Efficiency improving method for government and enterprise digital management

Abstract

The invention discloses an efficiency improvement method for government and enterprise digital management, and relates to the technical field of big data analysis, and the method comprises the steps: S1, building a business architecture interaction data synchronization network of each project type of a government and enterprise, marking a transaction log execution process of each project type of the government and enterprise, generating a business architecture interaction dynamic permission network of each project type of the government and enterprise, verifying a dependency relationship between transaction execution processes under the dynamic permission of the business architecture, endowing the transaction execution processes with priorities, and determining initial resource allocation of the transaction execution processes of each project type of the government and enterprise; and establishing a resource allocation income evaluation model of each project type of the government and enterprise, and generating an optimal resource allocation scheme of each project type of the government and enterprise. The method has the advantages that systematicness and adaptability of government and enterprise digital management are remarkably improved, and an overall solution considering compliance and efficiency is provided for complex business scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data analysis, in particular to an efficiency improvement method for digital management of government and enterprises. BACKGROUND

[0002] The current digital management of government and enterprises has the following problems: data island phenomenon is serious, multi-source heterogeneous data is difficult to effectively integrate; the permission management mechanism is rigid and cannot dynamically adapt to business needs; resource allocation relies on manual experience and lacks scientific and quantitative basis; the linear solidification of transaction processing process cannot identify the priority of the critical path; the single-dimensional evaluation system ignores the multi-dimensional correlation between resource input and value output, resulting in low management efficiency, unreasonable resource allocation and other problems. SUMMARY

[0003] To solve the above technical problems, the efficiency improvement method for digital management of government and enterprises is provided, which solves the above problems.

[0004] To achieve the above purpose, the technical scheme adopted by the present application is: The efficiency improvement method for digital management of government and enterprises comprises: S1, obtaining multi-source heterogeneous data of each project type of government and enterprises, establishing a business architecture knowledge graph of each project type of government and enterprises, performing semantic association according to the business architecture of each project type of government and enterprises, and generating a business architecture interaction data synchronization network of each project type of government and enterprises; S2, based on the business architecture interaction data synchronization network of each project type of government and enterprises, marking the transaction log execution process of each project type of government and enterprises, establishing a multi-level dynamic permission allocation model, and generating a business architecture interaction dynamic permission network of each project type of government and enterprises; S3, according to the business architecture interaction dynamic permission network of each project type of government and enterprises, verifying the dependency relationship between each transaction execution process under the business architecture dynamic permission, establishing a transaction execution process priority identification mechanism, assigning priority to the transaction execution process, and determining the initial resource allocation of the transaction execution process of each project type of government and enterprises; S4, screening the correlation state between the value income under the transaction execution process resource allocation of each project type of government and enterprises in history, establishing a resource allocation income evaluation model of each project type of government and enterprises, generating an optimal interval of transaction execution process resource allocation-value income of each project type of government and enterprises, correcting the initial resource allocation of the transaction execution process of each project type of government and enterprises, and obtaining the optimal resource allocation scheme of each project type of government and enterprises.

[0005] Preferably, step S1 specifically comprises: Based on the multi-source heterogeneous data of various project types of government and enterprises, the data is divided into structured data, semi-structured data and unstructured data, and the multi-head attention mechanism decoder is used to match the corresponding data types to generate a unified spatio-temporal reference data set of various project types of government and enterprises; According to the unified spatio-temporal reference data set of various project types of government and enterprises, the BiLSTM-CRF joint relation extraction model is substituted, the context semantic information of each structure data unit time in the unified spatio-temporal reference data of various project types of government and enterprises is marked by BiLSTM, and the context semantic information of each structure data unit time is input into CRF, the conditional probability of the entity / dependency relation label corresponding to the context semantic information of each structure data unit time is calculated, the data entity-relation probability distribution of various project types of government and enterprises is determined, and the method is as follows: ; Among them, is the conditional probability of the entity / dependency relation label when the unified spatio-temporal reference data of various project types of government and enterprises is given, is a binary variable of the entity / dependency relation label, is the context semantic information of each structure data in the unified spatio-temporal reference data of various project types of government and enterprises, is an entity / dependency relation label weight matrix, is the entity / dependency relation label offset item from the tth unit time to the t+1th unit time, is the entity / dependency relation label offset item from the tth unit time to the t+1th unit time, is all possible entity / dependency relation label sequences when the unified spatio-temporal reference data of various project types of government and enterprises.

[0006] Preferably, step S1 further comprises: According to the data entity-relation probability distribution of various project types of government and enterprises, according to the business architecture level of various project types of government and enterprises, the data entity-relation probability distribution corresponding to the business architecture level is substituted into the GNN graph neural network, the business architecture level is taken as a global entity, the data entity of the corresponding type of the government project is taken as an internal sub-node, and the data dependency relation of the corresponding type of the government project is taken as an edge weight, and a data self-synchronization network of the business architecture of various project types of government and enterprises is established. Based on the Jaccard-EditDistance mixed similarity, the Jaccard similarity algorithm is used to quantify the text character similarity of the data self-synchronization network internal sub-nodes of the business architecture of each project type of the government and enterprise, the EditDistance edit distance is used to quantify the text character operation change similarity of the data self-synchronization network internal sub-nodes of the business architecture of each project type of the government and enterprise, the mixed similarity between the internal sub-nodes of the data self-synchronization network of the business architecture of each project type of the government and enterprise is determined, and the business architecture interaction data synchronization network of each project type of the government and enterprise is established, as follows: ; Among them, is the mixed similarity between the kth sub-node in the data self-synchronization network of the business architecture of the ith project type of the government and enterprise and the k+1th sub-node in the data self-synchronization network of the business architecture of the ith project type of the government and enterprise, is the weight coefficient, is the text character similarity of the kth sub-node in the data self-synchronization network of the business architecture of the ith project type of the government and enterprise, is the text character operation change similarity of the kth sub-node in the data self-synchronization network of the business architecture of the ith project type of the government and enterprise.

[0007] Preferably, step S2 specifically comprises: Marking the transaction log execution process of each project type of the government and enterprise as the key value attribute feature of the internal node in the business architecture interaction data synchronization network of each project type of the government and enterprise, and obtaining the sub-node transaction process sequence in the business architecture interaction data synchronization network of each project type of the government and enterprise; According to the business architecture reality constraints of each project type of the government and enterprise, the transaction process state transition restriction condition is established; Based on the Lstm long short-term memory network, a multi-level dynamic permission allocation model is established, and according to the transaction process state transition restriction condition, the interaction path constraint between the sub-nodes of the business architecture interaction dynamic permission network of each project type of the government and enterprise is given, and the sub-node transaction process sequence in the business architecture interaction data synchronization network of each project type of the government and enterprise is taken as the input, and the business architecture interaction data synchronization constraint network of each project type of the government and enterprise is taken as the output; Using NMTF non-negative matrix triple decomposition, the dependence relationship strength of the transaction process pointing edge of the internal sub-node in the business architecture interaction data synchronization constraint network of each project type of the government and enterprise is verified; The CVAE conditional variational autoencoder is used to verify the transaction process context information of internal sub-nodes in the business architecture interaction data synchronization constraint network of various government and enterprise project types. The dependency strength of the transaction process pointing to the edge of the internal sub-node is used as the latent vector. The reconstruction error and KL divergence are used to form a permission allocation loss function to minimize the error between the reconstruction error permission allocation and the actual allocation of the permission allocation loss function and the difference between the generated distribution and the prior distribution of the KL divergence. Dynamic permissions are given to the sub-node transaction process sequences in the business architecture interaction data synchronization network of various government and enterprise project types, and a dynamic permission network for business architecture interaction of various government and enterprise project types is obtained.

[0008] Preferably, step S3 specifically includes: Based on the interactive dynamic permission network of the business architecture of various government and enterprise project types, the transaction process and its dynamic permission tags in the business architecture of various government and enterprise project types are obtained; According to recursive causal discovery, the causal relationship between the transaction processes of internal sub-nodes and their dynamic permission labels in the dynamic permission network of the business architecture interaction of various government and enterprise project types is calculated, and the dependency paths between the transaction processes of internal sub-nodes are determined to obtain the dynamic permission dependency relationship network of the business architecture interaction of various government and enterprise project types; Based on the business architecture interactive dynamic permission dependency network of various government and enterprise project types, the transaction process request open metadata of internal sub-nodes under the business architecture interactive dynamic permission is obtained and normalized to obtain the transaction process request feature data of the internal sub-nodes; Using the analytic hierarchy process, we synchronously constrain the dependency strength of transaction process-directed edges of internal sub-nodes in the network according to the business architecture interaction data of various government and enterprise project types, and assign weights to the transaction process request feature data of internal sub-nodes. Based on the weight of the transaction process request feature data of the internal sub-nodes and the transaction process request feature data of the internal sub-nodes, the transaction process priority of the internal sub-nodes in the business architecture interaction data synchronization constraint network of various government and enterprise project types is determined.

[0009] Preferably, step S3 further includes: Establish resource pool constraints for each type of government and enterprise project based on the total amount of resources that can be allocated per unit time in the business architecture of each type of government and enterprise project; Based on the business architecture interaction data of various government and enterprise project types, the transaction process priority and dynamic permission labels of internal sub-nodes in the network are synchronously constrained to form the transaction process priority vector of internal sub-nodes; Normalize the transaction process priority vector of the internal child nodes; Determine the request resource requirement range of the transaction process of the internal sub-node in the business architecture interaction data synchronization constraint network of each project type of government and enterprise; Maximize the allocation of the resource pool of each project type of government and enterprise to meet the request resource requirement range interval of the transaction process of the internal sub-node in the business architecture interaction data synchronization constraint network of each project type of government and enterprise as the objective function, the resource pool constraint of each project type of government and enterprise as the total constraint, and the normalized value of the transaction process priority vector of the internal sub-node as the resource allocation decision variable, to determine the initial resource allocation of the transaction execution process of each project type of government and enterprise.

[0010] Preferably, step S4 specifically includes: Obtain the historical resource allocation of the transaction execution process of each project type of government and enterprise; Benefit from the Bayesian network, verify the posterior probability of the corresponding value income influence of the transaction process request feature data of the internal sub-node under the historical resource allocation of the transaction execution process of each project type of government and enterprise, and determine the resource allocation-value income influence factor correlation matrix of the transaction execution process of each project type of government and enterprise; Using PCA principal component analysis, the resource allocation-value income influence factor correlation matrix of the transaction execution process of each project type of government and enterprise is processed by dimension reduction to obtain the resource allocation-value income influence factor correlation dimension reduction matrix of the transaction execution process of each project type of government and enterprise; Based on random forest, the resource allocation-value income influence factor correlation dimension reduction matrix of the transaction execution process of each project type of government and enterprise is taken as the root node, the resource allocation-value income influence factor of the transaction execution process is taken as the branch node, and the resource allocation-value income optimal interval of the transaction execution process is taken as the leaf node. A resource allocation income evaluation model of each project type of government and enterprise is established; The initial resource of the transaction execution process of each project type of government and enterprise is taken as the input of the resource allocation income evaluation model of each project type of government and enterprise, the constraint-sensitive split quality function is taken as the split criterion, and the optimal resource allocation scheme of each project type of government and enterprise is generated as the output; Wherein, the constraint-sensitive split quality function is specifically: ; Wherein, is a binary variable of the maximum income purity item and the resource constraint satisfaction item, is the value corresponding to the transaction process request feature data of the kth internal sub-node under the resource allocation of the transaction execution process of the ith project type of government and enterprise, is the average income of the transaction execution process of each project type of government and enterprise, is the resource allocation-value income influence factor correlation dimension reduction matrix of the transaction execution process of each project type of government and enterprise, a resource request vector for the execution process of the affairs of each project type of the government and enterprise, a total amount of resource pools for each project type of the government and enterprise, a resource allocation constraint parameter, a total number of internal child nodes.

[0011] Compared with the prior art, the present application has the following beneficial effects: The present application proposes an efficiency improvement scheme for digital management of government and enterprise, realizes deep fusion of multi-source data by constructing a business architecture knowledge graph, and breaks the information island; realizes intelligent real-time permission adjustment based on a dynamic permission network, and improves security compliance; establishes a scientific resource allocation mechanism through transaction priority identification and causal analysis, and optimizes key business processes; finally, a closed-loop optimization system is formed by combining a value benefit evaluation model, so that resource allocation not only meets the efficiency demand but also maximizes business value. The whole scheme realizes intelligent management of the whole link from data governance to decision optimization, significantly improves the systematicness and adaptability of digital management of government and enterprise, and provides an overall solution that takes into account compliance and efficiency for complex business scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A flowchart of the efficiency improvement method for digital management of government and enterprise. DETAILED DESCRIPTION

[0013] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only used as examples, and other obvious modifications can be thought of by those skilled in the art.

[0014] Referring to Figure 1 The efficiency improvement method for digital management of government and enterprise includes: S1, acquiring multi-source heterogeneous data of each project type of the government and enterprise, establishing a business architecture knowledge graph of each project type of the government and enterprise, performing semantic association according to the business architecture of each project type of the government and enterprise, and generating a business architecture interaction data synchronization network of each project type of the government and enterprise; Step S1 specifically includes: Based on the multi-source heterogeneous data of each project type of the government and enterprise, the structured data, semi-structured data and unstructured data are divided, the corresponding data type is matched by using a multi-head attention mechanism decoder, and a unified space-time reference data set of each project type of the government and enterprise is generated; the structured data, semi-structured data and unstructured data include: structured data (database table, Excel) of each project type of the government and enterprise, structured data (JSON / XML log) of each project type of the government and enterprise, and unstructured data (contract text, meeting minutes) of each project type of the government and enterprise According to the unified space-time reference data set of each project type of government and enterprise, the BiLSTM-CRF combined relation extraction model is substituted, the context semantic information of each structural data unit time in the unified space-time reference data of each project type of government and enterprise is marked by BiLSTM, the conditional probability of the context semantic information of each structural data unit time corresponding to the entity / dependency relation label is calculated by inputting CRF, the data entity-relation probability distribution of each project type of government and enterprise is determined, and the mode is as follows: ; Among them, is the conditional probability of the entity / dependency relation label at a given unified space-time reference data of each project type of government and enterprise, is a binary variable of the entity / dependency relation label, is the context semantic information of the tth unit time of each structural data in the unified space-time reference data of each project type of government and enterprise, is the entity / dependency relation label weight matrix, is the entity / dependency relation label offset item of the tth unit time to the t+1th unit time, is all possible entity / dependency relation label sequences of the unified space-time reference data of each project type of government and enterprise.

[0015] Step S1 also includes: According to the data entity-relation probability distribution of each project type of government and enterprise, according to the business architecture level of each project type of government and enterprise, the data entity-relation probability distribution corresponding to the business architecture level is substituted into the GNN graph neural network, the business architecture level is taken as a global entity, the data entity of the corresponding type of the government project is taken as an internal subnode, and the data dependency relation of the corresponding type of the government project is taken as an edge weight, to establish a data self-synchronization network of the business architecture of each project type of government and enterprise; Based on the Jaccard-EditDistance mixed similarity, the text character similarity of the internal subnode of the data self-synchronization network of the business architecture of each project type of government and enterprise is quantified by using the Jaccard similarity algorithm, the text character operation change similarity of the internal subnode of the data self-synchronization network of the business architecture of each project type of government and enterprise is quantified by using the EditDistance editing distance, the mixed similarity of the internal subnode of the data self-synchronization network of the business architecture of each project type of government and enterprise is determined, and the interactive data synchronization network of the business architecture of each project type of government and enterprise is established, and the mode is as follows: ; Among them, ​A mixed similarity between the kth sub-node and the k+1th sub-node within the data self-synchronization network of the business architecture of the ith project type of the government and enterprise, is a weight coefficient, is a text character similarity of the kth sub-node within the data self-synchronization network of the business architecture of the ith project type of the government and enterprise, is a text character operation change similarity of the kth sub-node within the data self-synchronization network of the business architecture of the ith project type of the government and enterprise.

[0016] In use, in combination with the contents in the above steps: As further contents, the scheme aligns the multi-source heterogeneous data of the government and enterprise through the multi-head attention mechanism, uses the BiLSTM-CRF model to extract entity relationships with high precision and calculate the probability distribution, constructs a business architecture knowledge graph based on the GNN graph neural network to realize dynamic synchronization, and realizes cross-system data consistency verification in combination with the Jaccard-EditDistance mixed similarity. The core point lies in that the automatic fusion of structured / unstructured data is realized, manual intervention is reduced, the entity relationship recognition accuracy is improved compared with traditional methods, the dynamic response speed of the business architecture is optimized from the hour level to the minute level, and the cross-system field matching error rate is reduced; Exemplary: Scenario: Digital management of “infrastructure project approval” of a certain municipal finance bureau Input data: structured: project budget table in Oracle database, semi-structured: JSON log of bidding system, and unstructured: scanned PDF version of contract; S1 execution process: Data unification: the multi-head attention model maps “budget amount” (Excel), “bid_amount” (JSON), and “contract total price” (PDF) to the unified field “project_budget”.

[0017] Knowledge extraction: BiLSTM-CRF extracts the triple 〈municipal finance bureau, reply, infrastructure project A〉 from the contract text with a confidence of 0.92.

[0018] Graph construction: GNN associates the “infrastructure approval” business architecture with “municipal finance bureau”, “project A”, “budget”, and the like, and the edge weight reflects the dependence strength of the approval process.

[0019] Dynamic synchronization: it is detected that the “bid_amount=1.2 billion” in the JSON log has a similarity of 0.98 with the “total price one hundred million two thousand yuan” in the contract text, and data correction is automatically triggered.

[0020] Output: Generate a dynamically updated "infrastructure project approval" business architecture synchronization network to support real-time data collaboration across departments.

[0021] S2, based on the business architecture interaction data synchronization network of each project type of government and enterprise, mark the transaction log execution process of each project type of government and enterprise, establish a multi-level dynamic permission allocation model, and generate a business architecture interaction dynamic permission network of each project type of government and enterprise; Step S2 specifically includes: Mark the transaction log execution process of each project type of government and enterprise as the key value attribute feature of the internal node in the business architecture interaction data synchronization network of each project type of government and enterprise, and obtain the transaction process sequence of the subnode in the business architecture interaction data synchronization network of each project type of government and enterprise; According to the business architecture reality constraints of each project type of government and enterprise, establish transaction process state transition restriction conditions; As further content, business architecture reality constraints are exemplified as: the procurement department can only proceed with the procurement transaction process after the financial approval is completed, and the corresponding transaction process state transition restriction condition is: financial approval completion triggers procurement execution; Based on the Lstm long short-term memory network, a multi-level dynamic permission allocation model is established, and according to the transaction process state transition restriction condition, the interaction path constraint between the subnodes of the business architecture interaction dynamic permission network of each project type of government and enterprise is given, and the transaction process sequence of the subnode in the business architecture interaction data synchronization network of each project type of government and enterprise is taken as the input, and the business architecture interaction data synchronization constraint network of each project type of government and enterprise is taken as the output; Use NMTF non-negative matrix triple decomposition to verify the dependency strength of the transaction process directed edge of the internal subnode in the business architecture interaction data synchronization constraint network of each project type of government and enterprise; Use CVAE conditional variational autoencoder to verify the transaction process context information of the internal subnode in the business architecture interaction data synchronization constraint network of each project type of government and enterprise, take the dependency strength of the transaction process directed edge of the internal subnode as the latent vector, use the reconstruction error and KL divergence, and assemble the permission allocation loss function, minimize the reconstruction error between the reconstruction and the actual allocation, and the difference between the generated distribution and the prior distribution of the KL divergence, give the dynamic permission of the transaction process sequence of the subnode in the business architecture interaction data synchronization network of each project type of government and enterprise, and obtain the business architecture interaction dynamic permission network of each project type of government and enterprise.

[0022] In use, the contents in the above steps are combined, As a further content, the scheme marks the transaction log as a key value attribute of the business architecture knowledge graph, models the process state transition with constraints using LSTM state machine, mines the internal node-permission relationship in the network architecture using non-negative matrix tri-factorization (NMTF), and generates context-aware dynamic permission allocation using conditional variational autoencoder (CVAE), and finally constructs a dynamic permission network that is highly consistent with business rules. The core point is to realize the temporal and spatial dynamics of permissions: the time dimension ensures operation time sequence compliance through the LSTM memory unit, the spatial dimension optimizes cross-department permission granularity through NMTF, and CVAE realizes adaptive security policy through hidden vector quantization of business risk. Compared with the traditional RBAC model, this scheme can reduce permission redundancy and improve the rule violation interception rate, and is especially suitable for real-time permission adjustment and strict compliance requirements in government and enterprise digital management scenarios such as government procurement and cross-department collaboration projects, which significantly reduces the cost of manual strategy maintenance while improving security.

[0023] Exemplary: Scenario, business architecture: financial approval → procurement tender → contract signing → acceptance of performance Transaction log example: json {"process ID": "P-2025-001", "department": "finance department", "operation": "budget approval", "status": "completed", "time": "2025-05-10T14:30:00"}; Permission network generation process: 1. Feature labeling: map the log to the dynamic attributes of the knowledge graph node "financial approval": {"status": "completed", "operator": "Li XX"}; 2. State transition constraints: set {{financial approval}, {procurement tender}} = 1 in the rule matrix, and other procurement-related transitions are initially set to 0; 3. Dynamic permission allocation: NMTF decomposition finds that the "tender officer" role needs to be associated with both the finance department (view budget) and the procurement office (publish tender). CVAE generates the permissions for this role: {"operable processes": ["procurement tender"], "data access range": ["budget amount"]}.

[0024] Conflict detection: when the same user applies for "tender execution" and "budget modification" permissions at the same time, manual review is triggered due to KL divergence exceeding the threshold.

[0025] S3, according to the interactive dynamic permission network of each project type of government and enterprise business architecture, verify the dependency relationship between each transaction execution process under the dynamic permission of the business architecture, establish a transaction execution process priority identification mechanism, assign priority to the transaction execution process, and determine the initial resource allocation of the transaction execution process of each project type of government and enterprise; The step S3 specifically comprises: Based on the business architecture interaction dynamic permission network of each project type of government and enterprise, the transaction processes and their dynamic permission labels in the business architecture of each project type of government and enterprise are obtained; According to the recursive causal discovery, the causal relationship between the transaction processes and their dynamic permission labels of the internal nodes in the business architecture interaction dynamic permission network of each project type of government and enterprise is calculated, the dependent path between the transaction processes of the internal nodes is determined, and the business architecture interaction dynamic permission dependency relationship network of each project type of government and enterprise is obtained; Based on the business architecture interaction dynamic permission dependency relationship network of each project type of government and enterprise, the transaction process request open metadata of the internal nodes under the business architecture interaction dynamic permission is normalized, and the transaction process request feature data of the internal nodes is obtained; Using the analytic hierarchy process, the transaction process request feature data of the internal nodes is given weight according to the dependency relationship strength of the transaction process directed edge of the internal nodes in the business architecture interaction data synchronization constraint network of each project type of government and enterprise; According to the transaction process request feature data weight of the internal nodes and the transaction process request feature data of the internal nodes, the transaction process priority of the internal nodes in the business architecture interaction data synchronization constraint network of each project type of government and enterprise is determined.

[0026] The step S3 further comprises: Based on the total amount of allocable resources under the business architecture of each project type of government and enterprise, the resource pool constraint of each project type of government and enterprise is established; According to the transaction process priority vector of the internal nodes in the business architecture interaction data synchronization constraint network of each project type of government and enterprise, the transaction process priority vector of the internal nodes is normalized; The transaction process priority vector of the internal nodes is normalized; The request resource demand range of the transaction process of the internal nodes in the business architecture interaction data synchronization constraint network of each project type of government and enterprise is determined; The maximum allocation of the resource pool of each project type of government and enterprise satisfies the request resource demand range interval of the transaction process of the internal nodes in the business architecture interaction data synchronization constraint network of each project type of government and enterprise as the objective function, the resource pool constraint of each project type of government and enterprise as the total constraint, and the normalized value of the transaction process priority vector of the internal nodes as the resource division decision variable, and the initial resource allocation of the transaction execution process of each project type of government and enterprise is determined; In use, the contents in the above steps are combined, As a further content, the transaction process dependency relationship is constructed by the dynamic permission network and the causal discovery, the multi-dimensional priority features (such as permission level, deadline) are quantified by combining the analytic hierarchy process (AHP), and the linear programming model is established based on the resource pool constraint to realize the optimal allocation of the initial resources. The technical core lies in that the recursive causal analysis ensures the business process compliance, the AHP weighting enhances the decision interpretability, and the constraint optimization improves the resource utilization rate. The strong binding of the permission label and the dependency path eliminates the overstepping operation; the priority-driven allocation improves the high-value transaction resource satisfaction rate; the dynamic normalization mechanism responds to the permission changes in real time; the explicit weight matrix and the causal diagram provide transparent basis for auditing, guaranteeing the strict compliance requirements and the resource allocation efficiency of the scene. S4, screen the correlation state between the value income of the transaction execution process resource allocation of each project type of the government and enterprise, establish a resource allocation income evaluation model of each project type of the government and enterprise, generate an optimal interval of resource allocation-value income of the transaction execution process of each project type of the government and enterprise, correct the initial resource allocation of the transaction execution process of each project type of the government and enterprise, and obtain an optimal resource allocation scheme of each project type of the government and enterprise; Step S4 specifically includes: Obtaining the transaction execution process resource allocation of each project type of the government and enterprise; Benefiting from the Bayesian network, verifying the posterior probability of the influence of the transaction process request feature data of the internal sub-node on the value income under the transaction execution process resource allocation of each project type of the government and enterprise, and determining the resource allocation-value income influence factor correlation matrix of the transaction execution process of each project type of the government and enterprise; Using the PCA principal component analysis method, performing dimension reduction processing on the resource allocation-value income influence factor correlation matrix of the transaction execution process of each project type of the government and enterprise, and obtaining a resource allocation-value income influence factor correlation reduced matrix of the transaction execution process of each project type of the government and enterprise; Based on the random forest, taking the resource allocation-value income influence factor correlation reduced matrix of the transaction execution process of each project type of the government and enterprise as the root node, taking the resource allocation-value income influence factor of the transaction execution process as the branch node, and taking the optimal interval of the resource allocation-value income of the transaction execution process as the leaf node, a resource allocation income evaluation model of each project type of the government and enterprise is established; Taking the initial resource of the transaction execution process of each project type of the government and enterprise as the input of the resource allocation income evaluation model of each project type of the government and enterprise, and taking the constraint-sensitive split quality function as the split criterion, an optimal resource allocation scheme of each project type of the government and enterprise is generated as the output; The constraint-sensitive split quality function is specifically as follows: ; Wherein, A binary variable for maximizing the profit purity item and the resource constraint satisfaction item, A value corresponding to the transaction process request feature data of the kth internal sub-node of the ith project type of the historical government-enterprise transaction execution process under the resource allocation of the transaction execution process, An average profit of the transaction execution process of each project type of the historical government-enterprise, A resource allocation-value profit influence factor correlation dimension reduction matrix of the transaction execution process of each project type of the government-enterprise, A resource request vector of the transaction execution process of each project type of the government-enterprise, A total amount of resource pool of each project type of the government-enterprise, A resource allocation constraint parameter (assigned according to the priority of the transaction execution process), A total number of internal sub-nodes.

[0027] In use, the above-mentioned contents in the steps are combined: As further content, the causal relationship between historical resource allocation and value profit is mined by a Bayesian network, the core influence factors are extracted by PCA dimension reduction, and a resource-profit optimal interval is generated by random forest modeling. Finally, the initial allocation scheme is corrected based on the constraint sensitivity criterion. The technical core lies in that causal reasoning ensures the scientificity of decision-making, dimension reduction processing improves the generalization ability of the model, and interval output enhances the management flexibility. By excluding subjective bias through causal correlation, the model after dimension reduction has stronger anti-noise performance, the optimal interval provides flexible space for resource adjustment, and the whole-process automation ensures the efficiency of resource division.

[0028] The above shows and describes the basic principles, main features and advantages of the present application. It should be understood by those skilled in the art that the present application is not limited by the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection claimed by the present application is defined by the appended claims and their equivalents.

Claims

1. An efficiency improvement method for digital management of government and enterprises, characterized by: include: S1. Acquire multi-source heterogeneous data of various government and enterprise project types, establish a business architecture knowledge graph for each type of government and enterprise project, perform semantic association based on the business architecture of each type of government and enterprise project, and generate a business architecture interaction data synchronization network for each type of government and enterprise project; S2. Based on the business architecture interaction data synchronization network of various government and enterprise project types, mark the transaction log execution process of various government and enterprise project types, establish a multi-level dynamic permission allocation model, and generate a business architecture interaction dynamic permission network for various government and enterprise project types; S3. Based on the interactive dynamic permission network of the business architecture for each type of government and enterprise project, verify the dependencies between each transaction execution process under the dynamic permission of the business architecture, establish a transaction execution process priority identification mechanism, assign transaction execution processes priorities, and determine the initial resource allocation for transaction execution processes for each type of government and enterprise project; S4. Filter the correlation between the value and benefit under the resource allocation of transaction execution processes of various types of government and enterprise projects in history, establish a resource allocation and benefit evaluation model for various types of government and enterprise projects, generate the optimal interval of resource allocation-value benefit for transaction execution processes of various types of government and enterprise projects, and revise the initial resource allocation of transaction execution processes of various types of government and enterprise projects to obtain the optimal resource allocation plan for various types of government and enterprise projects.

2. The efficiency improvement method for digital management of government and enterprises according to claim 1 is characterized in that: Step S1 specifically includes: Based on the multi-source heterogeneous data of various government and enterprise project types, the data is divided into structured data, semi-structured data and unstructured data, and a multi-head attention mechanism decoder is used to match the corresponding data types to generate a unified spatiotemporal benchmark data set for various government and enterprise project types; Based on the unified spatiotemporal benchmark data set for various government and enterprise project types, we substituted it into the BiLSTM-CRF joint relationship extraction model. We used BiLSTM to label the contextual semantic information of each structural data unit time in the unified spatiotemporal benchmark data for various government and enterprise project types, and input it into the CRF. We then calculated the conditional probability of the contextual semantic information corresponding to the entity / dependency relationship label for each structural data unit time, and determined the data entity-relationship probability distribution for various government and enterprise project types as follows: ； in, is the conditional probability of entity / dependency labels given the unified spatiotemporal benchmark data of various government and enterprise project types, is a binary variable of entity / dependency label, It is the contextual semantic information of each structural data in the unified spatiotemporal benchmark data of various project types of government and enterprises under the t-th unit time. is the entity / dependency label weight matrix, The entity / dependency label under the t-th unit time is transferred to the Entity / dependency label offset term per unit time, All possible entity / dependency label sequences when providing unified spatiotemporal benchmark data for various government and enterprise project types.

3. The efficiency improvement method for digital management of government and enterprises according to claim 2 is characterized in that: Step S1 further includes: According to the data entity-relationship probability distribution of each type of government and enterprise project, and according to the business architecture level of each type of government and enterprise project, the data entity-relationship probability distribution of the corresponding business architecture level is substituted into the GNN graph neural network. With the business architecture level as the global entity, the data entities of the corresponding types of government and enterprise projects as the internal sub-nodes, and the data dependency relationships of the corresponding types of government and enterprise projects as the edge weights, a data self-synchronization network for the business architecture of each type of government and enterprise project is established; Based on the Jaccard-EditDistance hybrid similarity, the Jaccard similarity algorithm is used to quantify the text character similarity of the sub-nodes in the data self-synchronization network of the business architecture of each project type of government and enterprise. The EditDistance edit distance is used to quantify the similarity of text character operation changes of the sub-nodes in the data self-synchronization network of the business architecture of each project type of government and enterprise. The hybrid similarity between the sub-nodes across the business architecture of the data self-synchronization network of the business architecture of each project type of government and enterprise is determined, and an interactive data synchronization network for the business architecture of each project type of government and enterprise is established. The method is as follows: ； in, is the mixed similarity between the kth child node in the data self-synchronization network of the business architecture of the i-th project type of government and enterprise and the k+1th child node in the data self-synchronization network of the business architecture of the i-th project type of government and enterprise, is the weight coefficient, is the text character similarity of the kth child node in the data self-synchronization network of the business architecture of the i-th project type of the government and enterprise, The similarity of text character operation changes of the kth child node in the data self-synchronization network of the business architecture of the i-th project type of the government and enterprise.

4. The efficiency improvement method for digital management of government and enterprises according to claim 3 is characterized in that: Step S2 specifically includes: Mark the transaction log execution process of each type of government and enterprise project as the key-value attribute feature of the internal node in the business architecture interactive data synchronization network of each type of government and enterprise project, and obtain the sub-node transaction process sequence in the business architecture interactive data synchronization network of each type of government and enterprise project; Establish transaction process state transfer restrictions based on the actual business architecture constraints of various government and enterprise project types; Based on the LSTM long short-term memory network, a multi-level dynamic permission allocation model is established. According to the transaction process state transfer constraints, the interaction path constraints between the sub-nodes of the business architecture interaction dynamic permission network of each type of government and enterprise project are given. The sub-node transaction process sequence in the business architecture interaction data synchronization network of each type of government and enterprise project is used as input, and the business architecture interaction data synchronization constraint network of each type of government and enterprise project is used as output. Using NMTF non-negative matrix tri-factorization, we verify the dependency strength of transaction process-directed edges of internal sub-nodes in the business architecture interaction data synchronization constraint network for various government and enterprise project types. The CVAE conditional variational autoencoder is used to verify the transaction process context information of internal sub-nodes in the business architecture interaction data synchronization constraint network of various government and enterprise project types. The dependency strength of the transaction process pointing to the edge of the internal sub-node is used as the latent vector. The reconstruction error and KL divergence are used to form a permission allocation loss function to minimize the error between the reconstruction error permission allocation and the actual allocation of the permission allocation loss function and the difference between the generated distribution and the prior distribution of the KL divergence. Dynamic permissions are given to the sub-node transaction process sequences in the business architecture interaction data synchronization network of various government and enterprise project types, and a dynamic permission network for business architecture interaction of various government and enterprise project types is obtained.

5. The efficiency improvement method for digital management of government and enterprises according to claim 4 is characterized in that: Step S3 specifically includes: Based on the interactive dynamic permission network of the business architecture of various government and enterprise project types, the transaction process and its dynamic permission tags in the business architecture of various government and enterprise project types are obtained; According to recursive causal discovery, the causal relationship between the transaction processes of internal sub-nodes and their dynamic permission labels in the dynamic permission network of the business architecture interaction of various government and enterprise project types is calculated, and the dependency paths between the transaction processes of internal sub-nodes are determined to obtain the dynamic permission dependency relationship network of the business architecture interaction of various government and enterprise project types; Based on the business architecture interactive dynamic permission dependency network of various government and enterprise project types, the transaction process request open metadata of internal sub-nodes under the business architecture interactive dynamic permission is obtained and normalized to obtain the transaction process request feature data of the internal sub-nodes; Using the analytic hierarchy process, we synchronously constrain the dependency strength of transaction process-directed edges of internal sub-nodes in the network according to the business architecture interaction data of various government and enterprise project types, and assign weights to the transaction process request feature data of internal sub-nodes. Based on the weight of the transaction process request feature data of the internal sub-nodes and the transaction process request feature data of the internal sub-nodes, the transaction process priority of the internal sub-nodes in the business architecture interaction data synchronization constraint network of various government and enterprise project types is determined.

6. The efficiency improvement method for digital management of government and enterprises according to claim 5 is characterized in that: Step S3 further includes: Establish resource pool constraints for each type of government and enterprise project based on the total amount of resources that can be allocated per unit time in the business architecture of each type of government and enterprise project; Based on the business architecture interaction data of various government and enterprise project types, the transaction process priority and dynamic permission labels of internal sub-nodes in the network are synchronously constrained to form the transaction process priority vector of internal sub-nodes; Normalize the transaction process priority vector of the internal child nodes; Determine the resource requirements for transaction processes within sub-nodes within the business architecture interaction data synchronization constraint network for each type of government and enterprise project; The objective function is to maximize the allocation of resource pools for each type of government and enterprise project to meet the request resource demand range of the transaction process of the internal sub-nodes in the business architecture interaction data synchronization constraint network of each type of government and enterprise project. The resource pool constraint of each type of government and enterprise project is used as the total amount constraint. The normalized value of the transaction process priority vector of the internal sub-node is used as the resource allocation decision variable to determine the initial resource allocation of the transaction execution process of each type of government and enterprise project.

7. The efficiency improvement method for digital management of government and enterprises according to claim 6 is characterized in that: Step S4 specifically includes: Obtain historical transaction execution process resource allocation for various project types of government and enterprises; Using Bayesian networks, we can verify the posterior probability of the value-benefit impact of transaction process request feature data on internal subnodes under the historical transaction execution process resource allocation of various government and enterprise project types, and determine the resource allocation-value-benefit influencing factor association matrix for transaction execution processes of various government and enterprise project types. Using the PCA principal component analysis method, we perform dimensionality reduction on the resource allocation-value benefit influencing factor correlation matrix of the transaction execution process of various government and enterprise project types, and obtain the resource allocation-value benefit influencing factor correlation dimensionality reduction matrix of the transaction execution process of various government and enterprise project types; Based on random forests, we establish a resource allocation benefit evaluation model for various types of government and enterprise projects, using the associated dimensionality reduction matrix of the resource allocation-value benefit influencing factors of the transaction execution process as the root node, the resource allocation-value benefit influencing factors of the transaction execution process as the branch nodes, and the optimal resource allocation-value benefit interval of the transaction execution process as the leaf node. The initial resources of the transaction execution process of each type of government and enterprise project are used as the input of the resource allocation benefit evaluation model for each type of government and enterprise project, and the constraint-sensitive split quality function is used as the splitting criterion to generate the optimal resource allocation plan for each type of government and enterprise project as the output; Among them, the constraint-sensitive splitting quality function is specifically: ； in, are binary variables of the income purity maximization term and the resource constraint satisfaction term, The corresponding value of the transaction process request feature data of the kth internal child node under the transaction execution process resource allocation of the i-th project type in the historical government and enterprise, The average revenue of transaction execution processes of various types of government and enterprise projects in history. It is the resource allocation-value benefit influencing factor correlation dimension reduction matrix of the transaction execution process of various project types of government and enterprises. It is the resource request vector of the transaction execution process of each project type of government and enterprise. The total amount of resource pools for various types of government and enterprise projects. Assign constraint parameters to resources, is the total number of internal child nodes.

Images