Data desensitization method and related equipment

Abstract

The invention relates to a data desensitization method and related equipment. The method comprises the following steps: acquiring a data access request of a user; according to the data access request, determining original data corresponding to the data access request; obtaining the sensitivity of each field of each sensitive entity in the original data; determining the security state of the user terminal according to the equipment fingerprint; obtaining index data of the current network encryption strength, and determining the network encryption strength according to the index data; determining a scene risk value according to the user terminal security state, the network encryption strength, the request time, the terminal position information, the operation behavior mode and the data flow compliance; determining the comprehensive risk of each field based on the sensitivity of each field, the scene risk value and the user credibility; and selecting a desensitization strategy of each field based on the comprehensive risk of each field, performing desensitization processing on each field of the original data, and returning desensitized data to the user. Differentiated desensitization processing is realized, and high-value data can be effectively protected.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data desensitization method, device, electronic device and computer-readable storage medium. Background Art

[0002] Data masking is a technology that protects data privacy by hiding, replacing, encrypting, or removing sensitive information. This ensures that personal or confidential information is not disclosed during data query, export, sharing, storage, or analysis. Masked data maintains business usability while complying with privacy regulations.

[0003] Dynamic desensitization refers to the real-time desensitization of sensitive data during data query and usage, based on user identity, permissions, and access context. In other words, when a user requests data, the system instantly desensitizes the data returned to the user based on pre-set policies and rules, without altering the original data itself.

[0004] However, current dynamic desensitization does not take into account differences in usage environments, resulting in insufficient protection of high-value data. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a data desensitization method and related equipment, which realize dynamic risk assessment through collaborative analysis of multiple dimensions such as data attributes, usage environment and access subjects, and achieve differentiated desensitization processing, which can effectively protect high-value data.

[0006] In a first aspect, an embodiment of the present invention provides a data desensitization method. The method comprises:

[0007] Obtaining a user's data access request, which includes the user's identity, device fingerprint, request time, and terminal location information;

[0008] Determining, according to the data access request, the original data corresponding to the data access request;

[0009] Obtain the sensitivity of each field of each sensitive entity in the original data;

[0010] Determine the user terminal security status based on device fingerprint;

[0011] Obtain indicator data of the current network encryption strength and determine the network encryption strength based on the indicator data;

[0012] Determine the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance;

[0013] Determine the comprehensive risk of each field based on its sensitivity, scenario risk value, and user trust;

[0014] Select a desensitization strategy for each field based on the comprehensive risk of each field, desensitize each field of the original data, and return the desensitized data to the user.

[0015] In one embodiment, before obtaining the sensitivity of each field of each sensitive entity in the original data, the method further includes:

[0016] Identify sensitive entities in the original data and determine the basic sensitivity of the sensitive entities;

[0017] Build a field relationship map based on the original data, extract field relationships, build an adjacency matrix, use the improved PageRank algorithm to calculate the PageRank value of each field, and perform normalization to obtain the correlation sensitivity of each field;

[0018] The sensitivity of each field is determined based on the base sensitivity and the associated sensitivity of each field.

[0019] In one embodiment, determining the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance includes:

[0020] Determine the respective risk values of user terminal security status, network encryption strength, temporal and spatial characteristics, operational behavior patterns, and data flow compliance;

[0021] The dynamic weight is determined by the entropy weight method and the scenario risk value is calculated.

[0022] In one embodiment, before determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust, the method further includes:

[0023] Determine user base trust based on user categories;

[0024] The user operation abnormality rate is counted through a sliding time window, and the user basic trust is corrected according to the user operation abnormality rate to obtain the user trust.

[0025] In one embodiment, determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust includes:

[0026] Determine the weight parameters of each field's sensitivity, scenario risk value, and user trust;

[0027] Calculate the overall risk.

[0028] In one embodiment, desensitizing each field of the original data and returning the desensitized data to the user includes:

[0029] Parse the format of the original data and extract fields;

[0030] Split each field into an independent processing object;

[0031] Execute the corresponding desensitization strategy for each field in parallel;

[0032] Output in the structure of original data, with additional desensitized metadata; the desensitized metadata includes the time of desensitization processing and the ID of the desensitization strategy used for each field;

[0033] Generates a log containing the hash value of the original data, desensitized metadata, the comprehensive risk of each field, and the weighted parameters of each field's sensitivity, scenario risk value, and user trust.

[0034] In one embodiment, the method further comprises:

[0035] The effectiveness of the desensitization strategy is periodically evaluated, and the weight parameters of the sensitivity of each field, scenario risk value, and user trust are dynamically optimized based on the reinforcement learning model of the deep Q network.

[0036] In a second aspect, an embodiment of the present invention provides a data desensitization device. The device includes:

[0037] A request acquisition module is used to obtain a user's data access request, wherein the data access request includes user identity, device fingerprint, request time and terminal location information;

[0038] The original data determination module is used to determine the original data corresponding to the data access request according to the data access request;

[0039] A sensitivity acquisition module is used to obtain the sensitivity of each field of each sensitive entity in the original data;

[0040] A user terminal security status determination module is used to determine the user terminal security status based on the device fingerprint;

[0041] The network encryption strength determination module is used to obtain the indicator data of the current network encryption strength and determine the network encryption strength based on the indicator data;

[0042] A scenario risk value determination module is used to determine the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance;

[0043] Comprehensive risk determination module, used to determine the comprehensive risk of each field based on the sensitivity of each field, scenario risk value, and user trust;

[0044] The desensitization execution module is used to select the desensitization strategy for each field based on the comprehensive risk of each field, desensitize each field of the original data, and return the desensitized data to the user.

[0045] In one embodiment, the apparatus further comprises:

[0046] The sensitivity determination module is used to identify sensitive entities in the original data and determine the basic sensitivity of sensitive entities; construct a field relationship map based on the original data, extract field relationships, build an adjacency matrix, use the improved PageRank algorithm to calculate the PageRank value of each field, and perform normalization to obtain the associated sensitivity of each field; determine the sensitivity of each field based on the basic sensitivity and the associated sensitivity of each field.

[0047] In one embodiment, the scenario risk value determination module includes:

[0048] The first submodule is used to determine the risk values of the user terminal security status, network encryption strength, time and space characteristics, operation behavior patterns, and data flow compliance;

[0049] The first calculation submodule is used to determine the dynamic weight through the entropy weight method and calculate the scenario risk value.

[0050] In one embodiment, the apparatus further comprises:

[0051] The user trust confirmation module is used to determine the user's basic trust according to the user category; the user operation abnormality rate is counted through a sliding time window, and the user's basic trust is corrected according to the user operation abnormality rate to obtain the user's trust.

[0052] In one embodiment, the comprehensive risk determination module includes:

[0053] The weight parameter determination submodule is used to determine the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust;

[0054] The second calculation submodule is used to calculate the comprehensive risk.

[0055] In one embodiment, the desensitization execution module includes:

[0056] The field extraction submodule is used to parse the format of the original data and extract fields;

[0057] Split submodule, used to split each field into independent processing objects;

[0058] The policy execution submodule is used to execute the corresponding desensitization policy for each field in parallel;

[0059] The output submodule is used to output the original data structure and append desensitization metadata; the desensitization metadata includes the time of desensitization processing and the ID of the desensitization strategy adopted for each field;

[0060] The log generation submodule is used to generate logs containing the hash value of the original data, desensitized metadata, the comprehensive risk of each field, and the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust.

[0061] In one embodiment, the apparatus further comprises:

[0062] The evaluation and weight update module is used to periodically evaluate the effectiveness of the desensitization strategy and dynamically optimize the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust based on the reinforcement learning model of the deep Q network.

[0063] In a third aspect, the present invention provides an electronic device, comprising:

[0064] at least one processor; and

[0065] a memory communicatively connected to the at least one processor; wherein,

[0066] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the above-mentioned data desensitization method.

[0067] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the above-mentioned data desensitization method when executed by a processor.

[0068] In this embodiment, by obtaining the user's data access request, the data access request includes the user identity, device fingerprint, request time and terminal location information; according to the data access request, determining the original data corresponding to the data access request; obtaining the sensitivity of each field of each sensitive entity in the original data; determining the user terminal security status according to the device fingerprint; obtaining the indicator data of the current network encryption strength, and determining the network encryption strength according to the indicator data; determining the scenario risk value according to the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance; determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust; selecting the desensitization strategy of each field based on the comprehensive risk of each field, desensitizing each field of the original data, and returning the desensitized data to the user. Differentiated desensitization processing is achieved, and high-value data can be effectively protected. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0070] Figure 1 This is a flow chart of a data desensitization method provided in Example 1 of the present invention;

[0071] Figure 2 This is a structural diagram of a data desensitization device provided in the second embodiment of the present invention;

[0072] Figure 3 It is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0073] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0074] Figure 1 This is a flow chart of a data desensitization method provided in the first embodiment of the present invention. This embodiment can be applied to a data desensitization device, such as Figure 1 As shown, this embodiment may include the following steps:

[0075] Step 101: Obtain a user's data access request, where the data access request includes user identification, device fingerprint, request time, and terminal location information.

[0076] In this step, the data access request is used to request access to the desensitized data after the original data has been processed. The data access request includes the user identity (such as user ID), device fingerprint, request time, and terminal location information. Of course, the data access request may also include other information, which is not listed here.

[0077] Device fingerprinting refers to the unique identification technology of terminal devices (such as mobile phones, computers, IoT devices, etc.). It generates a highly likely unique "fingerprint" by collecting the hardware and software features of the device, which is used to identify and track the device.

[0078] Step 102: Determine the original data corresponding to the user request according to the data access request.

[0079] In this step, according to the user's data access request, the original data corresponding to the user's request is determined, for example, it is determined that the user requests to access the medical records of a certain patient. The original data can be stored in an original database.

[0080] Step 103: Obtain the sensitivity of each field of each sensitive entity in the original data.

[0081] After determining the original data corresponding to the user request, field metadata of the original data is obtained, where the field metadata includes the sensitivity of each field of each sensitive entity.

[0082] In one embodiment, before step 103, the method further includes:

[0083] Step 103a: Identify the sensitive entities in the original data and determine the basic sensitivity of the sensitive entities.

[0084] For example, a Bidirectional Long Short-Term Memory-Conditional Random Field (BiLSTM-CRF) model can be used to identify sensitive entities in the original data and output the basic sensitivity of the sensitive entities.

[0085] In the BiLSTM-CRF model, the number of LSTM hidden layer units can be set to 128, the learning rate can be set to 0.001, and the dropout rate can be set to 0.3.

[0086] The CRF layer can use the negative log-likelihood loss function:

[0087] L=-∑logP(y|x)+λ‖θ‖ 2

[0088] Where x is the input sequence (such as the word sequence in the text), which is the input variable of the model; y is the label sequence (such as the label sequence in named entity recognition), which is the true label in the training data; P(y|x) is the conditional probability, which represents the probability of the model predicting the label sequence (y) when the input sequence (x) is given, and is calculated by the BiLSTM-CRF model; θ is the model parameter (including the weights of the BiLSTM and CRF layers), which is automatically learned during the training process and does not need to be manually set; λ is the regularization coefficient, which is used to balance the negative log-likelihood term and the regularization term in the loss function to prevent model overfitting, and is usually set to 0.001 to 0.01.

[0089] Step 103b: construct a field relationship graph based on the original data, extract the field relationship, construct an adjacency matrix, use the improved PageRank algorithm to calculate the PageRank value of each field, and perform normalization to obtain the correlation sensitivity of each field. The PageRank calculation formula can be:

[0090]

[0091] Among them, PR (p i ) is the node p i The PageRank value of the field indicates the importance of the field in the association graph and is updated through iterative calculation; d is the damping coefficient, which indicates the probability that the user continues to browse along the edge (rather than randomly jumping), and is usually fixed to an empirical value of 0.85; N is the total number of nodes in the field relationship graph (for example, the number of all fields), which is fixed in a single calculation; M(p i ) points to node p i The set of all neighbor nodes (i.e., the set of nodes with p i nodes with incoming edge connections), which changes dynamically with the graph structure; w ji is the edge weight, i.e. from node p j to p i The edge weight represents the strength of association between two fields (such as co-occurrence frequency, similarity, etc.), which is pre-calculated from the data; PR (p j ) is the node p j PageRank value, and PR (p i ) Similarly, updates are passed through the values of adjacent nodes; L(p i ) is the node p j The sum of outgoing edge weights is used to balance the influence of different nodes.

[0092] Step 103c: Determine the sensitivity of each field based on the basic sensitivity and the associated sensitivity of each field.

[0093] After obtaining the basic sensitivity and the associated sensitivity of each field, the sensitivity of each field can be calculated using the comprehensive calculation formula

[0094] S = α × basic sensitivity + (1-α) × associated sensitivity

[0095] Where S is the sensitivity and α is the domain adjustment parameter (for example, medical data can take a value of 0.7, and financial data can take a value of 0.6).

[0096] In some business scenarios, some fields explicitly require the retention of original content (data). A whitelist rule field can be configured in the device, and its sensitivity can be directly assigned a large negative number (such as -100).

[0097] Step 104: Determine the security status of the user terminal based on the device fingerprint.

[0098] By analyzing and processing device fingerprints, we can infer the device's software and hardware characteristics, and then determine the security status of the user terminal. For example, if it is determined that the user terminal is jailbroken (iOS system) / rooted (Android system), it is considered that there is a security risk in the user terminal.

[0099] Step 105: Obtain index data of the current network encryption strength, and determine the network encryption strength according to the index data.

[0100] Network encryption strength is an important indicator that measures the data protection capabilities of the data encryption process in the network. Related indicators include protocol security and cipher suite ratings. Based on these indicators, the network encryption strength can be determined.

[0101] Step 106 , determining the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance.

[0102] Based on the request time and terminal location information, it can be determined whether the user has abnormal dimensions such as non-working hours / cross-border access, that is, time and space characteristics.

[0103] Operation behavior patterns include weights of operation types such as query, export, and share.

[0104] Data flow compliance refers to whether it complies with predefined data sharing policies.

[0105] In one embodiment, step 106 includes the following sub-steps:

[0106] Sub-step 106a, determining the respective risk values of the user terminal security status, network encryption strength, time and space characteristics (corresponding to request time and terminal location information), operation behavior pattern, and data flow compliance.

[0107] Sub-step 106b: determining the dynamic weight by the entropy weight method and calculating the scenario risk value.

[0108] For example, the risk values for a user's current terminal security status, network encryption strength, temporal and spatial characteristics (corresponding to request time and terminal location information), operational behavior pattern, and data flow compliance are as follows (all ranging from 0 to 1, with higher values indicating greater risk):

[0109] elements <![CDATA[Risk value f i > Terminal security status 0.9 Network encryption strength 0.7 Operational behavior mode 0.8 Temporal and spatial characteristics 0.6 Data flow compliance 0.5

[0110] The weights of each factor are calculated based on the historical data of all personnel in the last five times. The five samples are as follows:

[0111]

[0112]

[0113] Calculation of the entropy value E1 of the terminal security status:

[0114]

[0115] Use the same method to calculate the entropy of other factors:

[0116] The entropy value of network encryption strength is E2 = 0.975;

[0117] The entropy value of the operation behavior pattern is E3 = 0.897;

[0118] The entropy value of time-space characteristics is E4=0.96;

[0119] The entropy value of data flow compliance is E5 = 0.969;

[0120] Calculate the initial weights:

[0121]

[0122] Exponential moving average update weights

[0123] Weight calculated in the last hour:

[0124]

[0125] Real-time weight calculation:

[0126]

[0127] Scenario risk value calculation:

[0128]

[0129] Determining dynamic weights through the entropy weight method can objectively reflect the importance of each factor at different stages based on the actual situation of the data, avoiding the arbitrariness of subjective weighting.

[0130] Step 107 : Determine the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust.

[0131] After obtaining the sensitivity, scenario risk value, and user trust of each field, the weight parameters of the sensitivity, scenario risk value, and user trust of each field are determined, and the comprehensive risk of each field is calculated using a weighted summation method.

[0132] For example, the overall risk

[0133] Score=γ1×S+γ2×R+γ3×(1-T)

[0134] Where γ1, γ2, and γ3 are the weights corresponding to the field sensitivity, scenario risk value, and user trust, respectively; S is the field sensitivity, R is the scenario risk value, and T is the user trust.

[0135] In one embodiment, before step 107, the method may further include:

[0136] Step 107a, determining the user's basic trustworthiness based on the user category;

[0137] Step 107b: Count the user operation abnormality rate through the sliding time window, and modify the user basic trust according to the user operation abnormality rate to obtain the user trust.

[0138] For example, the data desensitization device may assign a basic trust value of 0.8 to a first-level user (such as a chief physician) and a basic trust value of 0.5 to a second-level user (such as an intern), thereby determining the basic trust value of the user according to the user category.

[0139] The user operation abnormality rate is counted through a sliding time window (such as 30 minutes), and the user operation abnormality rate is normalized using the Sigmoid function and used as the correction coefficient β∈[0,1].

[0140] User trust

[0141] T = user basic trust × β

[0142] Where β is the correction coefficient.

[0143] Step 108 : Select a desensitization strategy for each field based on the comprehensive risk of each field, perform desensitization on each field of the original data, and return the desensitized data to the user.

[0144] Desensitization strategies can be divided into four levels and seven categories, for example:

[0145] Level 0 (Score < 0.1):

[0146] A. Return to original text

[0147] Applicable conditions:

[0148] The field sensitivity is extremely low (such as non-sensitive fields such as "gender" and "job title");

[0149] The business scenario clearly requires retaining the original content (data).

[0150] Level 1 (0.1≤Score<0.3):

[0151] B. Generalization

[0152] Implementation:

[0153] k-anonymity (k≥5): ensure that each equivalence class contains at least 5 records;

[0154] l-diversity (l ≥ 3): each equivalence class has at least three different values for the sensitive attribute;

[0155] Example: The original age "32 years old" is generalized to "30-35 years old".

[0156] C. Noise injection

[0157] Perturbation range: uniform random disturbance within ±15%;

[0158] Example: The original income of "10,000 yuan" is disturbed to "9,850 yuan" or "11,500 yuan".

[0159] Level 2 (0.3≤Score<0.7):

[0160] D. Rule Mask

[0161] Configuration parameters: retain the first N and the last M characters, and replace the rest with mask symbols (N, M, and mask symbols are configurable, the default is N=3, M=4, and the mask symbol is "*");

[0162] Example: The mobile phone number "13812345678" is masked as "138****5678".

[0163] E. Format Preserving Encryption (FPE)

[0164] Format compatibility: maintain encryption of the original data format and support numbers, letters, and mixed formats;

[0165] Example: The bank card number "6225880123456789" is encrypted to "7293519876543210".

[0166] Level 3 (Score ≥ 0.7):

[0167] F. Strong encryption

[0168] Implementation: Using SM4 / AES-256 algorithm;

[0169] Example: The ID number "110101199003077832" is encrypted to "2kDf8s..." (ciphertext).

[0170] G. Complete replacement: Replace the original data with a hash value or virtual identifier

[0171] Implementation:

[0172] Hash value replacement: Use HMAC-SHA256 to generate an irreversible identifier;

[0173] Virtual identity: Generates a false but consistent substitute value through a mapping table;

[0174] Example: Patient ID "P12345" is replaced with "VX9BZQ".

[0175] It should be understood that the above desensitization strategy is only an example, and the specific desensitization strategy can be set according to the business type, data type, etc.

[0176] Desensitizing each field of the original data and returning the desensitized data to the user may specifically include the following sub-steps:

[0177] Sub-step 108a: parse the format of the original data and extract fields.

[0178] Sub-step 108b: split each field into independent processing objects.

[0179] Sub-step 108c: executing the corresponding desensitization strategy for each field in parallel.

[0180] Sub-step 108d: Output the data in the original data structure and append desensitization metadata, including the desensitization processing time and the ID of the desensitization strategy used for each field.

[0181] Sub-step 108e generates a log including the hash value of the original data, the desensitized metadata, the comprehensive risk of each field, and the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust.

[0182] In one embodiment, the method further comprises:

[0183] In step 109, the effectiveness of the desensitization strategy is periodically evaluated, and the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust are dynamically optimized based on the reinforcement learning model of the deep Q network.

[0184] In this step, the privacy protection strength, data availability, and performance overhead of the desensitization strategy are periodically evaluated. The weight parameters of the sensitivity of each field, the scenario risk value, and the user trust are dynamically optimized based on the reinforcement learning model of the deep Q network to achieve adaptive desensitization strategy adjustment.

[0185] The state space, action space, reward function, and optimization parameters of the reinforcement learning model of the deep Q network are defined as follows:

[0186] State Space

[0187] 1. Current gamma coefficient value (γ1, γ2, γ3);

[0188] 2. Historical re-identification rate (e.g., the past 7 days);

[0189] 3. Data availability score (e.g., obtained by collecting feedback from business departments, 0-10 points).

[0190] Action Space

[0191] γ1 adjustment range: ±0.05;

[0192] γ2 adjustment range: ±0.03;

[0193] γ3 is automatically calculated (keeping γ1+γ2+γ3=1).

[0194] Reward Function

[0195]

[0196] Where ReIDRate is the re-identification rate, UsabilityScore is the usability score, StrategyChanged is the number of switching times, and II is the strategy switching indicator function (used to suppress frequent adjustments, it takes 1 when the current decision causes r to change, otherwise it takes 0).

[0197] Optimization parameters

[0198] Adjust once a week, and generate new parameters after training to directly update the weight parameters γ1, γ2, and γ3 of the comprehensive risk calculation formula.

[0199] In one embodiment, the method further comprises:

[0200] Step 110 , audit the desensitized data features in the tracking log, calculate the re-identification risk rate, and trigger an alert and notify the administrator when the re-identification risk rate exceeds a threshold.

[0201] In this step, the desensitized data features may be features such as query frequency, data volume standard deviation, operation type distribution entropy, and cross-module access ratio. The re-identification risk rate may be calculated based on the isolation forest model. If the re-identification risk rate slightly exceeds the threshold (i.e., a mild anomaly), an alarm is recorded and the administrator is notified. If the re-identification risk rate seriously exceeds the threshold (i.e., a severe anomaly), the desensitization strategy is automatically upgraded (e.g., using a higher-level desensitization strategy, or using the highest-level desensitization strategy, etc.).

[0202] In this embodiment, by obtaining the user's data access request, the data access request includes the user identity, device fingerprint, request time and terminal location information; according to the data access request, determining the original data corresponding to the data access request; obtaining the sensitivity of each field of each sensitive entity in the original data; determining the user terminal security status according to the device fingerprint; obtaining the indicator data of the current network encryption strength, and determining the network encryption strength according to the indicator data; determining the scenario risk value according to the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance; determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust; selecting the desensitization strategy of each field based on the comprehensive risk of each field, desensitizing each field of the original data, and returning the desensitized data to the user. Differentiated desensitization processing is achieved, and high-value data can be effectively protected.

[0203] Corresponding to the data desensitization method of the present invention, the present invention also provides a data desensitization system, Figure 2 This is a structural diagram of a data desensitization device. Figure 2 As shown, the data desensitization device includes:

[0204] Request acquisition module 201, used to obtain a user's data access request, wherein the data access request includes user identity, device fingerprint, request time and terminal location information;

[0205] The original data determination module 202 is used to determine the original data corresponding to the data access request according to the data access request;

[0206] The sensitivity acquisition module 203 is used to obtain the sensitivity of each field of each sensitive entity in the original data;

[0207] The user terminal security status determination module 204 is used to determine the user terminal security status based on the device fingerprint;

[0208] The network encryption strength determination module 205 is used to obtain the index data of the current network encryption strength and determine the network encryption strength according to the index data;

[0209] Scenario risk value determination module 206, configured to determine a scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance;

[0210] Comprehensive risk determination module 207, for determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user's trust;

[0211] The desensitization execution module 208 is used to select a desensitization strategy for each field based on the comprehensive risk of each field, perform desensitization processing on each field of the original data, and return the desensitized data to the user.

[0212] In one embodiment, the apparatus further comprises:

[0213] The sensitivity determination module is used to identify sensitive entities in the original data and determine the basic sensitivity of sensitive entities; construct a field relationship map based on the original data, extract field relationships, build an adjacency matrix, use the improved PageRank algorithm to calculate the PageRank value of each field, and perform normalization to obtain the associated sensitivity of each field; determine the sensitivity of each field based on the basic sensitivity and the associated sensitivity of each field.

[0214] In one embodiment, the scenario risk value determination module 206 includes:

[0215] The first submodule is used to determine the risk values of the user terminal security status, network encryption strength, time and space characteristics, operation behavior patterns, and data flow compliance;

[0216] The first calculation submodule is used to determine the dynamic weight through the entropy weight method and calculate the scenario risk value.

[0217] In one embodiment, the apparatus further comprises:

[0218] The user trust confirmation module is used to determine the user's basic trust according to the user category; the user operation abnormality rate is counted through a sliding time window, and the user's basic trust is corrected according to the user operation abnormality rate to obtain the user's trust.

[0219] In one embodiment, the comprehensive risk determination module 207 includes:

[0220] The weight parameter determination submodule is used to determine the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust;

[0221] The second calculation submodule is used to calculate the comprehensive risk.

[0222] In one embodiment, the desensitization execution module 208 includes:

[0223] The field extraction submodule is used to parse the format of the original data and extract fields;

[0224] Split submodule, used to split each field into independent processing objects;

[0225] The policy execution submodule is used to execute the corresponding desensitization policy for each field in parallel;

[0226] The output submodule is used to output the original data structure and append desensitization metadata; the desensitization metadata includes the time of desensitization processing and the ID of the desensitization strategy adopted for each field;

[0227] The log generation submodule is used to generate logs containing the hash value of the original data, desensitized metadata, the comprehensive risk of each field, and the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust.

[0228] In one embodiment, the apparatus further comprises:

[0229] The evaluation and weight update module is used to periodically evaluate the effectiveness of the desensitization strategy and dynamically optimize the weight parameters of the sensitivity of each field, the scenario risk value, and the user trust based on the reinforcement learning model of the deep Q network.

[0230] A data desensitization device provided in an embodiment of the present invention can execute a data desensitization method provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0231] Figure 3 A schematic block diagram of an electronic device 30 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0232] like Figure 3As shown, the electronic device 30 includes at least one processor 31 and a memory, such as a read-only memory (ROM) 32, a random access memory (RAM) 33, etc., which is communicatively connected to the at least one processor 31. The memory stores a computer program that can be executed by the at least one processor. The processor 31 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 32 or the computer program loaded from the storage unit 38 into the random access memory (RAM) 33. Various programs and data required for the operation of the electronic device 30 can also be stored in the RAM 33. The processor 31, ROM 32, and RAM 33 are connected to each other via a bus 34. An input / output (I / O) interface 35 is also connected to the bus 34.

[0233] Multiple components in the electronic device 30 are connected to the I / O interface 35, including an input unit 36, such as a keyboard, a mouse, etc.; an output unit 37, such as various types of displays, speakers, etc.; a storage unit 38, such as a magnetic disk, an optical disk, etc.; and a communication unit 39, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 39 allows the electronic device 30 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0234] The processor 31 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the processor 31 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 31 performs the various methods and processes described above, such as the data desensitization method.

[0235] In some embodiments, the data desensitization method may be implemented as a computer program that is tangibly contained in a computer-readable storage medium, such as a storage unit 38. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 30 via the ROM 32 and / or the communication unit 39. When the computer program is loaded into the RAM 33 and executed by the processor 31, one or more steps of the data desensitization method described above may be performed. Alternatively, in other embodiments, the processor 31 may be configured to perform the data desensitization method in any other appropriate manner (e.g., by means of firmware).

[0236] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0237] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0238] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0239] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0240] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0241] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0242] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0243] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A data desensitization method, characterized in that: The method comprises: Obtaining a user's data access request, which includes the user's identity, device fingerprint, request time, and terminal location information; Determining, according to the data access request, the original data corresponding to the data access request; Obtain the sensitivity of each field of each sensitive entity in the original data; Determine the user terminal security status based on device fingerprint; Obtain indicator data of the current network encryption strength and determine the network encryption strength based on the indicator data; Determine the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance; Determine the comprehensive risk of each field based on its sensitivity, scenario risk value, and user trust; Select a desensitization strategy for each field based on the comprehensive risk of each field, desensitize each field of the original data, and return the desensitized data to the user.

2. The method according to claim 1, characterized in that Before obtaining the sensitivity of each field of each sensitive entity in the original data, the method further includes: Identify sensitive entities in the original data and determine the basic sensitivity of the sensitive entities; Build a field relationship map based on the original data, extract field relationships, build an adjacency matrix, use the improved PageRank algorithm to calculate the PageRank value of each field, and perform normalization to obtain the correlation sensitivity of each field; The sensitivity of each field is determined based on the base sensitivity and the associated sensitivity of each field.

3. The method according to claim 1, characterized in that Determining the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance includes: Determine the respective risk values of user terminal security status, network encryption strength, temporal and spatial characteristics, operational behavior patterns, and data flow compliance; The dynamic weight is determined by the entropy weight method and the scenario risk value is calculated.

4. The method according to claim 1, wherein Before determining the comprehensive risk of each field based on the sensitivity of each field, the scenario risk value, and the user trust, the following steps may also be included: Determine user base trust based on user categories; The user operation abnormality rate is counted through a sliding time window, and the user basic trust is corrected according to the user operation abnormality rate to obtain the user trust.

5. The method according to claim 1, characterized in that The comprehensive risk of each field is determined based on the sensitivity of each field, the scenario risk value, and the user trust, including: Determine the weight parameters of each field's sensitivity, scenario risk value, and user trust; Calculate the overall risk.

6. The method according to claim 1, characterized in that The desensitizing process is performed on each field of the original data, and the desensitized data is returned to the user, including: Parse the format of the original data and extract fields; Split each field into an independent processing object; Execute the corresponding desensitization strategy for each field in parallel; Output in the structure of original data, with additional desensitized metadata; the desensitized metadata includes the time of desensitization processing and the ID of the desensitization strategy used for each field; Generates a log containing the hash value of the original data, desensitized metadata, the comprehensive risk of each field, and the weighted parameters of each field's sensitivity, scenario risk value, and user trust.

7. The method according to claim 6, characterized in that Also includes: The effectiveness of the desensitization strategy is periodically evaluated, and the weight parameters of the sensitivity of each field, scenario risk value, and user trust are dynamically optimized based on the reinforcement learning model of the deep Q network.

8. A data desensitization device, characterized in that: include: A request acquisition module is used to obtain a user's data access request, wherein the data access request includes user identity, device fingerprint, request time and terminal location information; The original data determination module is used to determine the original data corresponding to the data access request according to the data access request; A sensitivity acquisition module is used to obtain the sensitivity of each field of each sensitive entity in the original data; A user terminal security status determination module is used to determine the user terminal security status based on the device fingerprint; The network encryption strength determination module is used to obtain the indicator data of the current network encryption strength and determine the network encryption strength based on the indicator data; A scenario risk value determination module is used to determine the scenario risk value based on the user terminal security status, network encryption strength, request time and terminal location information, operation behavior pattern, and data flow compliance; Comprehensive risk determination module, used to determine the comprehensive risk of each field based on the sensitivity of each field, scenario risk value, and user trust; The desensitization execution module is used to select the desensitization strategy for each field based on the comprehensive risk of each field, desensitize each field of the original data, and return the desensitized data to the user.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the data masking method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the data desensitization method according to any one of claims 1 to 7 when executed.

Images