Sensitive information privacy calculation multi-dimensional security risk evaluation method

Abstract

The invention relates to the technical field of information security, in particular to a sensitive information privacy calculation multi-dimensional security risk evaluation method, which comprises the following steps of: cooperatively collecting sensitive data information by setting a plurality of distributed nodes, and constructing the collected sensitive data information into corresponding feature vectors; performing privacy calculation based on the constructed feature vector; performing multi-dimensional assessment of the security risk based on a recovery result of the feature vector; according to the method, the feature vectors are encrypted layer by layer by using a multi-stage differential privacy mechanism, original data exposure is avoided, the risk of privacy disclosure is reduced, and personalized encryption is realized by dynamically adjusting the noise variance, and privacy protection and data validity are both considered; meanwhile, a distributed dynamic key management and decryption mechanism is introduced, and the key is prevented from being maliciously inferred through multi-round decryption calculation; and dynamically dividing risk level intervals, automatically matching feature evaluation results, and quickly identifying risk levels.

Description

Technical Field

[0001] The present invention relates to the technical field of information security, and specifically to a multi-dimensional security risk assessment method for sensitive information privacy computing. Background Art

[0002] In the current information security field, privacy protection and security risk assessment of sensitive information are key technical difficulties. However, at this stage, security assessment of sensitive information faces the following challenges: Traditional encryption methods cannot fully protect the privacy of sensitive information, especially in distributed data storage and computing environments, which are prone to privacy leakage due to data transmission or centralized processing; The static key management mechanism is easily inferred by malicious intent, single-point data decryption has a high attack success rate, and there is a lack of dynamic protection strategies for multi-level encryption and joint decryption mechanisms; The risk assessment model is single and fails to comprehensively consider the impact of multi-dimensional characteristics on risk. The assessment results are not accurate enough and there is a lack of dynamic adjustment noise mechanism, which leads to reduced data validity or insufficient privacy protection. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] To solve the above technical problems, the present invention provides the following technical solution: a multi-dimensional security risk assessment method for sensitive information privacy computing, comprising the following steps: By setting up multiple distributed nodes to collaboratively collect sensitive data information, the collected sensitive data information is constructed into corresponding feature vectors; Privacy calculation is performed based on the constructed feature vector, specifically: Set up multiple encryption levels, dynamically encrypt the feature vectors layer by layer, and adjust the degree of dynamic encryption in real time through dynamic adjustment functions, and formulate a distributed dynamic key decryption mechanism to try to decrypt the encrypted data in a homomorphic way, including: Set up multiple rounds of decryption calculations, generate different encryption key pairs based on the hash function in each round, and set up multiple decryption spaces to locally decrypt the feature vector using subkeys. Determine whether to restore the encrypted feature vector based on the number of participants in the decryption space. Based on the recovery results of the feature vector, a multi-dimensional assessment of security risks is performed, specifically: A security risk assessment model is constructed based on the risk score of the features in the feature vector, and the recovery result of the feature vector is used as the input of the model. At the same time, the risk level interval is divided, and the output result of the model is matched with the risk level interval to realize security risk assessment.

[0005] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the dynamic encryption of the feature vector layer by layer is specifically as follows: The privacy protection of the base layer is to add basic noise to the constructed feature vector, then we have:

[0006] Sensitive layer privacy protection is to add enhanced noise to highly sensitive features, then,

[0007] Global privacy protection is to add global noise to the comprehensive results of all nodes, then,

[0008] in, represents the extracted abnormal distribution features, represents the constructed feature vector, Indicates the total number of distributed nodes. It represents the result after adding basic noise to the privacy protection layer of the base layer. It represents the result after the privacy protection layer of the sensitive layer adds enhanced noise to the highly sensitive features. It represents the result after adding global noise to the global privacy protection layer, and the comprehensive result after encrypting the local data of each node. represents a Gaussian noise function with a mean of 0 and a variance of The normal distribution of , , Represent the noise variance of the base layer, sensitive layer and global layer respectively, satisfying the formula and upload the encrypted data to the central server.

[0009] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the real-time adjustment of the degree of dynamic encryption is specifically as follows: When adjusting the base layer noise variance, the base noise variance is adjusted according to the volatility of the data in the eigenvector, then,

[0010] in, represents the control coefficient, Represents the feature vector The variance of represents the noise variance of the base layer; When the noise variance of the sensitive layer is adjusted, the noise variance of the sensitive layer is dynamically adjusted according to the sensitivity index of the sensitive feature, then,

[0011] in, represents the control coefficient, Indicates the feature vector Sensitive features in the feature vector The importance of the implementation is determined by the implementer according to the actual application scenario. represents the noise variance of the sensitive layer; When adjusting the global layer noise method, the global noise variance is dynamically adjusted according to the mean of the eigenvectors of all nodes, then,

[0012] in, represents the control coefficient, Represents the mean of the eigenvectors of all nodes, which is the mean of the eigenvectors of all distributed nodes. represents the noise variance of the global layer.

[0013] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the generation of different encryption key pairs based on the hash function is as follows: Initialize security seed , and set the hash function used to generate a unique round identifier, then,

[0014] in, Indicates safe seeds. Indicates rounds, Indicates The round identifier corresponding to the round, Represents a cryptographic hash function, a function with a fixed output length of 256 bits; Generate a key pair based on the round identifier, then we have,

[0015] in, Represents a key generation function, which outputs the corresponding key pair according to the dynamically generated hash value. Indicates the generated key pair. Indicates The public key corresponding to the round is distributed to the distributed nodes to encrypt the feature vector. Indicates The private key corresponding to the round is distributed to the central server to decrypt the encrypted feature vector.

[0016] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the local decryption of the feature vector using the subkey is specifically as follows: When the distributed node encrypts the feature vector, it uses the public key in the generated key pair to encrypt the local feature vector, then,

[0017] in, Indicates The rounds correspond to the encrypted feature vectors, which are uploaded to the central server for homomorphic decryption; Divide the private key of the key pair into The molecular key is,

[0018] in, Indicates the division The sub-keys are allocated to the corresponding decryption space in the central server. Indicates the total number of subkeys; Each decryption space uses the assigned subkey to locally decrypt the feature vector and generates the corresponding local decryption result, then,

[0019] in, Indicates that the decryption space uses The decryption result of the molecular key pair feature vector.

[0020] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the method of judging whether to perform result recovery for the encrypted feature vector according to the number of participants in the decryption space is as follows: For the local decryption result , set the threshold of the number of joint decryption spaces , the number of spaces involved in feature vector decryption is , then there is, If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of joint decryption spaces meets the feature vector recovery requirements, and starts the decryption space corresponding to all subkeys for local decryption, and combines all local results , the combined result is used as the recovery result of the feature vector, and the recovery result is used as the evaluation basis for the subsequent multi-dimensional evaluation of security risks; If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of decryption space joints cannot meet the feature vector recovery requirements, and the feature vector recovery request of the current round is rejected until the number of decryption space joints meets the requirements.

[0021] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the security risk assessment model constructed based on the risk score of the features in the feature vector is specifically as follows: Constructing the comprehensive risk scoring function, we have:

[0022] in, represents the first Features, Indicates the weight coefficient corresponding to the feature, which is set by the implementer according to the actual application scenario. Indicates The amount of risk data corresponding to each feature, Indicates the total amount of data involved in the evaluation, Indicates The comprehensive risk score corresponding to each feature; Based on the calculated risk score, the risk level is assessed, in, Indicates The comprehensive risk score corresponding to each feature is Indicates the total number of feature types, Represents the risk assessment result for the input features, which is used to implement multi-dimensional assessment of security risks.

[0023] As a preferred solution of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention, the security risk assessment is specifically as follows: For risk assessment results , and divide it into 3 interval levels according to the same interval length, and satisfy the formula , and the interval length satisfies ; According to the divided interval levels, security risk assessment is carried out, specifically: According to the input features, the interval level is automatically matched according to the output results of the risk assessment model. If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the first-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the second-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the third level of risk warning.

[0024] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a multi-dimensional security risk assessment method for sensitive information privacy computing when executing the computer program.

[0025] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a multi-dimensional security risk assessment method for sensitive information privacy computing.

[0026] Beneficial effects of the present invention: The present invention uses a multi-level differential privacy mechanism to encrypt feature vectors layer by layer, avoiding exposure of original data and reducing the risk of privacy leakage. It achieves personalized encryption by dynamically adjusting the noise variance, taking into account both privacy protection and data validity. At the same time, a distributed dynamic key management and decryption mechanism is introduced to prevent the key from being maliciously inferred through multiple rounds of decryption calculations; Construct a multi-dimensional risk assessment model based on feature vectors, optimize the contribution of key features through weight allocation, and improve the accuracy of assessment results; Dynamically divide risk levels, automatically match feature assessment results, and quickly identify risk levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing 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. Among them: Figure 1 This is a schematic diagram of the overall method steps of the multi-dimensional security risk assessment method for sensitive information privacy computing of the present invention. DETAILED DESCRIPTION

[0028] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0029] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0030] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0031] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0032] Meanwhile, in the description of the present invention, it should be noted that the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0033] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. Example

[0034] Reference Figure 1 , is an embodiment of the present invention, and provides a multi-dimensional security risk assessment method for sensitive information privacy computing, comprising the following steps: S1: Collection and processing of sensitive information data.

[0035] Specifically, the collection of sensitive information data is done through a federated learning framework, with multiple distributed nodes collaboratively collecting sensitive data information. At the same time, an encrypted communication channel is established between the node and the server through a secure communication protocol for subsequent data transmission. For the collected sensitive data information, the sensitive data information is processed by extracting sensitive features. The specific implementation is as follows: set up distributed nodes, then we have, , and each node collects independent local sensitive data information ; Based on the collected local sensitive data information, the sensitive feature types are divided into Types include attack characteristics, abnormal distribution characteristics, and traffic characteristics, specifically: Extract attack features based on the number of abnormal data requests per unit time , then there is,

[0036] in, The number of data requests showing abnormal attacks. Representation Node The total amount of local sensitive data information collected, represents the extracted attack features, Based on the distribution of data entropy, we can extract abnormal distribution features, then we have:

[0037] in, Indicates that it belongs to the category The number of Representation Node The total amount of local sensitive data information collected, represents the calculated entropy distribution result, which is the extracted abnormal distribution feature. Based on the maximum flow peak and average flow, the flow characteristics are extracted,

[0038] in, Representation Node Local data The average value of all traffic records in is the average load of the node. Representing a collection of data The elements, is the index sequence number of the element in the data set, satisfying the formula , Representation Node The number of recorded flow values ​​in ; Based on the extracted features, the feature vector is constructed and encrypted, then, Constructing feature vectors , specifically:

[0039] in, represents the extracted attack features, represents the extracted abnormal distribution features, represents the extracted traffic features, Representation Node The maximum flow rate in Represents the constructed feature vector.

[0040] It should be noted that for the constructed feature vector , and transmitted to the central server using the established encrypted communication channel for subsequent privacy computing.

[0041] S2: Perform privacy computing based on sensitive data information processing results.

[0042] Specifically, the privacy calculation based on the sensitive data information processing result is to perform privacy calculation based on the feature vector uploaded to the central server.

[0043] Furthermore, before uploading to the central server, the extracted feature vector is dynamically encrypted using a multi-level differential privacy mechanism. The multi-level differential privacy mechanism sets multiple encryption levels and dynamically encrypts the feature vector layer by layer, so as to improve the validity of the data while ensuring privacy and reduce the success rate of single-point data attacks. It includes basic layer privacy protection, sensitive layer privacy protection and global privacy protection. The specific encryption process is as follows: The privacy protection of the base layer is to add basic noise to the constructed feature vector, then we have:

[0044] Sensitive layer privacy protection is to add enhanced noise to highly sensitive features, then,

[0045] Global privacy protection is to add global noise to the comprehensive results of all nodes, then,

[0046] in, It represents the result after adding basic noise to the privacy protection layer of the base layer. It represents the result after the privacy protection layer of the sensitive layer adds enhanced noise to the highly sensitive features. It represents the result after adding global noise to the global privacy protection layer, and the comprehensive result after encrypting the local data of each node. represents a Gaussian noise function with a mean of 0 and a variance of The normal distribution of , , Represent the noise variance of the base layer, sensitive layer and global layer respectively, satisfying the formula , and upload the encrypted data to the central server. The encrypted data is homomorphically decrypted in the central server to obtain the global feature vector, thereby completing the privacy calculation of sensitive data.

[0047] It should be noted that the noise variance of the base layer, sensitive layer and global layer is adjusted in real time by setting a dynamic adjustment function. The specific adjustments are as follows: When adjusting the base layer noise variance, the base noise variance is adjusted according to the volatility of the data in the eigenvector, then, in, represents the control coefficient, Represents the feature vector The variance of represents the noise variance of the base layer; When the noise variance of the sensitive layer is adjusted, the noise variance of the sensitive layer is dynamically adjusted according to the sensitivity index of the sensitive feature, then,

[0048] in, represents the control coefficient, Indicates the feature vector Sensitive features in the feature vector The importance of the implementation is determined by the implementer according to the actual application scenario. represents the noise variance of the sensitive layer; When adjusting the global layer noise method, the global noise variance is dynamically adjusted according to the mean of the eigenvectors of all nodes, then,

[0049] in, represents the control coefficient, Represents the mean of the eigenvectors of all nodes, which is the mean of the eigenvectors of all distributed nodes. represents the noise variance of the global layer.

[0050] The dynamic adjustment of noise variance can not only avoid the decrease in data validity caused by excessive noise, but also prevent the risk of privacy leakage caused by too little noise. At the same time, personalized encryption can be achieved based on the data distribution characteristics.

[0051] Furthermore, the homomorphic decryption of the encrypted data is achieved by formulating a distributed dynamic key decryption mechanism, and the distributed dynamic decryption mechanism includes dynamic key management and distributed decryption. The dynamic key management is to set multiple rounds of decryption calculations, and use a different encryption key pair in each round to prevent the ciphertext from being maliciously managed and inferred. At the same time, dynamic key updates are introduced to make the key pair time-effective and unique, as follows: Initialize security seed , and set the hash function used to generate a unique round identifier, then,

[0052] in, Indicates safe seeds. Indicates rounds, Indicates The round identifier corresponding to the round, Represents a cryptographic hash function, a function with a fixed output length of 256 bits; Generate a key pair based on the round identifier, then we have,

[0053] in, Represents a key generation function, which outputs the corresponding key pair according to the dynamically generated hash value. Indicates the generated key pair. Indicates The public key corresponding to the round is distributed to the distributed nodes to encrypt the feature vector. Indicates The private key corresponding to the round is distributed to the central server to decrypt the encrypted feature vector.

[0054] When the distributed node encrypts the feature vector, it uses the public key in the generated key pair to encrypt the local feature vector, then,

[0055] in, Indicates The rounds correspond to the encrypted feature vectors, which are uploaded to the central server for homomorphic decryption; The central server uses the private key of the key pair Perform homomorphic decryption of the feature vector, specifically: Divide the private key of the key pair into The molecular key is,

[0056] in, Indicates the division The sub-keys are allocated to the corresponding decryption space in the central server. Indicates the total number of subkeys; Each decryption space uses the assigned subkey to locally decrypt the feature vector and generates the corresponding local decryption result, then,

[0057] in, Indicates that the decryption space uses The decryption result of the molecular key pair feature vector.

[0058] According to the local decryption results, a joint decryption mechanism is set up to recover the feature vector, specifically: For the local decryption result , set the threshold for the number of joint decryption spaces , then there is, If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of joint decryption spaces meets the feature vector recovery requirements, and starts the decryption space corresponding to all subkeys for local decryption, and combines all local results , the combined result is used as the recovery result of the feature vector, and the recovery result is used as the evaluation basis for the subsequent multi-dimensional evaluation of security risks; If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of decryption space joints cannot meet the feature vector recovery requirements, and the feature vector recovery request of the current round is rejected until the number of decryption space joints meets the requirements.

[0059] S3: Based on the results of privacy computing, conduct a multi-dimensional assessment of security risks.

[0060] Specifically, the multi-dimensional assessment of security risks is performed by constructing a risk assessment model, taking the recovery results corresponding to the feature vectors as the input of the model, and implementing the multi-dimensional assessment of security risks according to the output results of the model. The specific implementation is as follows: The recovery result corresponding to the feature vector is used as the input of the risk assessment model, then, For the restored result corresponding to the eigenvector , and satisfies the formula ,in, represents the recovery result corresponding to the first distributed node feature vector, Indicates The recovery result corresponding to the feature vector of distributed nodes; For each distributed node corresponding to the feature vector, it satisfies the formula ,in, represents the first feature in the feature vector, represents the first Features The feature vector corresponding to each distributed node is used as the input of the risk assessment model; Construct a risk assessment model based on risk assessment, specifically: Constructing the comprehensive risk scoring function, we have:

[0061] in, represents the first Features, Indicates the weight coefficient corresponding to the feature, which is set by the implementer according to the actual application scenario. Indicates The amount of risk data corresponding to each feature, Indicates the total amount of data involved in the evaluation, Indicates The comprehensive risk score corresponding to each feature; Based on the calculated risk score, the risk level is assessed,

[0062] in, Indicates The comprehensive risk score corresponding to each feature is Indicates the total number of feature types, Represents the risk assessment result for the input features, which is used to implement multi-dimensional assessment of security risks, specifically: The risk assessment results are used to implement a multi-dimensional assessment of security risks, specifically: For risk assessment results , and divide it into 3 interval levels according to the same interval length, and satisfy the formula , and the interval length satisfies ; According to the divided interval levels, security risk assessment is carried out, specifically: According to the input features, the interval level is automatically matched according to the output results of the risk assessment model. If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the first-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the second-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the third level of risk warning.

[0063] It should be noted that for the determined risk warning level, corresponding emergency measures are taken according to the warning level. The specific emergency measures are set by the implementers according to the actual application scenario and are not described in detail in this embodiment.

[0064] Furthermore, if the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0065] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0066] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0067] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment may not be described (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those features that are not relevant to implementing the invention).

[0068] It will be appreciated that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but will be a routine task of design, fabrication, and production for those of ordinary skill having the benefit of this disclosure without undue experimentation.

[0069] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-dimensional security risk assessment method for sensitive information privacy computing, characterized by : Includes the following steps By setting up multiple distributed nodes to collaboratively collect sensitive data information, the collected sensitive data information is constructed into corresponding feature vectors; Privacy calculation is performed based on the constructed feature vector, specifically: Set up multiple encryption levels, dynamically encrypt the feature vectors layer by layer, and adjust the degree of dynamic encryption in real time through dynamic adjustment functions, and formulate a distributed dynamic key decryption mechanism to try to decrypt the encrypted data in a homomorphic way, including: Set up multiple rounds of decryption calculations, generate different encryption key pairs based on the hash function in each round, and set up multiple decryption spaces to use subkeys to partially decrypt the feature vectors. Determine whether to restore the results for the encrypted feature vectors based on the number of participants in the decryption space; A multi-dimensional assessment of security risks is performed based on the recovery results of the feature vector, specifically: A security risk assessment model is constructed based on the risk score of the features in the feature vector, and the recovery result of the feature vector is used as the input of the model. At the same time, the risk level interval is divided, and the output result of the model is matched with the risk level interval to realize security risk assessment.

2. The multi-dimensional security risk assessment method for sensitive information privacy computing according to claim 1 is characterized in that The dynamic encryption of the feature vector layer by layer is specifically as follows: The privacy protection of the base layer is to add basic noise to the constructed feature vector, then we have: Sensitive layer privacy protection is to add enhanced noise to highly sensitive features, then, Global privacy protection is to add global noise to the comprehensive results of all nodes, then, in, represents the extracted abnormal distribution features, represents the constructed feature vector, Indicates the total number of distributed nodes. It represents the result after adding basic noise to the privacy protection layer of the base layer. It represents the result after the privacy protection layer of the sensitive layer adds enhanced noise to the highly sensitive features. It represents the result after adding global noise to the global privacy protection layer, and the comprehensive result after encrypting the local data of each node. represents a Gaussian noise function with a mean of 0 and a variance of The normal distribution of , , Represent the noise variance of the base layer, sensitive layer and global layer respectively, satisfying the formula and upload the encrypted data to the central server.

3. The sensitive information privacy computing multi-dimensional security risk assessment method according to claim 2, characterized in that: The real-time adjustment of the degree of dynamic encryption is specifically as follows: When adjusting the base layer noise variance, the base noise variance is adjusted according to the volatility of the data in the eigenvector, then, in, represents the control coefficient, Represents the feature vector The variance of represents the noise variance of the base layer; When the noise variance of the sensitive layer is adjusted, the noise variance of the sensitive layer is dynamically adjusted according to the sensitivity index of the sensitive feature, then, in, represents the control coefficient, Indicates the feature vector Sensitive features in the feature vector The importance of the implementation is determined by the implementer according to the actual application scenario. represents the noise variance of the sensitive layer; When adjusting the global layer noise method, the global noise variance is dynamically adjusted according to the mean of the eigenvectors of all nodes, then, in, represents the control coefficient, Represents the mean of the eigenvectors of all nodes, which is the mean of the eigenvectors of all distributed nodes. represents the noise variance of the global layer.

4. The method for multi-dimensional security risk assessment of sensitive information privacy computing according to claim 3, characterized in that: The generation of different encryption key pairs based on the hash function is as follows: Initialize security seed , and set the hash function used to generate a unique round identifier, then, in, Indicates safe seeds. Indicates rounds, Indicates The round identifier corresponding to the round, Represents a cryptographic hash function, a function with a fixed output length of 256 bits; Generate a key pair based on the round identifier, then we have, in, Represents a key generation function, which outputs the corresponding key pair according to the dynamically generated hash value. Indicates the generated key pair. Indicates The public key corresponding to the round is distributed to the distributed nodes to encrypt the feature vector. Indicates The private key corresponding to the round is distributed to the central server to decrypt the encrypted feature vector.

5. The method for multi-dimensional security risk assessment of sensitive information privacy computing according to claim 4, characterized in that: The specific method of using the subkey to locally decrypt the feature vector is as follows: When the distributed node encrypts the feature vector, it uses the public key in the generated key pair to encrypt the local feature vector, then, in, Indicates The rounds correspond to the encrypted feature vectors, which are uploaded to the central server for homomorphic decryption; Divide the private key of the key pair into The molecular key is, in, Indicates the division The sub-keys are allocated to the corresponding decryption space in the central server. Indicates the total number of subkeys; Each decryption space uses the assigned subkey to locally decrypt the feature vector and generates the corresponding local decryption result, then, in, Indicates that the decryption space uses The decryption result of the molecular key pair feature vector.

6. The method for multi-dimensional security risk assessment of sensitive information privacy computing according to claim 5, characterized in that: The specific method of judging whether to perform result recovery on the encrypted feature vector according to the number of participants in the decryption space is as follows: For the local decryption result , set the threshold of the number of joint decryption spaces , the number of spaces involved in feature vector decryption is , then there is, If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of joint decryption spaces meets the feature vector recovery requirements, and starts the decryption space corresponding to all subkeys for local decryption, and combines all local results , the combined result is used as the recovery result of the feature vector, and the recovery result is used as the evaluation basis for the subsequent multi-dimensional evaluation of security risks; If the number of spaces participating in feature vector decryption and the joint number threshold satisfy the formula , indicating that the number of decryption space joints cannot meet the feature vector recovery requirements, and the feature vector recovery request of the current round is rejected until the number of decryption space joints meets the requirements.

7. The method for multi-dimensional security risk assessment of sensitive information privacy computing according to claim 6, characterized in that: The security risk assessment model constructed based on the risk score of the features in the feature vector is specifically as follows: Constructing the comprehensive risk scoring function, we have: in, represents the first Features, Indicates the weight coefficient corresponding to the feature, which is set by the implementer according to the actual application scenario. Indicates The amount of risk data corresponding to each feature, Indicates the total amount of data involved in the evaluation, Indicates The comprehensive risk score corresponding to each feature; Based on the calculated risk score, the risk level is assessed, in, Indicates The comprehensive risk score corresponding to each feature is Indicates the total number of feature types, Represents the risk assessment result for the input features, which is used to implement multi-dimensional assessment of security risks.

8. The sensitive information privacy computing multi-dimensional security risk assessment method according to claim 7, characterized in that: The security risk assessment is as follows: For risk assessment results , and divide it into 3 interval levels according to the same interval length, and satisfy the formula , and the interval length satisfies ; According to the divided interval levels, security risk assessment is carried out, specifically: According to the input features, the interval level is automatically matched according to the output results of the risk assessment model. If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the first-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is , which is the second-level risk warning level; If the formula is satisfied , indicating the characteristics The corresponding risk assessment level is It is the third level of risk warning.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images