Data protection method and device based on differential privacy, medium and equipment

Abstract

The embodiment of the invention discloses a data protection method and device based on differential privacy, a medium and equipment, and relates to the technical field of data protection.The data protection method comprises the steps that firstly, a privacy budget is determined, the privacy protection degree of data is measured, then the data is preprocessed, the data can be suitable for differential privacy processing, and the data protection efficiency is improved. Noise is added by adopting a fusion protection mechanism, the fusion protection mechanism is obtained based on fusion of a Laplacian mechanism and a Gaussian mechanism, the two fused mechanisms are also suitable for numeric data and are symmetrically distributed, and Laplacian is close to Gaussian normal distribution, so that the noise is reduced, and the noise is reduced. According to the method, the data can be effectively fused and the validity of the data can be kept, and a mechanism obtained by fusion has the advantages of processing extreme values by a Laplace mechanism and processing continuous data by a Gaussian mechanism, so that the requirement that the same type of data has different characteristics is met under the condition of ensuring privacy protection and data availability; and the effect of data protection based on differential privacy is improved.

Description

Technical Field

[0001] The present application relates to the field of data protection technology, and specifically to a data protection method, apparatus, medium and equipment based on differential privacy. Background Art

[0002] Differential privacy is a method in cryptography that aims to maximize the accuracy of data queries when querying from a statistical database while minimizing the chance of identifying its records. Data protection technology based on differential privacy adds random interference elements during data processing, making it difficult for individual data to be singled out, thereby effectively avoiding potential data leakage risks. These interference elements are usually called noise. Adding noise will use different mechanisms depending on the type of data, but existing methods all use a single mechanism. Faced with the rapidly growing demand for data, even the same type of data will have different characteristics. The means of adding noise under a single mechanism can no longer meet the needs of diversified data characteristics, resulting in the existing data protection based on differential privacy being poor. Summary of the invention

[0003] The main purpose of this application is to provide a data protection method, device, medium and equipment based on differential privacy, aiming to solve the problem of poor data protection effect based on differential privacy in the prior art.

[0004] To achieve the above objectives, the technical solutions adopted in the embodiments of the present application are as follows:

[0005] In a first aspect, an embodiment of the present application provides a data protection method based on differential privacy, comprising the following steps:

[0006] Determine the privacy budget based on the need to protect data;

[0007] Determine the amount of noise to add to the data based on the privacy budget;

[0008] Preprocess the data to obtain the original data;

[0009] According to the noise scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

[0010] In a possible implementation manner of the first aspect, according to the noise scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data before data protection is achieved, the method further includes:

[0011] The Laplace mechanism and the Gaussian mechanism are integrated to obtain a fusion protection mechanism.

[0012] In a possible implementation of the first aspect, a Laplace mechanism and a Gaussian mechanism are integrated to obtain a fusion protection mechanism, including:

[0013] According to the Laplace mechanism and the Gaussian mechanism, a Laplace distribution curve and a Gaussian distribution curve are obtained respectively;

[0014] The Laplace distribution curve and the Gaussian distribution curve are fused to obtain a fusion protection mechanism.

[0015] In a possible implementation manner of the first aspect, a Laplace distribution curve and a Gaussian distribution curve are fused to obtain a fusion protection mechanism, including:

[0016] In the same coordinate system, respectively obtain the first ordinate value on the Laplace distribution curve and the second ordinate value on the Gaussian distribution curve corresponding to each abscissa value;

[0017] Determine a target ordinate value according to an average value of the first ordinate value and the second ordinate value;

[0018] According to each abscissa value and the target ordinate value, a fusion curve is determined to obtain a fusion protection mechanism.

[0019] In a possible implementation of the first aspect, according to the noise scale and based on the fusion protection mechanism, noise is added to the original data to obtain target data to implement data protection, including:

[0020] According to the noise scale and based on the fusion protection mechanism, noise with distribution that conforms to the fusion curve is added to the original data to obtain the target data to achieve data protection.

[0021] In a possible implementation manner of the first aspect, after determining the scale of noise added to the data according to the privacy budget, the method further includes:

[0022] Obtaining a first similarity between the Laplace distribution curve and the fusion curve, and a second similarity between the Gaussian distribution curve and the fusion curve;

[0023] The difference between the first similarity and the second similarity is used as an adjustment coefficient, and the noise scale is adjusted according to the adjustment coefficient to obtain a noise target scale;

[0024] According to the noise scale, noise is added to the original data based on the fusion protection mechanism to obtain the target data to achieve data protection, including:

[0025] According to the noise target scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data to achieve data protection.

[0026] In a possible implementation manner of the first aspect, preprocessing the data to obtain original data includes:

[0027] Clean the data to obtain the original data.

[0028] In a second aspect, an embodiment of the present application provides a data protection device based on differential privacy, including:

[0029] Budget determination module, which is used to determine the privacy budget according to the data protection requirements;

[0030] Noise determination module, which is used to determine the scale of noise added to the data according to the privacy budget;

[0031] Preprocessing module, the preprocessing module is used to preprocess the data to obtain the original data;

[0032] The noise adding module is used to add noise to the original data according to the noise scale and based on the fusion protection mechanism to obtain the target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

[0033] In a third aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program. When the computer program is loaded and executed by a processor, the differential privacy-based data protection method provided in any one of the first aspects above is implemented.

[0034] In a fourth aspect, an embodiment of the present application provides an electronic device, including a processor and a memory, wherein:

[0035] The memory is used to store computer programs;

[0036] The processor is used to load and execute a computer program so that the electronic device executes a data protection method based on differential privacy as provided in any one of the first aspects above.

[0037] Compared with the prior art, the beneficial effects of this application are:

[0038] The embodiments of the present application propose a data protection method, apparatus, medium and equipment based on differential privacy. The method includes: determining a privacy budget according to the protection requirements of the data; determining the scale of noise added to the data according to the privacy budget; preprocessing the data to obtain the original data; adding noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism. This application first determines the privacy budget and measures the degree of privacy protection for the data. It then preprocesses the data to make it suitable for differential privacy processing. Noise is then added to the original data obtained through preprocessing. The noise is added using a fusion protection mechanism. This mechanism is based on the fusion of the Laplace mechanism and the Gaussian mechanism. Since the two fused mechanisms are both applicable to numerical data and their distributions are symmetrical, and the Laplace is close to the Gaussian normal distribution, they can be effectively fused and retain the validity of the data. The fused mechanism has the advantages of the Laplace mechanism for processing extreme values ​​and the Gaussian mechanism for processing continuous data. While ensuring privacy protection and data availability, it meets the needs of the same type of data with different characteristics, thereby improving the effect of data protection based on differential privacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of the structure of an electronic device of a hardware operating environment involved in an embodiment of the present application;

[0040] Figure 2 A flowchart of a data protection method based on differential privacy provided in an embodiment of the present application;

[0041] Figure 3 A schematic diagram of a Laplace distribution curve, a Gaussian distribution curve, and a fusion curve in a data protection method based on differential privacy provided in an embodiment of the present application;

[0042] Figure 4 A schematic diagram of a module of a data protection device based on differential privacy provided in an embodiment of the present application;

[0043] Markings in the figure: 101 - processor, 102 - communication bus, 103 - network interface, 104 - user interface, 105 - memory. DETAILED DESCRIPTION

[0044] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0045] The main solution provided by the embodiment of the present application is: determine the privacy budget according to the protection requirements of the data; determine the scale of noise added to the data according to the privacy budget; preprocess the data to obtain the original data; add noise to the original data according to the noise scale and based on the fusion protection mechanism to obtain the target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

[0046] Differential privacy is a method in cryptography that aims to maximize the accuracy of data queries when querying from a statistical database while minimizing the chance of identifying its records. Data protection technology based on differential privacy adds random interference elements during data processing, making it difficult for individual data to be singled out, thereby effectively avoiding potential data leakage risks. These interference elements are usually called noise, and adding noise will use different mechanisms depending on the type of data. Differential privacy is a mathematical framework used to quantify the degree of protection of personal privacy during data release or algorithm processing, and these mechanisms are used as a way to achieve it.

[0047] For example, the index mechanism is mainly used to process queries for non-numerical data. Its core idea is to assign a score to each possible query output through a scoring function, and then determine the final query result based on these scores and the preset privacy budget. For example, in the recommendation system, the index mechanism can be used to protect the user's privacy information while providing personalized recommendation services to users. Specifically, the user's personal information (such as age, gender, interests, etc.) can be used as the input of the scoring function, and the recommended items can be used as the output in the query space, and then the final recommendation result can be determined through the index mechanism. In this way, even if the attacker has a lot of background knowledge, he cannot accurately infer the user's personal information through data analysis.

[0048] For numerical data, Gaussian or Laplace mechanisms are usually used to achieve privacy protection by adding noise that conforms to the Gaussian distribution or Laplace distribution to the data. The difference is that Laplace noise is a symmetrical noise distribution, and the noise distribution is discrete, which may have a greater impact on the precision and accuracy of the data. It has a long-tail feature, which makes the mechanism perform better when processing extreme values; Gaussian noise is a symmetrical continuous noise distribution, usually used to process continuous data. The noise distribution is continuous, which may affect the outliers and extreme values ​​of the data.

[0049] Although existing methods provide multiple mechanisms, the implementation process all adopts a single mechanism. Faced with the rapidly growing demand for data, even the same type of data will have different characteristics. For example, the same numerical data may have continuity characteristics or numerical response characteristics. The means of adding noise under a single mechanism can no longer meet the needs of diversified data characteristics, and cannot take into account the protection of privacy and the validity of data in the differential privacy processing process, resulting in the poor effect of existing data protection based on differential privacy.

[0050] Therefore, the present application provides a means to first determine the privacy budget and measure the degree of privacy protection of the data, then preprocess the data to make it suitable for differential privacy processing, and then add noise to the original data obtained by preprocessing. The addition of noise adopts a fusion protection mechanism, and the mechanism is based on the fusion of the Laplace mechanism and the Gaussian mechanism. Since the two fused mechanisms are both applicable to numerical data and their distributions are symmetrical, and the Laplace is close to the Gaussian normal distribution, it can effectively fuse and retain the validity of the data. The fused mechanism has the advantages of the Laplace mechanism for processing extreme values ​​and the Gaussian mechanism for processing continuous data. While ensuring privacy protection and data availability, it meets the needs of the same type of data with different characteristics, and improves the effect of data protection based on differential privacy.

[0051] Refer to the attached Figure 1 , attached Figure 1This is a schematic diagram of the structure of an electronic device of the hardware operating environment involved in the embodiment of the present application. The electronic device may include: a processor 101, such as a central processing unit (CPU), a communication bus 102, a user interface 104, a network interface 103, and a memory 105. Among them, the communication bus 102 is used to realize the connection and communication between these components. The user interface 104 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 104 may also include a standard wired interface and a wireless interface. The network interface 103 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WI-FI) interface). The memory 105 may optionally be a storage device independent of the aforementioned processor 101. The memory 105 may be a high-speed random access memory (RAM) memory, or it may be a stable non-volatile memory (NVM), such as at least one disk storage; the processor 101 may be a general-purpose processor, including a central processing unit, a network processor, etc., or it may be a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component.

[0052] Those skilled in the art will appreciate that Figure 1 The structure shown in the figure does not constitute a limitation on the electronic device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0053] As attached Figure 1 As shown, the memory 105 as a storage medium may include an operating system, a network communication module, a user interface module, and a data protection device based on differential privacy.

[0054] In the attached Figure 1 In the electronic device shown, the network interface 103 is mainly used for data communication with the network server; the user interface 104 is mainly used for data interaction with the user; the processor 101 and the memory 105 in the present application can be set in the electronic device, and the electronic device calls the differential privacy-based data protection device stored in the memory 105 through the processor 101, and executes the differential privacy-based data protection method provided in the embodiment of the present application.

[0055] Refer to the attached Figure 2 Based on the hardware device of the foregoing embodiment, the embodiment of the present application provides a data protection method based on differential privacy, comprising the following steps:

[0056] S10: Determine the privacy budget based on the need to protect data.

[0057] In the specific implementation process, the data mentioned in this embodiment is the data that needs to be protected. The privacy budget is a key parameter in differential privacy technology, which is used to quantify the degree of privacy protection. The smaller the privacy budget value, the stronger the indistinguishability and the stronger the security of the data. When the privacy budget is equal to 0, different outputs are completely indistinguishable; the larger the privacy budget value, the weaker the indistinguishability. When the privacy budget is equal to infinity, different outputs can definitely be distinguished. The value range of the privacy budget is usually determined according to the protection requirements for the data. It is usually determined by the maximum post-belief corresponding to the data query request (that is, the probability that the preset data set constitutes the user data set), and the preset privacy budget can be updated according to the value range. In this way, the quantitative relationship between the privacy budget and the possibility of differential attacks can be achieved, thereby maximizing data availability while ensuring data privacy.

[0058] S20: Determine the scale of noise added to the data based on the privacy budget.

[0059] In the specific implementation process, since the privacy budget directly determines the security and protection of the data, the scale of noise added to the data needs to be determined based on the privacy budget. In differential privacy protection, sensitivity is an indicator to measure the impact of data changes on query results, and privacy loss is used to quantify the risk of privacy leakage. By calculating sensitivity and privacy loss, the scale of noise that needs to be added can be determined to ensure that the requirements of the privacy budget are met.

[0060] S30: Preprocess the data to obtain original data.

[0061] In the specific implementation process, the data is preprocessed, such as data cleaning, format conversion, etc., to obtain the original data for adding noise to ensure that the data is suitable for differential privacy processing. For example, data cleaning, that is, preprocessing the data to obtain the original data, includes:

[0062] Clean the data to obtain the original data.

[0063] Data cleaning is an important part of data preprocessing. It refers to the process of identifying, correcting and removing erroneous, incomplete, inconsistent or repeated information in the data set through a series of technical means. There are usually processing methods such as deleting missing values, filling the mean, and interpolation methods to improve data accuracy, consistency and integrity. It should be noted that the embodiments of the present application are aimed at numerical data, and the data has many characteristics, such as user behavior data, including online time, number of clicks, etc., which have continuity and extreme value characteristics, because users will be online for a long time, and sometimes there will be special circumstances that cause the number of clicks to increase sharply; another example is medical and health data, including physiological parameters and probability of disease, etc., which not only have continuity and extreme value characteristics, but also belong to response data, because in data query, only the corresponding numerical results need to be output according to the query situation. Similarly, the data can also be user transaction amount data, academic research data, enterprise operation data, etc.

[0064] S40: adding noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

[0065] In the specific implementation process, the Gaussian mechanism and the Laplace mechanism achieve privacy protection by adding noise that conforms to the Gaussian distribution or Laplace distribution to the data. The difference is that Laplace noise is a symmetrical noise distribution, which is discrete and may have a greater impact on the precision and accuracy of the data. It has a long-tail feature, which makes the mechanism perform better when processing extreme values; Gaussian noise is a symmetrical continuous noise distribution, which is usually used to process continuous data. The noise distribution is continuous and may have an impact on the outliers and extreme values ​​of the data. The data after adding noise through the fusion protection mechanism is the target data that achieves data protection. After completing differential privacy processing, it can be used for publication or third-party analysis, while ensuring that personal privacy will not be leaked.

[0066] In one embodiment, before adding noise to the original data based on the fusion protection mechanism according to the noise scale and obtaining the target data to implement data protection, the method further includes:

[0067] The Laplace mechanism and the Gaussian mechanism are integrated to obtain a fusion protection mechanism.

[0068] In the specific implementation process, by fusing the Laplace mechanism and the Gaussian mechanism, a fusion protection mechanism is obtained in advance for rapid application in the use phase. Since the addition of noise is actually the addition of noise that conforms to the corresponding distribution of the Laplace mechanism and the Gaussian mechanism, in the fusion mechanism phase, it is only necessary to follow the Laplace distribution curve and the Gaussian distribution curve. That is: the Laplace mechanism and the Gaussian mechanism are fused to obtain a fusion protection mechanism, including:

[0069] According to the Laplace mechanism and the Gaussian mechanism, a Laplace distribution curve and a Gaussian distribution curve are obtained respectively;

[0070] The Laplace distribution curve and the Gaussian distribution curve are fused to obtain a fusion protection mechanism.

[0071] In one embodiment, a method for obtaining a fusion protection mechanism by fusing distribution curves is provided. The Laplace distribution curve and the Gaussian distribution curve can represent the Laplace mechanism and the Gaussian distribution mechanism respectively. Then, it is considered to fuse the distribution curves to obtain a new curve to represent the fusion protection mechanism. Specifically, the Laplace distribution curve and the Gaussian distribution curve are fused to obtain the fusion protection mechanism, including:

[0072] In the same coordinate system, respectively obtain the first ordinate value on the Laplace distribution curve and the second ordinate value on the Gaussian distribution curve corresponding to each abscissa value;

[0073] Determine a target ordinate value according to an average value of the first ordinate value and the second ordinate value;

[0074] According to each abscissa value and the target ordinate value, a fusion curve is determined to obtain a fusion protection mechanism.

[0075] In the specific implementation process, the Laplace distribution curve is as shown in the attached figure. Figure 3 As shown in the Lap L ace in the figure, it is a symmetrical distribution curve, similar to the bell curve, but with a longer tail. The Gaussian distribution curve is shown in the figure below. Figure 3 As shown in Gauss i an, it is also a symmetrical distribution curve, a typical bell-shaped curve, also known as normal distribution. Considering that the Laplace curve is also close to the standard normal distribution as a whole, the two have a basis for fusion. Putting the two curves in the same coordinate system, it is necessary to ensure that the symmetry axes of the two coincide, so as to ensure that the fusion curve does not change the most basic quasi-normal distribution. The specific curve distribution form of the Gaussian mechanism and the Laplace mechanism can be obtained according to a single mechanism, that is, the distribution curve that needs to be satisfied by adding noise to the processed data using the existing Gaussian mechanism or Laplace mechanism.

[0076] In this way, the same horizontal coordinate value will correspond to a point on the two curves respectively, and two vertical coordinate values ​​will be obtained. The two coordinate values ​​are averaged to obtain the target vertical coordinate value. In this way, a horizontal coordinate corresponds to a target vertical coordinate value, and a new point is obtained. Then all the points can be obtained in a one-to-one correspondence manner, thereby obtaining a new curve composed of these points, which is the fusion curve, as shown in the attached figure. Figure 3As shown in Fusion, the fusion protection mechanism is obtained by using the fusion curve. In the noise adding stage, a random noise generation algorithm is used to ensure that the added noise meets the distribution represented by the fusion curve. That is, according to the noise scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data to achieve data protection, including:

[0077] According to the noise scale and based on the fusion protection mechanism, noise with distribution that conforms to the fusion curve is added to the original data to obtain the target data to achieve data protection.

[0078] In one embodiment, in order to further improve the effect of privacy protection, on the basis of the above-mentioned fusion protection mechanism, the noise scale is adjusted to make it more in line with the privacy protection requirements when the fusion protection mechanism is currently adopted. Specifically, after determining the noise scale added to the data according to the privacy budget, the method further includes:

[0079] Obtaining a first similarity between the Laplace distribution curve and the fusion curve, and a second similarity between the Gaussian distribution curve and the fusion curve;

[0080] The difference between the first similarity and the second similarity is used as an adjustment coefficient, and the noise scale is adjusted according to the adjustment coefficient to obtain a noise target scale.

[0081] In the specific implementation process, the Laplace distribution curve and Gaussian distribution, which are the basis of fusion, can calculate a similarity relative to the fusion curve. For example, the similarity is measured by cosine similarity, Euclidean distance, and Manhattan distance. The similarity is obtained in the form of a percentage. The difference between the two similarities will get a signed percentage vertical, which is used as an adjustment coefficient to adjust the noise scale. By quantifying the noise scale, the current noise scale is regarded as a unit, the difference in similarity is summed with 100% and then multiplied by a unit of the noise scale, which is equivalent to adjusting the noise scale by the absolute value of the difference. The sign of the difference indicates whether it needs to be increased or decreased, which needs to be defined according to the actual situation.

[0082] Based on the above steps, according to the noise scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data to achieve data protection, including:

[0083] According to the noise target scale and based on the fusion protection mechanism, noise is added to the original data to obtain the target data to achieve data protection.

[0084] In this embodiment, the privacy budget is first determined to measure the degree of privacy protection of the data, and then the data is preprocessed to make it suitable for differential privacy processing. Subsequently, noise is added to the original data obtained by preprocessing. The addition of noise adopts a fusion protection mechanism, and the mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism. Since the two fused mechanisms are both applicable to numerical data and their distributions are symmetrical, and the Laplace is close to the Gaussian normal distribution, they can be effectively fused and retain the validity of the data. The fused mechanism has the advantages of the Laplace mechanism for processing extreme values ​​and the Gaussian mechanism for processing continuous data. While ensuring privacy protection and data availability, it meets the needs of the same type of data with different characteristics, thereby improving the effect of data protection based on differential privacy.

[0085] Refer to the attached Figure 4 Based on the same inventive concept as in the above-mentioned embodiment, the embodiment of the present application further provides a data protection device based on differential privacy, including:

[0086] Budget determination module, which is used to determine the privacy budget according to the data protection requirements;

[0087] Noise determination module, which is used to determine the scale of noise added to the data according to the privacy budget;

[0088] Preprocessing module, the preprocessing module is used to preprocess the data to obtain the original data;

[0089] The noise adding module is used to add noise to the original data according to the noise scale and based on the fusion protection mechanism to obtain the target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

[0090] Those skilled in the art should understand that the division of the modules in the embodiment is merely a division of logical functions, which can be fully or partially integrated into one or more actual carriers in actual applications, and these modules can be implemented entirely in the form of software called by a processing unit, or entirely in the form of hardware, or in the form of a combination of software and hardware. It should be noted that each module in the data protection device based on differential privacy in this embodiment corresponds one-to-one to each step in the data protection method based on differential privacy in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned data protection method based on differential privacy, which will not be repeated here.

[0091] Based on the same inventive concept as in the aforementioned embodiments, an embodiment of the present application further provides a computer-readable storage medium storing a computer program. When the computer program is loaded and executed by a processor, a data protection method based on differential privacy as provided in an embodiment of the present application is implemented.

[0092] Based on the same inventive concept as in the above-mentioned embodiment, an embodiment of the present application further provides an electronic device, including a processor and a memory, wherein:

[0093] The memory is used to store computer programs;

[0094] The processor is used to load and execute a computer program so that the electronic device executes a data protection method based on differential privacy as provided in an embodiment of the present application.

[0095] In some embodiments, the computer readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or various devices including one or any combination of the above memories. The computer may be various computing devices including intelligent terminals and servers.

[0096] In some embodiments, executable instructions may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0097] As an example, executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, such as in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files storing one or more modules, subroutines, or code portions).

[0098] By way of example, executable instructions may be deployed to be executed on one computing device, or on multiple computing devices located at one site, or on multiple computing devices distributed across multiple sites and interconnected by a communication network.

[0099] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0100] 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] 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 a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an 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 a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a multimedia terminal device (which can be a mobile phone, a computer, a television receiver, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0102] In summary, the present application provides a data protection method, device, medium and equipment based on differential privacy, which method includes: determining a privacy budget based on the protection requirements of the data; determining the scale of noise added to the data based on the privacy budget; preprocessing the data to obtain the original data; adding noise to the original data based on the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism. This application first determines the privacy budget and measures the degree of privacy protection for the data. It then preprocesses the data to make it suitable for differential privacy processing. Noise is then added to the original data obtained through preprocessing. The noise is added using a fusion protection mechanism. This mechanism is based on the fusion of the Laplace mechanism and the Gaussian mechanism. Since the two fused mechanisms are both applicable to numerical data and their distributions are symmetrical, and the Laplace is close to the Gaussian normal distribution, they can be effectively fused and retain the validity of the data. The fused mechanism has the advantages of the Laplace mechanism for processing extreme values ​​and the Gaussian mechanism for processing continuous data. While ensuring privacy protection and data availability, it meets the needs of the same type of data with different characteristics, thereby improving the effect of data protection based on differential privacy.

[0103] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A data protection method based on differential privacy, characterized in that: The following steps are involved: Determine the privacy budget based on the need to protect data; Determining a noise scale to be added to the data according to the privacy budget; Preprocessing the data to obtain original data; According to the noise scale and based on a fusion protection mechanism, noise is added to the original data to obtain target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of the Laplace mechanism and the Gaussian mechanism.

2. The data protection method based on differential privacy according to claim 1, characterized in that: Before adding noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection, the method further includes: The Laplace mechanism and the Gaussian mechanism are integrated to obtain the integrated protection mechanism.

3. The data protection method based on differential privacy according to claim 2, characterized in that: The fusing the Laplace mechanism and the Gaussian mechanism to obtain the fusion protection mechanism includes: According to the Laplace mechanism and the Gaussian mechanism, a Laplace distribution curve and a Gaussian distribution curve are obtained respectively; The Laplace distribution curve and the Gaussian distribution curve are fused to obtain the fusion protection mechanism.

4. The data protection method based on differential privacy according to claim 3, characterized in that: The fusing the Laplace distribution curve and the Gaussian distribution curve to obtain the fusion protection mechanism includes: In the same coordinate system, respectively obtain a first ordinate value on the Laplace distribution curve and a second ordinate value on the Gaussian distribution curve corresponding to each abscissa value; Determine a target ordinate value according to an average value of the first ordinate value and the second ordinate value; According to the abscissa values ​​and the target ordinate value, a fusion curve is determined to obtain the fusion protection mechanism.

5. The data protection method based on differential privacy according to claim 4, characterized in that: The adding noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection includes: According to the noise scale and based on a fusion protection mechanism, noise with a distribution that conforms to the fusion curve is added to the original data to obtain target data to achieve data protection.

6. The data protection method based on differential privacy according to claim 4, characterized in that: After determining the scale of noise added to the data according to the privacy budget, the method further includes: Obtaining a first similarity between the Laplace distribution curve and the fusion curve, and a second similarity between the Gaussian distribution curve and the fusion curve; Using the difference between the first similarity and the second similarity as an adjustment coefficient, and adjusting the noise scale according to the adjustment coefficient to obtain a noise target scale; The adding noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection includes: According to the noise target scale and based on a fusion protection mechanism, noise is added to the original data to obtain target data to achieve data protection.

7. The data protection method based on differential privacy according to claim 1, characterized in that: The preprocessing of the data to obtain original data includes: The data is cleaned to obtain original data.

8. A data protection device based on differential privacy, characterized in that: include: A budget determination module, the budget determination module is used to determine a privacy budget according to data protection requirements; A noise determination module, the noise determination module is used to determine the scale of noise added to the data according to the privacy budget; A preprocessing module, wherein the preprocessing module is used to preprocess the data to obtain original data; A noise adding module is used to add noise to the original data according to the noise scale and based on a fusion protection mechanism to obtain target data to achieve data protection; wherein the fusion protection mechanism is obtained based on the fusion of Laplace mechanism and Gaussian mechanism.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is loaded and executed by the processor, the data protection method based on differential privacy as described in any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: comprising a processor and a memory, wherein: The memory is used to store computer programs; The processor is used to load and execute the computer program so that the electronic device performs the data protection method based on differential privacy as described in any one of claims 1-7.

Images