Multi-department government affair cooperation method, equipment, medium and product

Abstract

The invention provides a multi-department government affair cooperation method and device, a medium and a product, and relates to the technical field of data security, the method comprises the following steps: obtaining original data matrixes input in the government affair cooperation process of different departments to carry out dimension reduction, and generating feature vectors of the original data matrixes of the departments; calculating a KL divergence value to determine the privacy protection degree of each feature vector; splicing the feature vectors of the departments to obtain an n-dimensional feature vector matrix, and combining the feature vector matrix with a rank missing matrix with a preset rank of n-1 to obtain a joint feature matrix; a loss function used for balancing the privacy protection degree of original data and information loss is constructed, and a feature vector and a joint feature matrix which minimize a loss function value are determined. According to the method, the joint feature matrix has no unique solution, the problem that original data are reconstructed due to the fact that model parameters are prone to being reversely deduced is solved, and the problem that privacy protection and information loss of the original data are difficult to balance is solved through the constructed loss function.

Description

Technical Field

[0001] This invention relates to the field of data security technology, and in particular to a method, device, medium and product for multi-departmental government collaboration. Background Technology

[0002] With the continuous advancement of e-government digitalization, cross-domain data collaboration between different departments is becoming increasingly frequent. In scenarios such as emergency response, timely and accurate data collaboration is crucial, directly impacting the effectiveness and timeliness of countermeasures. However, current data collaboration faces numerous challenges. On the one hand, due to sensitive issues involving citizen privacy and institutional permissions, data often requires manual approval or anonymization. Manual approval processes are cumbersome, consuming significant time and manpower, resulting in extremely low collaboration efficiency and failing to meet the urgent time requirements of emergency tasks. On the other hand, while anonymization protects privacy to some extent, it can cause severe information loss, significantly compromising data integrity and accuracy, thereby affecting the precision of decision-making and potentially leading to biased or erroneous decisions.

[0003] To address data privacy and collaboration issues, traditional federated learning methods have emerged. This approach ensures data security by maintaining "data within the domain," meaning raw data remains local to each department and is not directly transmitted externally. Collaboration occurs only through sharing model parameters and computation results, allowing participants to train the model while protecting their local data privacy. However, this method is not without its flaws and still carries the risk of model parameter leakage. Attackers could potentially analyze model parameters to infer information from the raw data, jeopardizing data security. While secure multi-party computation excels in security, ensuring participants complete computational tasks without disclosing their own data, its computational complexity increases exponentially with the number of participants. In large-scale model environments, the sheer volume and complexity of data to be processed make real-time response difficult for secure multi-party computation, failing to meet the demands of rapid data processing in practical applications. Summary of the Invention

[0004] The embodiments of the present invention provide a method, device, medium and product for multi-departmental government collaboration, which aims to solve the problems that the model parameters of existing government data collaboration are easily reverse-derived and the original data is reconstructed, and that it is difficult to balance the privacy protection and information loss of the original data.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a multi-departmental government collaboration method, comprising the following steps: Obtain the raw data matrix input from multiple departments participating in government collaboration; The original data matrix is ​​dimensionality reduced using an encoder to generate feature vectors for each department's original data matrix, where: , In the formula, Z is the feature vector of the department; X is the original data matrix; W e b is the encoder weight matrix; e Here, represents the bias vector; ReLU represents the activation function. This represents the mapping function from the original data matrix X to the virtual eigenvector Z; The privacy protection level of the feature vector is determined by calculating the KL divergence value; the KL divergence value is calculated using the following formula: , , In the formula, Represents the coding distribution Compared with the reference distribution KL divergence function between them; μ It is an eigenvector Z The mean vector; Σ is the eigenvector. Z The covariance matrix; is the preset KL divergence threshold; k is the dimension of the original data matrix; Represents the coding distribution; Indicates the reference distribution; Represents a normal distribution; An n-dimensional feature vector matrix is ​​obtained by concatenating the feature vectors of each department. This feature vector matrix is ​​then combined with a rank-deficient matrix of rank n-1 to obtain a joint feature matrix, where: , In the formula, F Represents a joint feature matrix of multiple departments. These represent virtual feature vectors for n government departments. This indicates a row-by-row concatenation operation, which is a joint matrix formed by horizontally concatenating the feature matrices of different departments. M Let be a rank-deficient matrix, where ; By constructing a loss function to balance the degree of privacy protection of the original data and information loss, the feature vector and joint feature matrix that minimize the value of the loss function are determined. The loss function is calculated using the following formula: , In the formula, L The loss function; Y Label the target decision; W d The decoder weight matrix is ​​the matrix corresponding to the joint feature matrix; This indicates information loss; λγ is the privacy protection strength coefficient; γ is the feature sparsity coefficient; Represents the encoder weight matrix W e The L1 norm; The joint feature matrix is ​​sent to multiple departments participating in government collaboration, so that these departments can make collaborative decisions based on the joint feature matrix.

[0006] Furthermore, the original data matrix also identifies key features through a two-layer attention mechanism, including: The local attention weights are calculated using the following formula: , In the formula, For the original data matrix, the first... One feature; x j For the original data matrix, the first... j One characteristic, d The total number of features in the original data matrix; Represents the original data matrix. Local attention weights for each feature; W a This is the attention weight matrix; The global attention weights are calculated using the following formula: , In the formula, Represents the original data matrix. Global attention weights for each feature; Represents the original data matrix. Features Cosine similarity to the target decision label Y; This represents the maximum cosine similarity between all features of the original data matrix and the target decision label; A comprehensive weight is calculated using the local attention weights and the total attention weights. Features with a comprehensive weight greater than or equal to a preset comprehensive weight threshold are identified as key features of the original data matrix. The comprehensive weight is calculated using the following formula: , In the formula, Represents the original data matrix. The combined weight of each feature.

[0007] Furthermore, before dimensionality reduction via the encoder, the original data matrix is ​​adjusted in dimensionality based on feedback from changes in information loss and privacy protection levels, calculated using the following formula: , In the formula, The dimensions of the original data matrix after adjustment; The dimensions of the original data matrix before adjustment; Indicates the change in information loss; This represents the change in the level of privacy protection. This represents the learning rate.

[0008] Furthermore, the joint feature matrix is ​​also verified for invertibility using the following formula: , In the formula, Represents the matrix condition number; The Jacobian matrix representing the joint characteristic matrix; This represents the largest singular value of the Jacobian matrix; Represents the smallest singular value of the Jacobian matrix; when When the value is greater than the preset value, it indicates that the verification result of the invertibility of the joint feature matrix meets the standard.

[0009] Furthermore, the parameters, including the encoder weight matrix, decoder weight matrix, and bias vector, are adjusted using gradient descent and adjusted according to the following formula: ; In the formula, These are the adjusted parameters; These are the parameters before adjustment; The learning rate; For effect enhancement; This represents the gradient of the loss function; This represents the gain coefficient.

[0010] Furthermore, the collaborative decision-making determines the optimal decision based on the joint decision probability distribution of the joint feature matrix, calculated using the following formula: ; In the formula, S is the probability distribution after mapping the joint feature matrix through the decision weight matrix; W s This is the decision weight matrix; The decision with the highest probability is selected as the optimal decision based on the probability distribution.

[0011] Furthermore, the optimal decision also incorporates a comprehensive score for the decision-making effect based on the degree of privacy protection, and the comprehensive score is calculated using the following formula: , In the formula, Indicates the decision effect score. Indicates the accuracy of the prediction; The time-consuming nature of traditional decision-making; The time required for the optimal decision; This represents the KL divergence value. When the decision performance score is greater than a preset score threshold, it indicates that the best decision performance is better than the traditional decision performance.

[0012] Secondly, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores at least one program, which is executed by the processor to implement the multi-departmental government collaboration method as described above.

[0013] Thirdly, the present invention provides a computer-readable storage medium storing at least one program, which is executed by a processor to implement the multi-departmental government collaboration method described above.

[0014] Fourthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-departmental government collaboration method described above.

[0015] The above technical solution has the following technical effects: This invention reduces the dimensionality of the original data matrices input from multiple departments participating in government collaboration, generating feature vectors for each department's original data matrix. The KL divergence value is calculated to determine the privacy protection level of each feature vector. An n-dimensional feature vector matrix is ​​obtained by concatenating the feature vectors from each department. This feature vector matrix is ​​then combined with a rank-defining matrix with a preset rank of n-1 to obtain a joint feature matrix. A loss function is constructed to balance the privacy protection level and information loss of the original data, determining the feature vectors and joint feature matrix that minimize the loss function value. In this invention, the joint feature matrix has no unique solution, solving the problem that model parameters can be easily reverse-derived and the original data reconstructed. The constructed loss function addresses the difficulty in balancing privacy protection and information loss in the original data.

[0016] In a further embodiment, the present invention also identifies key features of the original data matrix through a two-layer attention mechanism, filters redundant features such as identifiers, and adjusts the dimensions of the original data matrix by combining feedback on information loss and changes in the degree of privacy protection, thereby achieving an optimal balance between privacy protection and information integrity.

[0017] In a further embodiment, the decision with the highest probability is obtained through probability distribution as the best decision, and the best decision for collaboration among different departments is determined. At the same time, the degree of privacy protection is combined to comprehensively score the decision effect, and the effectiveness of the best decision relative to traditional decision is judged, thereby improving the quality of government collaboration. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a multi-departmental government collaboration method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the various embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosure of the present invention, primarily used to illustrate the embodiments and to explain the operating principles of the embodiments in conjunction with the relevant descriptions in the specification. With reference to these drawings, those skilled in the art should be able to understand other possible implementations and the advantages of the present invention. Components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0020] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0021] Example 1: Figure 1 This is a flowchart illustrating a multi-departmental government collaboration method according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method of this embodiment includes the following steps: Obtain the raw data matrix input from multiple departments participating in government collaboration; In one specific implementation, this embodiment of government collaboration takes the intelligent dispatching scenario of flood emergency supplies as an example. It requires quickly coordinating real-time disaster data (such as population distribution and house damage rate) from the civil affairs department for a township, and inventory data (such as tents, lifeboats, and medicines) from the emergency management department for b material reserve points. Civil Affairs Department Data Matrix n1 is the number of samples, and d1 is the number of features; Social Security Department Data Matrix n2 is the number of samples, and d2 is the number of features; The original data matrix is ​​dimensionality reduced using an encoder to generate feature vectors for each department's original data matrix, where: , In the formula, Z is the feature vector of the department; X is the original data matrix; W e b is the encoder weight matrix; e Here, represents the bias vector; ReLU represents the activation function. This symbol represents the mapping function from the original data matrix X to the virtual feature vector Z. It clearly shows the process of the original data being transformed by the feature editor. In one specific implementation, the dimension of the feature vector after dimensionality reduction ; In one specific implementation, the hash value of the original data matrix is ​​recorded on the blockchain. Once the feature vector is generated, the original data matrix is ​​immediately deleted. Different departments can only access the Z-axis of a specific dimension; for example, the civil affairs department cannot see "logistics timeliness." While reducing data dimensionality, the statistical patterns required for business operations are accurately preserved, and sensitive individual information is completely removed. This achieves the goal of data usability without visibility—that is, retaining only the statistical features needed for cross-departmental collaboration while keeping other sensitive information invisible—in order to achieve the goal of government collaboration.

[0022] In one specific implementation, the original data matrix also identifies key features through a two-layer attention mechanism, including: The local attention weights are calculated using the following formula: , In the formula, For the original data matrix, the first... One feature; x j For the original data matrix, the first... j One characteristic, d The total number of features in the original data matrix; Represents the original data matrix. The local attention weights of each feature, for example, in civil affairs data, the weight of the building damage level is 0.92, while the weight of the administrative division code is 0.05; W a This is the attention weight matrix; The global attention weights are calculated using the following formula: , In the formula, Represents the original data matrix Global attention weights for each feature; Represents the original data matrix. Features In one specific implementation, the cosine similarity to the target decision label Y represents historical decision results or reference standards related to business objectives, used to measure the relevance of current features to business needs. Y, as a historical decision label, can be decision results formed in similar past scenarios, such as material dispatch plans or rescue priorities. Its role is to provide a business-oriented reference for judging feature importance, ensuring that the selected features are closely related to actual government task objectives. This represents the maximum cosine similarity between all features of the original data matrix and the target decision label; In this embodiment, the local attention weight represents the importance of the feature dimension; the global attention weight represents the correlation between the feature and the target label.

[0023] The overall weight is calculated using local attention weights and total attention weights; In one specific implementation, the overall weight is calculated using the following formula: , In the formula, Represents the original data matrix. The combined weight of each feature.

[0024] In one specific implementation, features with a comprehensive weight greater than or equal to a preset comprehensive weight threshold, such as 0.3, are identified as key features of the original data matrix. At the same time, other features, such as direct identifiers including ID card numbers and phone numbers, are automatically filtered to achieve the purpose of de-identification.

[0025] Calculate the KL divergence value to determine the degree of privacy protection of the feature vector; In one specific implementation, the KL divergence value is calculated using the following formula: , , In the formula, Represents the coding distribution Compared with the reference distribution KL divergence function between them; μ It is an eigenvector Z The mean vector; Σ is the eigenvector. Z The covariance matrix; is the preset KL divergence threshold; k is the dimension of the original data matrix; Represents the coding distribution; Indicates the reference distribution; Represents a normal distribution; An n-dimensional feature vector matrix is ​​obtained by concatenating the feature vectors of each department. This feature vector matrix is ​​then combined with a rank-missing matrix with a preset rank of n-1 to obtain a joint feature matrix, where: , In the formula, F Represents a joint feature matrix of multiple departments. These represent virtual feature vectors for n government departments. This indicates a row-by-row concatenation operation, which is a joint matrix formed by horizontally concatenating the feature matrices of different departments. M Let be a rank-deficient matrix, where ; In one specific implementation, when At that time, the solution space of the system of equations has a dimension of 1, requiring at least n linearly independent equations to determine a unique solution, while the attacker can only obtain... The equations are mathematically proven to be unsolvable in reverse, meaning that the processed eigenvector matrix cannot be used to derive the original data matrix.

[0026] In one specific implementation, the joint feature matrix is ​​also verified for invertibility using the following formula: , In the formula, This represents the matrix condition number. The larger the value, the more ill-conditioned the matrix is, and the less stable the inverse solution is. The Jacobian matrix representing the joint characteristic matrix; This represents the maximum singular value of the Jacobian matrix, which measures the matrix's ability to amplify changes in the input in the most sensitive direction. A larger value indicates that a small perturbation in the input in that direction will lead to a significant change in the output. The value represents the minimum singular value of the Jacobian matrix, which measures the matrix's ability to amplify input changes in the least sensitive direction. The smaller the value, the weaker the impact of input changes on the output in that direction. when When the value is greater than the preset value, it indicates that the verification result of the invertibility of the joint feature matrix meets the standard.

[0027] In one specific implementation, before dimensionality reduction by the encoder, the original data matrix is ​​adjusted in dimensionality based on feedback from information loss and changes in privacy protection levels, calculated using the following formula: , In the formula, The dimensions of the original data matrix after adjustment; The dimensions of the original data matrix before adjustment; Indicates the change in information loss; This represents the change in the level of privacy protection. This represents the learning rate.

[0028] In one specific implementation, the initial dimension k(0) of the original data matrix is ​​min(d1,d2) / 2, and the upper limit k max The value is min(d1,d2)×0.6.

[0029] In one specific implementation, the system has a built-in government affairs feature dictionary to ensure that the features after dimensionality reduction are interpretable.

[0030] By constructing a loss function to balance the degree of privacy protection of the original data with information loss, the feature vectors and joint feature matrix that minimize the value of the loss function are determined. In one specific implementation, the loss function is calculated using the following formula: , In the formula, L The loss function;Y Label the target decision; W d The decoder weight matrix is ​​the matrix corresponding to the joint feature matrix; This represents information loss, specifically the error between the decoded data matrix and the output target. λ γ is the privacy protection strength coefficient; γ is the feature sparsity coefficient; Represents the encoder weight matrix W e The L1 norm is calculated as the sum of the absolute values ​​of all elements in the matrix. Its function is to achieve feature sparsity. By constraining the sparsity of the weight matrix, the model is prompted to prioritize the retention of features that are more critical to the task, reducing redundant information. This improves the interpretability and computational efficiency of features while ensuring model performance.

[0031] In one specific implementation, parameters, including the encoder weight matrix, decoder weight matrix, and bias vector, are adjusted using gradient descent, and adjusted according to the following formula: ; In the formula, These are the adjusted parameters; These are the parameters before adjustment; The learning rate; In one specific implementation, performance gain refers to the improvement in task performance, such as the accuracy of resource scheduling, under the current parameter adjustments. It measures the degree to which parameter updates improve task performance. It can be determined by comparing task performance before and after the parameter update, such as prediction error and decision accuracy. When the value is less than 5%, it means that the current parameter adjustment has limited effect on improving the task performance. At this time, the gain coefficient ξ is triggered to participate in the parameter update in order to further optimize the model parameters. This represents the gradient of the loss function; Indicates the gain coefficient; The joint feature matrix is ​​sent to multiple departments involved in government collaboration so that they can make collaborative decisions based on the joint feature matrix.

[0032] In one specific implementation, collaborative decision-making determines the optimal decision based on the joint decision probability distribution of the joint feature matrix, calculated using the following formula: ; In the formula, S is the probability distribution after mapping the joint feature matrix through the decision weight matrix; W s The decision weight matrix has dimensions of . Here, m represents the number of decision categories, and k represents the dimension of the joint feature matrix. In one specific implementation, the softmax function transforms the result of the joint feature matrix F after being mapped by the weight matrix into a probability distribution. Specifically, the softmax function exponentially normalizes and normalizes the input values, ensuring that each element in the output decision vector S is a probability value within the interval [0,1], and the sum of all elements is 1. This characteristic allows S to intuitively reflect the likelihood of different decision schemes. By selecting the element with the highest probability, i.e., S* = arg max (S), the optimal decision scheme, such as the best material scheduling scheme, can be determined, which meets the requirements for decision clarity and executability in government scenarios. The decision with the highest probability is selected as the optimal decision based on the probability distribution.

[0033] In one specific implementation, the optimal decision also incorporates the degree of privacy protection to comprehensively score the decision's effectiveness, calculated using the following formula: , In the formula, Indicates the decision effect score. Indicates the accuracy of the prediction; The time-consuming nature of traditional decision-making; The time taken for optimal decision-making is considered. In one specific implementation, time is included as part of the effectiveness score and is closely related to the target decision. For example, government scenarios such as disaster emergency response and material dispatch have extremely high timeliness requirements. Flood relief needs to meet the golden 72-hour response requirement, and traditional manual coordination or complex calculation methods are difficult to meet the standard due to excessive time consumption. One of the core objectives of this embodiment is to improve the timeliness of emergency response by reducing the computational complexity of cross-departmental collaboration through a federated computing mechanism, thereby achieving rapid response. Therefore, time consumption, as a key indicator for measuring the efficiency of the solution, is included in the comprehensive verification system to fully evaluate the technical effect. It is necessary not only to ensure prediction accuracy and privacy security, but also to verify its advantages in time efficiency, ensuring the feasibility of the solution in actual government scenarios. This represents the KL divergence value. When the decision performance score is greater than the preset score threshold, it indicates that the best decision performance is better than the traditional decision performance.

[0034] Example 2: Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention, as shown below. Figure 2As shown, the device includes a processor 201, a memory 202, a bus 203, and a computer program stored in the memory 202 and executable on the processor 201. The processor 201 includes one or more processing cores. The memory 202 is connected to the processor 201 via the bus 203. The memory 202 is used to store program instructions. When the processor executes the computer program, it implements the steps in the above-described method embodiment of Embodiment 1 of the present invention.

[0035] Furthermore, as an executable solution, the electronic device can be a computer unit, which can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. The computer unit may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above-described structure of the computer unit is merely an example and does not constitute a limitation on the computer unit. It may include more or fewer components, or combine certain components, or use different components. For example, the computer unit may also include input / output devices, network access devices, buses, etc., and this embodiment of the invention does not limit this.

[0036] Furthermore, as an executable solution, the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the computer unit, connecting various parts of the entire computer unit via various interfaces and lines.

[0037] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the computer unit by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0038] Example 3: The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in the embodiments of the present invention.

[0039] If the modules / units integrated in the computer unit are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0040] Example 4: The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-departmental government collaboration method described above.

[0041] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A method for multi-departmental government collaboration, characterized in that, Includes the following steps: Obtain the raw data matrix input from multiple departments participating in government collaboration; The original data matrix is ​​dimensionality reduced using an encoder to generate feature vectors for each department's original data matrix, where: ， In the formula, Z is the feature vector of the department; X is the original data matrix; W e b is the encoder weight matrix; e Here, represents the bias vector; ReLU represents the activation function. This represents the mapping function from the original data matrix X to the virtual eigenvector Z; The privacy protection level of the feature vector is determined by calculating the KL divergence value of the feature vector; the KL divergence value is calculated using the following formula: ， ， In the formula, Represents the coding distribution Compared with the reference distribution KL divergence function between them; μ It is an eigenvector Z The mean vector; Σ is the eigenvector. Z The covariance matrix; is the preset KL divergence threshold; k is the dimension of the original data matrix; Represents the coding distribution; Indicates the reference distribution; Represents a normal distribution; An n-dimensional feature vector matrix is ​​obtained by concatenating the feature vectors of each department. This feature vector matrix is ​​then combined with a rank-deficient matrix of rank n-1 to obtain a joint feature matrix, where: ， In the formula, F Represents a joint feature matrix of multiple departments. These represent virtual feature vectors for n government departments. This indicates a row-by-row concatenation operation, which is a joint matrix formed by horizontally concatenating the feature matrices of different departments. M Let be a rank-deficient matrix, where ; By constructing a loss function to balance the degree of privacy protection of the original data and information loss, the feature vector and joint feature matrix that minimize the value of the loss function are determined. The loss function is calculated using the following formula: ， In the formula, L The loss function; Y Label the target decision; W d The decoder weight matrix is ​​the matrix corresponding to the joint feature matrix; This indicates information loss; λ γ is the privacy protection strength coefficient; γ is the feature sparsity coefficient; Represents the encoder weight matrix W e The L1 norm; The joint feature matrix is ​​sent to multiple departments participating in government collaboration, so that these departments can make collaborative decisions based on the joint feature matrix.

2. The multi-departmental government collaboration method according to claim 1, characterized in that, The original data matrix also identifies key features through a two-layer attention mechanism, including: The local attention weights are calculated using the following formula: ， In the formula, For the original data matrix, the first... One feature; x j For the original data matrix, the first... j One characteristic, d The total number of features in the original data matrix; Represents the original data matrix. Local attention weights for each feature; W a This is the attention weight matrix; The global attention weights are calculated using the following formula: ， In the formula, Represents the original data matrix. Global attention weights for each feature; Represents the original data matrix. Features Cosine similarity to the target decision label Y; This represents the maximum cosine similarity between all features of the original data matrix and the target decision label; A comprehensive weight is calculated using the local attention weights and the total attention weights. Features with a comprehensive weight greater than or equal to a preset comprehensive weight threshold are identified as key features of the original data matrix. The comprehensive weight is calculated using the following formula: ， In the formula, Represents the original data matrix. The combined weight of each feature.

3. The multi-departmental government collaboration method according to claim 1, characterized in that, Before dimensionality reduction using the encoder, the original data matrix is ​​adjusted for dimensionality based on feedback from information loss and changes in privacy protection levels, calculated using the following formula: ， In the formula, The dimensions of the original data matrix after adjustment; The dimensions of the original data matrix before adjustment; Indicates the change in information loss; This represents the change in the level of privacy protection. This represents the learning rate.

4. The multi-departmental government collaboration method according to claim 1, characterized in that, The joint feature matrix is ​​also verified for invertibility using the following formula: ， In the formula, Represents the matrix condition number; The Jacobian matrix representing the joint characteristic matrix; This represents the largest singular value of the Jacobian matrix; Represents the smallest singular value of the Jacobian matrix; when When the value is greater than the preset value, it indicates that the verification result of the invertibility of the joint feature matrix meets the standard.

5. The multi-departmental government collaboration method according to claim 1, characterized in that, The parameters, including the encoder weight matrix, decoder weight matrix, and bias vector, are adjusted using gradient descent and are adjusted according to the following formula: ； In the formula, These are the adjusted parameters; These are the parameters before adjustment; The learning rate; For effect enhancement; This represents the gradient of the loss function; This represents the gain coefficient.

6. The multi-departmental government collaboration method according to claim 1, characterized in that, The collaborative decision-making process determines the optimal decision based on the joint decision probability distribution of the joint feature matrix, calculated using the following formula: ； In the formula, S is the probability distribution after mapping the joint feature matrix through the decision weight matrix; W s This is the decision weight matrix; The decision with the highest probability is selected as the optimal decision based on the probability distribution.

7. The multi-departmental government collaboration method according to claim 6, characterized in that, The optimal decision also incorporates a comprehensive score for the decision-making effectiveness based on the degree of privacy protection, and this comprehensive score is calculated using the following formula: ， In the formula, Indicates the decision effect score. Indicates the accuracy of the prediction; The time-consuming nature of traditional decision-making; The time required for the optimal decision; This represents the KL divergence value. When the decision performance score is greater than a preset score threshold, it indicates that the best decision performance is better than the traditional decision performance.

8. An electronic device, characterized in that, It includes a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the multi-departmental government collaboration method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one program, which is executed by a processor to implement the multi-departmental government collaboration method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-departmental government collaboration method as described in any one of claims 1 to 7.

Images