Data desensitization processing method and device, equipment, storage medium and product

Abstract

The invention belongs to the technical field of data processing, and discloses a data desensitization processing method and device, equipment, a storage medium and a product. According to the application, the sensitive type corresponding to the sensitive service data is determined according to the sensitive service data and the associated information, then the data desensitization processing mode is determined according to the sensitive type, and then the sensitive service data is subjected to data desensitization processing according to the data desensitization processing mode and the desensitization intensity factor to obtain the desensitized service data. The sensitive type corresponding to the sensitive service data is determined firstly, then the data desensitization processing mode is determined according to the sensitive type, the sensitive service data of different sensitive types can be subjected to data desensitization processing, the desensitization intensity factor is combined, and compared with existing desensitization according to the highest privacy requirement, the data desensitization processing method has the advantages that the data desensitization processing efficiency is improved. According to the method, the desensitization intensity can be dynamically controlled on demand according to a specific scene, and then data desensitization processing is effectively carried out on the sensitive service data.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data desensitization processing method, device, equipment, storage medium and product. Background Art

[0002] At present, when the data to be desensitized has diverse desensitization requirements, it is usually difficult to take into account the desensitization effect of all scenarios using a simple desensitization algorithm, while the use of a complex algorithm model requires a large training sample size to obtain a good desensitization effect. In the operator field, the users of data after desensitization include business analysts, account managers and other roles. Different roles have different requirements for data desensitization. At present, desensitization is usually not accurately differentiated, but desensitized according to the highest privacy requirements, resulting in the loss of many analyzable attributes of the desensitized data. Therefore, how to dynamically control the intensity of data desensitization and effectively perform data desensitization processing has become an urgent problem to be solved. Summary of the invention

[0003] The main purpose of this application is to provide a data desensitization processing method, device, equipment, storage medium and product, aiming to solve the technical problem of how to dynamically control the intensity of data desensitization and effectively perform data desensitization processing.

[0004] To achieve the above purpose, the present application provides a data desensitization processing method, which comprises the following steps:

[0005] Determine the sensitivity type corresponding to the sensitive business data according to the sensitive business data and the associated information;

[0006] Determine the data desensitization processing method according to the sensitive type;

[0007] The sensitive business data is desensitized according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

[0008] Optionally, the step of determining the sensitivity type corresponding to the sensitive business data according to the sensitive business data and the associated information specifically includes:

[0009] Obtain sensitive business data and corresponding related information;

[0010] Determine business data characteristics corresponding to the sensitive business data and associated information characteristics corresponding to the associated information;

[0011] Performing a full connection operation on the business data feature and the associated information feature to obtain a target data feature;

[0012] Determine the sensitivity type corresponding to the sensitive business data according to the target data characteristics.

[0013] Optionally, the step of determining the business data characteristics corresponding to the sensitive business data specifically includes:

[0014] Performing word vector conversion on the sensitive business data to obtain a converted word vector set;

[0015] Input the word vector set into the convolution layer to obtain the feature set corresponding to each convolution kernel;

[0016] Input the feature set into the maximum pooling layer to obtain the pooling results corresponding to each convolution kernel, and construct a pooling feature set according to the pooling results;

[0017] The pooled feature set is input into the fully connected layer to obtain the business data features corresponding to the sensitive business data.

[0018] Optionally, the step of determining a data desensitization processing method according to the sensitive type specifically includes:

[0019] When the sensitive type is a sensitive numerical type, determining that the data desensitization processing method is a Laplace algorithm;

[0020] When the sensitive type is a sensitive name type, the data desensitization processing method is determined to be a natural language processing algorithm.

[0021] Optionally, the step of performing data desensitization processing on the sensitive business data according to the data desensitization processing mode and the desensitization intensity factor to obtain the desensitized business data specifically includes:

[0022] When the data desensitization processing method is the Laplace algorithm, the Laplace function value is determined according to the Laplace noise distribution width;

[0023] The sensitive business data is desensitized according to the Laplace function value and the desensitization intensity factor to obtain desensitized business data.

[0024] Optionally, the step of performing data desensitization processing on the sensitive business data according to the data desensitization processing mode and the desensitization intensity factor specifically includes:

[0025] When the data desensitization processing method is a natural language processing algorithm, the sensitive business data is input into an embedding layer to obtain a converted vector set;

[0026] Inputting the converted vector set into a position encoding layer to obtain an initial position vector, and determining a target position vector according to the initial position vector and a desensitization strength factor;

[0027] Input the target position vector into the encoder layer to obtain an encoding result;

[0028] Input the encoding result to the decoder layer to obtain a decoding result;

[0029] The decoding result is input into the linear layer to obtain the desensitized business data.

[0030] In addition, to achieve the above purpose, the present application also provides a data desensitization processing device, the data desensitization processing device comprising:

[0031] A sensitive type determination module, used to determine the sensitive type corresponding to the sensitive business data according to the sensitive business data and associated information;

[0032] A desensitization processing method determination module is used to determine the data desensitization processing method according to the sensitive type;

[0033] The data desensitization processing module is used to perform data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

[0034] In addition, to achieve the above-mentioned purpose, the present application also proposes a data desensitization processing device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the data desensitization processing method described above.

[0035] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the data desensitization processing method described above are implemented.

[0036] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the data desensitization processing method described above.

[0037] This application determines the sensitive type corresponding to the sensitive business data based on the sensitive business data and related information, and then determines the data desensitization processing method based on the sensitive type, and then performs data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data. This application first determines the sensitive type corresponding to the sensitive business data, and then determines the data desensitization processing method based on the sensitive type. It can perform data desensitization processing on sensitive business data of different sensitive types separately, and combined with the desensitization intensity factor, compared to the existing desensitization according to the highest privacy requirements, the above method of this application can realize dynamic control of the desensitization intensity on demand according to the specific scenario, and thus effectively perform data desensitization processing on sensitive business data. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0040] Figure 1 This is a flowchart of the first embodiment of the data desensitization processing method of this application;

[0041] Figure 2 This is an example of the overall data desensitization processing flow of an embodiment of the data desensitization processing method of this application;

[0042] Figure 3 This is a flow chart of the second embodiment of the data desensitization processing method of this application;

[0043] Figure 4 This is the discriminator model structure of an embodiment of the data desensitization processing method of this application;

[0044] Figure 5 This is the generator model structure of an embodiment of the data desensitization processing method of this application;

[0045] Figure 6 This is a structural block diagram of the first embodiment of the data desensitization processing device of this application;

[0046] Figure 7 It is a structural diagram of a data desensitization processing device in a hardware operating environment involved in an embodiment of the present application.

[0047] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0049] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0050] The main solution of the embodiment of the present application is: determine the sensitive type corresponding to the sensitive business data according to the sensitive business data and related information; determine the data desensitization processing method according to the sensitive type; desensitize the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data.

[0051] At present, when the data to be desensitized has diverse desensitization requirements, it is usually difficult to take into account the desensitization effect of all scenarios using a simple desensitization algorithm, while a complex algorithm model requires a large training sample size to achieve a good desensitization effect. In the operator field, the users of data after desensitization include business analysts, account managers and other roles. Different roles have different requirements for data desensitization. At present, desensitization is usually not accurately differentiated, but desensitized according to the highest privacy requirements, resulting in the loss of many analyzable attributes of the desensitized data.

[0052] This application determines the sensitive type corresponding to the sensitive business data based on the sensitive business data and related information, and then determines the data desensitization processing method based on the sensitive type, and then performs data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data. This application first determines the sensitive type corresponding to the sensitive business data, and then determines the data desensitization processing method based on the sensitive type. It can perform data desensitization processing on sensitive business data of different sensitive types separately, and combined with the desensitization intensity factor, compared to the existing desensitization according to the highest privacy requirements, the above method of this application can realize dynamic control of the desensitization intensity on demand according to the specific scenario, and thus effectively perform data desensitization processing on sensitive business data.

[0053] It should be noted that the execution subject of this application can be a computing service device with data processing, network communication and program running functions, such as a computer, etc., or an electronic device capable of realizing the above functions, a data desensitization processing device, etc. The following takes the data desensitization processing device as an example to illustrate this embodiment and the following embodiments.

[0054] Based on this, the present application embodiment provides a data desensitization processing method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the data desensitization processing method of this application.

[0055] In this embodiment, the data desensitization processing method includes the following steps:

[0056] Step S10: Determine the sensitivity type corresponding to the sensitive business data according to the sensitive business data and the associated information.

[0057] It is understandable that the sensitive business data in this embodiment can be user data stored in the background of the business support system. At present, there are generally several sensitive encryption scenarios such as user sensitive value confusion, such as consumption amount, balance, etc., and subject identity information hiding, such as company name, address, name, etc. The associated information can be the type of sensitive business data, such as consumption amount, balance, etc.

[0058] In a specific implementation, the sensitive type corresponding to the sensitive business data may be determined based on the sensitive business data and associated information. The sensitive type may include a sensitive value and a sensitive name.

[0059] Step S20: Determine a data desensitization processing method according to the sensitive type.

[0060] It should be understood that the data desensitization processing method can be determined according to the sensitivity type, that is, sensitive values ​​and sensitive names correspond to different data desensitization processing methods.

[0061] Step S30: performing data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

[0062] It is understandable that the desensitization intensity factor refers to a factor that can control the desensitization intensity. Specifically, the desensitization intensity factor can be set to 0 to indicate partial desensitization, and the desensitization intensity factor can be set to 1 to indicate complete desensitization. In this embodiment, sensitive business data can be desensitized according to the data desensitization processing method combined with the desensitization intensity factor to obtain desensitized business data.

[0063] In the specific implementation, refer to Figure 2 , Figure 2 This is an example of the overall data desensitization process of an embodiment of the data desensitization method of this application. Figure 3 As shown, sensitive business data may include [charge amount, 2300.00, 1115.00, 965.00], [customer name, Jiangsu AAA Industrial Co., Ltd., Beijing BBB Technology Co., Ltd., Zhang Zhengfeng], and the sensitive type may be determined by the discriminator, including sensitive values ​​and sensitive names, and then the sensitive data and sensitive names may be partially or completely desensitized by the generator to obtain desensitized business data, which may include the charge amount obtained by partial desensitization [2350.60, 1093.75, 931.00], obtained by complete desensitization [5111.25, 2376.1, 475.41], the customer name obtained by partial desensitization [Jiangsu **** Industrial Co., Ltd., Beijing **** Technology Co., Ltd., Zhang *feng], and the customer name obtained by complete desensitization [**** Co., Ltd., **** Co., Ltd., Zhang *].

[0064] This embodiment determines the sensitive type corresponding to the sensitive business data based on the sensitive business data and the associated information, then determines the data desensitization processing method based on the sensitive type, and then performs data desensitization processing on the sensitive business data based on the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data. This embodiment first determines the sensitive type corresponding to the sensitive business data, and then determines the data desensitization processing method based on the sensitive type. It can perform data desensitization processing on sensitive business data of different sensitive types respectively, and combined with the desensitization intensity factor, compared with the existing desensitization according to the highest privacy requirements, the above method of this embodiment can realize dynamic control of the desensitization intensity on demand according to the specific scenario, thereby effectively performing data desensitization processing on sensitive business data.

[0065] refer to Figure 3 , Figure 3 This is a flow chart of the second embodiment of the data desensitization processing method of this application.

[0066] Based on the above first embodiment, in this embodiment, step S10 includes:

[0067] Step S101: Acquire sensitive business data and corresponding associated information.

[0068] It should be understood that reference Figure 4 , Figure 4 This is the discriminator model structure of an embodiment of the data desensitization processing method of this application. Figure 4 The discriminator shown can determine the sensitive type corresponding to the sensitive business data, and specifically can obtain the sensitive business data Input Data and the corresponding related information Input Related Info. For example, the Input Data is [2300.00, 1115.00, 965.00], and the Input Related Info is the consumption amount.

[0069] Step S102: Determine the business data characteristics corresponding to the sensitive business data and the associated information characteristics corresponding to the associated information.

[0070] Furthermore, in order to accurately obtain the business data features corresponding to the sensitive business data, in this embodiment, the step of determining the business data features corresponding to the sensitive business data specifically includes: performing word vector conversion on the sensitive business data to obtain a converted word vector set; inputting the word vector set into the convolution layer to obtain a feature set corresponding to each convolution kernel; inputting the feature set into the maximum pooling layer to obtain the pooling results corresponding to each convolution kernel, and constructing a pooling feature set based on the pooling results; inputting the pooling feature set into the fully connected layer to obtain the business data features corresponding to the sensitive business data.

[0071] Understandably, in Figure 4 In the discriminator shown, Word Embedding-1: The sensitive business data is converted into word vectors through the publicly known word2Vec method, and the word vector dimension is d. Let X i ∈R d is the word vector of the i-th word in the sentence. The input data length is n, and the converted word vector set can be expressed as:

[0072] Convolutional Layer-1: Convolutional layer, which uses h convolution kernels of different sizes, each of which is responsible for extracting a feature of a different size. Suppose the sizes of the convolution kernels are h1, h2, …, h h , the width of the convolution kernel is the same as the dimension of the word vector, that is, d. Let W i is the weight matrix of the i-th convolution kernel, with dimension h i ×d;b i is the skew value of the i-th convolution kernel, then for the i-th convolution kernel, the corresponding feature set is: c i = relu(W i *X+b i ).

[0073] Max Pooling-1: The maximum pooling layer extracts the most significant features from the feature set output by the convolution layer. For the i-th convolution kernel output, the pooling result is: p i =max(c i ), the pooled feature set is: P = (p1, p2, ..., p h ).

[0074] Fully Connected Layer-1: Fully connected layer, further extracts feature information from the pooled feature set output by the pooling layer. Suppose the weight of this layer is W1, the skew value is b1, the output is O1, and the business data feature corresponding to the sensitive business data is: O1=relu(W1·P+b1).

[0075] It should be understood that the steps of determining the associated information features corresponding to the associated information can refer to the above-mentioned steps of determining the business data features. Specifically, Word Embedding-2: convert the input associated information into word vectors, and the processing method is the same as WordEmbedding-1. Convolutional Layer-2: perform convolution operation on the input associated information, and the processing method is the same as Convolutional Layer-1. Max Pooling-2: perform maximum pooling processing on the output of Convolutional Layer-2, and the processing method is the same as Max Pooling-1. Full Connected Layer-2: perform full connection operation on the output of Max Pooling-2, and the processing method is the same as Full Connected Layer-1, and its output is set to be the associated information feature O2.

[0076] Step S103: Perform a full connection operation on the business data feature and the associated information feature to obtain the target data feature.

[0077] It can be understood that Full Connected Layer-3: merges the business data feature O1 output by Full Connected Layer-1 and the associated information feature output by Full Connected Layer-2, and outputs the final result, i.e., the target data feature, through a full connection operation. Assuming the weight of this layer is W0, the skew value is b0, and the output is O0, the target data feature is: O0 = softmax(W0·concat(O1,P2)+b0).

[0078] Step S104: Determine the sensitivity type corresponding to the sensitive business data according to the target data characteristics.

[0079] In the specific implementation, Output Class: obtains the sensitive type corresponding to the target data feature of the final output from the upper layer. The sensitive type may include: 0-non-sensitive, 1-sensitive value, 2-sensitive name.

[0080] This embodiment obtains sensitive business data and corresponding associated information, then determines the business data features corresponding to the sensitive business data and the associated information features corresponding to the associated information, then performs a full-connect operation on the business data features and the associated information features to obtain target data features, and then determines the sensitive type corresponding to the sensitive business data based on the target data features. This embodiment determines the business data features corresponding to the sensitive business data and the associated information features corresponding to the associated information, then performs a full-connect operation on the business data features and the associated information features, and determines the sensitive type corresponding to the sensitive business data based on the target data features, so that the sensitive type corresponding to the sensitive business data can be accurately obtained.

[0081] Based on the above embodiments, in this embodiment, the step S20 includes: when the sensitive type is a sensitive numerical type, determining that the data desensitization processing method is a Laplace algorithm; when the sensitive type is a sensitive name type, determining that the data desensitization processing method is a natural language processing algorithm.

[0082] Understandably, referring to Figure 5 , Figure 5 This is the generator model structure of an embodiment of the data desensitization processing method of this application. Figure 5 The generator shown can perform data desensitization processing on sensitive business data of sensitive numeric types and sensitive name types respectively.

[0083] It should be understood that for non-sensitive types, they can be output as is, that is, sensitive business data will not be desensitized. For sensitive numerical types, the overall probability and statistical related characteristics need to be retained for the needs of business analysts, but not for account managers. Therefore, the Laplace algorithm is used for desensitization to retain the analyzability of the data. At the same time, when the desensitization intensity factor value is completely desensitized, a more disordered disturbance coefficient is introduced to destroy the analyzability of the data. For sensitive name types, a method based on Transformer, that is, a natural language processing algorithm, and the introduction of a desensitization intensity factor is used to perform refined content desensitization and remove individual characteristics as needed.

[0084] In the specific implementation, Figure 5 The input of the generator algorithm structure in may include sensitive business data Input Data, sensitive type Sensitive Type, and desensitization intensity factor Sensitive Intensity. Branches are called according to the sensitive type corresponding to the sensitive business data. When the type is a sensitive value, the subsequent algorithm is called Figure 5 The Laplace algorithm on the left generates desensitized business data; when the type is a sensitive name, call Figure 5The Transformer-based algorithm on the right side of the middle generates desensitized business data. When called, the desensitization intensity factor is passed down. When the factor value is 0, it indicates partial desensitization, and when the value is 1, it indicates complete desensitization. For example, the input data of the generator is [1,1,[2300.00,1115.00,965.00]].

[0085] Furthermore, in order to effectively perform data desensitization on sensitive business data of sensitive numerical types, in this embodiment, the step S30 includes: when the data desensitization processing method is the Laplace algorithm, determining the Laplace function value according to the Laplace noise distribution width; performing data desensitization on the sensitive business data according to the Laplace function value and the desensitization intensity factor to obtain desensitized business data.

[0086] It is understandable that Laplace Noise: Laplace noise is injected into sensitive business data, adding data desensitization characteristics. The processed data can roughly retain the overall trend, distribution shape, frequency statistics and other characteristics of the original data while adding obfuscation, thereby achieving the overall analysis value of the data without exposing individual information. The Laplace function value is:

[0087]

[0088] Among them, α is a random variable and b is a scale parameter that controls the distribution width of the noise.

[0089] On this basis, when the desensitization intensity is completely desensitized, more disordered noise needs to be introduced. Here, random numbers are directly used. Let the desensitization intensity factor be m. When it is completely desensitized, m can be 1. The sensitive business data is x, and the output is y. The business data after desensitization is:

[0090]

[0091] Furthermore, in order to effectively perform data desensitization processing on sensitive business data of sensitive name type, in this embodiment, the step S30 also includes: when the data desensitization processing method is a natural language processing algorithm, the sensitive business data is input into the embedding layer to obtain a converted vector set; the converted vector set is input into the position encoding layer to obtain an initial position vector, and a target position vector is determined according to the initial position vector and the desensitization intensity factor; the target position vector is input into the encoder layer to obtain an encoding result; the encoding result is input into the decoder layer to obtain a decoding result; the decoding result is input into the linear layer to obtain the desensitized business data.

[0092] It can be understood that the Embedding Layer converts each character of the input sensitive business data into a vector identifier. For each input character, let the converted vector be represented by e i , then the output of the embedding layer, that is, the converted vector set is: E X =(e1,e2,…,e n ).

[0093] Positional Encoding: Positional encoding layer adds position information to the vector so that the contextual position information can be perceived during subsequent feature extraction. The initial position vector of the i-th position is:

[0094]

[0095] Where d is the dimension of the embedding vector, j is the index of the dimension, and i is the position.

[0096] In this embodiment, a desensitization intensity factor can be added while adding the position vector. The target position vector is: E′ X =concat(m,E X +PE).

[0097] Encoder Layer: Encoder layer, which can use the multi-head self-attention mechanism of the Transformer algorithm to extract feature information. Q , W K ,W V Represent the query, key, and value of the self-attention mechanism respectively, and d k is the dimension of the vector, then the self-attention calculation formula is:

[0098]

[0099] Assume the number of heads of the multi-head attention mechanism is h, and the input of the first head is head i =Attention(E′ X ). W0 is the weight of multi-head attention, and the output of multi-head attention is: MultiHead(E′ X )=Concat(head1,head2,…,head h )·W0. Then apply residual connection and layer normalization, and the encoding result is: EncoderLayer(E′ X ) = LayerNorm(E′) X +MultiHead(E′ X )).

[0100] Decoder Layer: The calculation process of the decoder layer is similar to that of the encoder layer, except that the input of this layer is E′ Y , that is, the output of the position vector is added, and similarly we can get:

[0101]

[0102] head i =Attention(E′ Y )

[0103] MultiHead(E′ Y )=Concat(head1,head2,…,head h )·W0

[0104] In addition to calculating the above multi-head attention output, the decoder layer also needs to pay attention to the input sequence

[0105] LayerNorm(E′ Y +MultiHead(E′ Y )+Attention(E′ Y ,E′ X )).

[0106] Linear Layer: Linear layer maps the output of the decoder to the probability distribution of the target vocabulary. Let the learning weight of this layer be W out , the final output, that is, the desensitized business data is: P(Y) = softmax(W out DecoderLayer(E′ Y )).

[0107] This embodiment implements data desensitization processing through the above-mentioned algorithm model, which realizes the process from sensitive business data identification to accurate data desensitization processing, while retaining the analytical characteristics of the desensitized data, which can be used in scenarios such as business analysis and decision-making, and has high practical value.

[0108] In addition, the desensitized business data can continue to be encrypted. Specifically, because similar application scenarios exist in multiple business systems, the sample data of multiple systems can be combined to train the model to improve the robustness of the model. In order to protect the privacy of data in each system, federated learning is introduced to perform cross-system sample data fusion training. For business scenarios, the federated averaging algorithm of horizontal federated learning is used for learning.

[0109] When using this training method, the server first sends the model, initial parameters and fixed weights to each business system. Each business system trains 3 rounds based on local samples and multiplies the trained W by the weight Upload to the server, the server performs weighted summation of W uploaded by all systems, the formula is as follows:

[0110]

[0111] Where n is the number of business systems, The fixed setting is 1 / n. Different scenarios can be set to take weights based on the proportion of the total number of samples according to the actual verification effect.

[0112] The server sends the calculated parameter W to each business system to update the local parameters, and then uploads the parameters again after 3 rounds of local training, and iterates continuously until the model converges. This method will upload the weighted parameters of local training to the server. In theory, there is a possibility that the server can infer local data based on the parameters, which poses a threat to data privacy, so an encryption algorithm is introduced in the transmission of parameters for encryption. Because the processing of multiplying the parameters by the weights is completed locally on the client, only addition calculations are required on the server, so the semi-homomorphic encryption algorithm Paillier is used to encrypt the data values ​​uploaded locally by the client to ensure the security of data privacy. The specific processing method is as follows: 1. All client business systems upload their own encrypted public keys to the server. 2. The server sends all public keys to all clients. 3. When the client uploads parameters, it uses all public keys to encrypt a version of the data and then uploads it to the server. 4. The server groups by public key, accumulates and sums all client encrypted parameters corresponding to each public key, and sends them to the client corresponding to this public key. The client uses the private key to decrypt and update the local parameters.

[0113] Reference Figure 6 , Figure 6 This is a structural block diagram of the first embodiment of the data desensitization processing device of this application.

[0114] like Figure 6 As shown, the data desensitization processing device proposed in the embodiment of the present application includes:

[0115] A sensitive type determination module 10, configured to determine the sensitive type corresponding to the sensitive business data according to the sensitive business data and associated information;

[0116] A desensitization processing method determination module 20, used to determine a data desensitization processing method according to the sensitive type;

[0117] The data desensitization processing module 30 is used to perform data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

[0118] This embodiment determines the sensitive type corresponding to the sensitive business data based on the sensitive business data and the associated information, then determines the data desensitization processing method based on the sensitive type, and then performs data desensitization processing on the sensitive business data based on the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data. This embodiment first determines the sensitive type corresponding to the sensitive business data, and then determines the data desensitization processing method based on the sensitive type. It can perform data desensitization processing on sensitive business data of different sensitive types respectively, and combined with the desensitization intensity factor, compared with the existing desensitization according to the highest privacy requirements, the above method of this embodiment can realize dynamic control of the desensitization intensity on demand according to the specific scenario, thereby effectively performing data desensitization processing on sensitive business data.

[0119] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present application. In practical applications, technicians in this field can select part or all of it according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0120] In addition, for technical details not described in detail in this embodiment, please refer to the data desensitization processing method provided in any embodiment of the present application, which will not be repeated here.

[0121] Based on the above-mentioned first embodiment of the data desensitization processing device of the present application, a second embodiment of the data desensitization processing device of the present application is proposed.

[0122] In this embodiment, the sensitive type determination module 10 is also used to obtain sensitive business data and corresponding associated information; determine the business data characteristics corresponding to the sensitive business data and the associated information characteristics corresponding to the associated information; perform a full connection operation on the business data characteristics and the associated information characteristics to obtain target data characteristics; and determine the sensitive type corresponding to the sensitive business data based on the target data characteristics.

[0123] Furthermore, the sensitive type determination module 10 is also used to perform word vector conversion on the sensitive business data to obtain a converted word vector set; input the word vector set to the convolution layer to obtain a feature set corresponding to each convolution kernel; input the feature set to the maximum pooling layer to obtain a pooling result corresponding to each convolution kernel, and construct a pooling feature set based on the pooling result; input the pooling feature set to the fully connected layer to obtain business data features corresponding to the sensitive business data.

[0124] Furthermore, the desensitization processing method determination module 20 is also used to determine that the data desensitization processing method is a Laplace algorithm when the sensitive type is a sensitive numerical type; and to determine that the data desensitization processing method is a natural language processing algorithm when the sensitive type is a sensitive name type.

[0125] Furthermore, the data desensitization processing module 30 is also used to determine the Laplace function value according to the Laplace noise distribution width when the data desensitization processing method is the Laplace algorithm; perform data desensitization processing on the sensitive business data according to the Laplace function value and the desensitization intensity factor to obtain the desensitized business data.

[0126] Furthermore, the data desensitization processing module 30 is also used to input the sensitive business data into the embedding layer to obtain a converted vector set when the data desensitization processing method is a natural language processing algorithm; input the converted vector set into the position encoding layer to obtain an initial position vector, and determine the target position vector based on the initial position vector and the desensitization intensity factor; input the target position vector into the encoder layer to obtain an encoding result; input the encoding result into the decoder layer to obtain a decoding result; input the decoding result into the linear layer to obtain the desensitized business data.

[0127] Other embodiments or specific implementation methods of the data desensitization processing device of the present application can refer to the above-mentioned method embodiments and will not be repeated here.

[0128] The present application provides a data desensitization processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the data desensitization processing method in the above-mentioned embodiment one.

[0129] Reference below Figure 7 , which shows a schematic diagram of the structure of a data desensitization processing device suitable for implementing an embodiment of the present application. The data desensitization processing device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The data desensitization processing device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0130] like Figure 7As shown, the data desensitization processing device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the data desensitization processing device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the data desensitization processing device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a data desensitization processing device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided alternatively.

[0131] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0132] The data desensitization processing device provided by the present application adopts the data desensitization processing method in the above embodiment, which can solve the technical problem of how to dynamically control the data desensitization intensity and effectively perform data desensitization processing. Compared with the prior art, the beneficial effects of the data desensitization processing device provided by the present application are the same as the beneficial effects of the data desensitization processing method provided by the above embodiment, and the other technical features in the data desensitization processing device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0133] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0134] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0135] The present application provides a computer-readable storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the data desensitization processing method in the above-mentioned embodiment.

[0136] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0137] The above-mentioned computer-readable storage medium may be included in the data desensitizing processing device; or it may exist independently without being assembled into the data desensitizing processing device.

[0138] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the data desensitization processing device, the data desensitization processing device: determines the sensitive type corresponding to the sensitive business data according to the sensitive business data and related information; determines the data desensitization processing method according to the sensitive type; performs data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain the desensitized business data.

[0139] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0140] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0141] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0142] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned data desensitization processing method, and can solve the technical problem of how to dynamically control the intensity of data desensitization and effectively perform data desensitization processing. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the data desensitization processing method provided in the above-mentioned embodiment, and will not be repeated here.

[0143] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned data desensitization processing method when executed by a processor.

[0144] The computer program product provided by this application can solve the technical problem of how to dynamically control the intensity of data desensitization and effectively perform data desensitization processing. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the data desensitization processing method provided by the above embodiment, which will not be repeated here.

[0145] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A data desensitization processing method, characterized in that: The data desensitization processing method comprises the following steps: Determine the sensitivity type corresponding to the sensitive business data according to the sensitive business data and the associated information; Determine the data desensitization processing method according to the sensitive type; The sensitive business data is desensitized according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

2. The data desensitization processing method according to claim 1, characterized in that: The step of determining the sensitive type corresponding to the sensitive business data according to the sensitive business data and the associated information specifically includes: Obtain sensitive business data and corresponding related information; Determine business data characteristics corresponding to the sensitive business data and associated information characteristics corresponding to the associated information; Performing a full connection operation on the business data feature and the associated information feature to obtain a target data feature; Determine the sensitivity type corresponding to the sensitive business data according to the target data characteristics.

3. The data desensitization processing method according to claim 2, characterized in that: The step of determining the business data characteristics corresponding to the sensitive business data specifically includes: Performing word vector conversion on the sensitive business data to obtain a converted word vector set; Input the word vector set into the convolution layer to obtain the feature set corresponding to each convolution kernel; Input the feature set into the maximum pooling layer to obtain the pooling results corresponding to each convolution kernel, and construct a pooling feature set according to the pooling results; The pooled feature set is input into the fully connected layer to obtain the business data features corresponding to the sensitive business data.

4. The data desensitization processing method according to claim 1, characterized in that: The step of determining the data desensitization processing method according to the sensitive type specifically includes: When the sensitive type is a sensitive numerical type, determining that the data desensitization processing method is a Laplace algorithm; When the sensitive type is a sensitive name type, the data desensitization processing method is determined to be a natural language processing algorithm.

5. The data desensitization processing method according to claim 4, characterized in that: The step of performing data desensitization processing on the sensitive business data according to the data desensitization processing mode and the desensitization intensity factor to obtain the desensitized business data specifically includes: When the data desensitization processing method is the Laplace algorithm, the Laplace function value is determined according to the Laplace noise distribution width; The sensitive business data is desensitized according to the Laplace function value and the desensitization intensity factor to obtain desensitized business data.

6. The data desensitization processing method according to claim 4, characterized in that: The step of performing data desensitization processing on the sensitive business data according to the data desensitization processing mode and the desensitization intensity factor specifically includes: When the data desensitization processing method is a natural language processing algorithm, the sensitive business data is input into an embedding layer to obtain a converted vector set; Inputting the converted vector set into a position encoding layer to obtain an initial position vector, and determining a target position vector according to the initial position vector and a desensitization strength factor; Input the target position vector into the encoder layer to obtain an encoding result; Input the encoding result to the decoder layer to obtain a decoding result; The decoding result is input into the linear layer to obtain the desensitized business data.

7. A data desensitization processing device, characterized in that: The data desensitization processing device comprises: A sensitive type determination module, used to determine the sensitive type corresponding to the sensitive business data according to the sensitive business data and associated information; A desensitization processing method determination module is used to determine the data desensitization processing method according to the sensitive type; The data desensitization processing module is used to perform data desensitization processing on the sensitive business data according to the data desensitization processing method and the desensitization intensity factor to obtain desensitized business data.

8. A data desensitization processing device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the data desensitization processing method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the data desensitization processing method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the data desensitization processing method as claimed in any one of claims 1 to 6 are implemented.

Images