturbing the real data containing the numerical attribute in the real data subset to be processed based on the privacy budget, the sensitivity corresponding to the numerical attribute, and the Laplace mechanism to obtain first sample data;

[0069] perturbing the real data containing the category attribute in the real data subset to be processed based on the privacy budget and the exponential mechanism to obtain second sample data;

[0070] The first sample data and the second sample data constitute a target sample data set.

[0071] For example, in this embodiment, see Figure 4 As shown in the figure, before perturbing the real data, the data attribute set is traversed to select unused data. The perturbation mechanism is then determined based on whether the data is a numerical attribute. For numerical attributes, the real data is perturbed and noised using the Laplace mechanism that satisfies differential privacy based on the data range. For categorical attributes, the data is perturbed and noised using the exponential mechanism that satisfies differential privacy based on the data set of the categorical attribute.

[0072] Specifically, for numerical attributes, when generating sample data, the data owner sets a privacy budget ε and the sensitivity Δf for the numerical attribute. Sensitivity refers to the maximum change in the query function output when changing a single item in a dataset. The data owner inputs the privacy budget and sensitivity into the Laplace function to sample a noise value. This noise value is then added to the true data to obtain noisy data that satisfies ε-differential privacy, forming the first sample data. For example, if the true data is 1, and the noise value obtained after perturbation sampling is 0.001, then 1 and 0.001 are added to obtain the sample data.

[0073] For categorical attributes, we assign corresponding weights to different real data using a scoring function and perturb them using an exponential mechanism. This yields noisy data that satisfies ε-differential privacy, forming the second sample data. It should be noted that, unlike numerical attributes, the perturbation output of a categorical attribute is a value in the attribute set. For example, if the categorical attribute is work location, and the work locations include A, B, and C, then the perturbation output of the categorical attribute is one of A, B, and C.

[0074] In summary, in the privacy computing scenario, this embodiment reduces the manpower and material costs consumed in the data desensitization or data generation process by traversing the real data set to obtain the attribute set; divides the data attributes into two categories: numerical attributes and categorical attributes, and automatically generates perturbation rules by traversing the data, that is, different perturbation methods are used for different attributes, and each attribute is sampled and perturbed using a single piece of data. While ensuring that the requirements of differential privacy are met, the sample data is made as close as possible to the actual distribution of the data, the availability of the sample data is improved, and to a certain extent, the problem of over-desensitization caused by the use of data desensitization technology is solved. It can be seen that this embodiment eliminates the privacy leakage risk of sample data released in the privacy computing scenario through differential privacy technology, and provides data users with sample data that meets the real distribution as much as possible.

[0075] In a second aspect, an embodiment of the present application also provides a trusted environment sample data publishing device.

[0076] In one embodiment, the trusted environment sample data publishing device includes:

[0077] An attribute acquisition module, configured to traverse a preset real data set to acquire at least one data attribute corresponding to the real data set;

[0078] an attribute adding module, configured to add the data attributes to the real data in the real data subset to obtain a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set;

[0079] A data perturbation module, configured to perturb the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set;

[0080] A data publishing module is used to publish the target sample data set.

[0081] Furthermore, in one embodiment, the attribute adding module is specifically configured to:

[0082] Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

[0083] Furthermore, in one embodiment, the data attributes include numerical attributes and category attributes.

[0084] Furthermore, in one embodiment, the perturbation mechanism corresponding to the numerical attribute is a Laplace mechanism, and the perturbation mechanism corresponding to the categorical attribute is an exponential mechanism.

[0085] Furthermore, in one embodiment, the data perturbation module is specifically configured to:

[0086] perturbing the real data containing the numerical attribute in the real data subset to be processed based on the privacy budget, the sensitivity corresponding to the numerical attribute, and the Laplace mechanism to obtain first sample data;

[0087] perturbing the real data containing the category attribute in the real data subset to be processed based on the privacy budget and the exponential mechanism to obtain second sample data;

[0088] The first sample data and the second sample data constitute a target sample data set.

[0089] Among them, the functional implementation of each module in the above-mentioned trusted environment sample data publishing device corresponds to each step in the above-mentioned trusted environment sample data publishing method embodiment, and its functions and implementation processes are no longer repeated here.

[0090] In a third aspect, an embodiment of the present application provides a trusted environment sample data publishing device, which may be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0091] Reference Figure 5 , Figure 5 FIG1 is a schematic diagram of the hardware structure of the trusted environment sample data publishing device involved in the embodiment of the present application. In the embodiment of the present application, the trusted environment sample data publishing device may include a processor, a memory, a communication interface, and a communication bus.

[0092] The communication bus may be of any type and is used to interconnect the processor, memory, and communication interface.

[0093] Communication interfaces include input / output (I / O) interfaces, physical interfaces, and logical interfaces, which are used to interconnect components within the trusted environment sample data distribution device, as well as interfaces used to interconnect the trusted environment sample data distribution device with other devices (such as other computing devices or user devices). Physical interfaces can be Ethernet, fiber, or ATM interfaces; user devices can be displays, keyboards, and other devices.

[0094] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0095] The processor may be a general-purpose processor that can call a trusted environment sample data publishing program stored in a memory and execute the trusted environment sample data publishing method provided in the embodiments of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the trusted environment sample data publishing program is called can refer to the various embodiments of the trusted environment sample data publishing method of the present application, and will not be repeated here.

[0096] Those skilled in the art will understand that Figure 5The hardware structure shown in the figure does not constitute a limitation to the present application and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0097] In a fourth aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0098] The readable storage medium of the present application stores a trusted environment sample data publishing program, wherein when the trusted environment sample data publishing program is executed by a processor, the steps of the trusted environment sample data publishing method as described above are implemented.

[0099] Among them, the method implemented when the trusted environment sample data publishing program is executed can refer to the various embodiments of the trusted environment sample data publishing method of this application, and will not be repeated here.

[0100] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0101] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0102] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0103] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0104] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0105] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of the present application.

[0106] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for publishing sample data in a trusted environment, characterized in that: The trusted environment sample data publishing method includes: Traversing a preset real data set to obtain at least one data attribute corresponding to the real data set; Adding the data attribute to the real data in the real data subset so that each real data in the real data subset includes a data attribute, thereby obtaining a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set; Perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set; Publishing the target sample data set; Among them, for each data attribute, a single real data is used for perturbation. data attributes, each time a sample data is synthesized, it will be used Real data.

2. The trusted environment sample data publishing method according to claim 1, characterized in that: Adding the data attribute to the real data in the real data subset includes: Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

3. The trusted environment sample data publishing method according to claim 1, characterized in that: The data attributes include numerical attributes and category attributes.

4. The trusted environment sample data publishing method according to claim 3, characterized in that: The perturbation mechanism corresponding to the numerical attribute is the Laplace mechanism, and the perturbation mechanism corresponding to the categorical attribute is the exponential mechanism.

5. The trusted environment sample data publishing method according to claim 4, characterized in that: The method of perturbing the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set includes: perturbing the real data containing the numerical attribute in the real data subset to be processed based on the privacy budget, the sensitivity corresponding to the numerical attribute, and the Laplace mechanism to obtain first sample data; perturbing the real data containing the category attribute in the real data subset to be processed based on the privacy budget and the exponential mechanism to obtain second sample data; The first sample data and the second sample data constitute a target sample data set.

6. A trusted environment sample data publishing device, characterized in that: The trusted environment sample data publishing device includes: An attribute acquisition module, configured to traverse a preset real data set to acquire at least one data attribute corresponding to the real data set; an attribute adding module, configured to add the data attribute to the real data in the real data subset, so that each piece of real data in the real data subset contains a data attribute, thereby obtaining a real data subset to be processed, wherein the real data subset includes real data randomly selected from the real data set; A data perturbation module, configured to perturb the real data in the real data subset to be processed according to a preset privacy budget and a preset perturbation mechanism to obtain a target sample data set; A data publishing module, which is used to publish the target sample data set; Among them, for each data attribute, a single real data is used for perturbation. data attributes, each time a sample data is synthesized, it will be used Real data.

7. The trusted environment sample data publishing device according to claim 6, characterized in that: The attribute adding module is specifically used for: Each data attribute is assigned to the real data in the real data subset, so that each piece of real data in the real data subset contains the data attribute.

8. The trusted environment sample data publishing device according to claim 6, characterized in that: The data attributes include numerical attributes and category attributes.

9. A trusted environment sample data publishing device, characterized in that: The trusted environment sample data publishing device includes a processor, a memory, and a trusted environment sample data publishing program stored in the memory and executable by the processor, wherein when the trusted environment sample data publishing program is executed by the processor, the steps of the trusted environment sample data publishing method as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a trusted environment sample data publishing program, wherein when the trusted environment sample data publishing program is executed by a processor, the steps of the trusted environment sample data publishing method according to any one of claims 1 to 5 are implemented.

Images