Task disassembly and federation allocation method and system based on multi-level knowledge graph

Abstract

The invention relates to the technical field of artificial intelligence and organization management, and discloses a task disassembly and federation allocation method and system based on a multi-level knowledge graph, and the method comprises the steps: converting an unstructured to-be-processed document into a structured task carrier for task allocation through employing a semantic analysis means, and synchronously extracting key information; constructing an organization unit and member individual dimension double-layer knowledge graph, and realizing association through a cross-dimension association algorithm; combining privacy protection and a federated learning mechanism, calculating a task and organization unit right and responsibility association degree to match the organization units, locally quantifying member capabilities by the organization units, encrypting the member capabilities, and aggregating data to generate a task list; according to member task execution feedback data, the task allocation weight is dynamically adjusted, and the overall load balance of the system is guaranteed. According to the method, an analysis-modeling-allocation-optimization closed loop is integrally formed, the task allocation accuracy and the system response efficiency are remarkably improved, and the efficient cooperation requirements of enterprise and public institution cross-department report tasks are met.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and organizational management technology, and in particular to a task decomposition and federation allocation method and system based on a multi-level knowledge graph. Background Art

[0002] When enterprises and institutions collaborate across departments to complete reporting tasks such as listed company annual reports and audit office enterprise work reports, existing task decomposition and allocation technologies have multiple pain points, making it difficult to meet the needs of efficient, secure, and dynamic collaboration. These pain points are mainly reflected in: First, the traditional model for task breakdown relies entirely on staff experience. This requires analyzing reports chapter by chapter, clarifying task boundaries and departmental responsibilities, and is time-consuming and significantly impacted by staff capabilities. Experienced staff members can spend 2-5 days breaking down a single report, while new staff members require an additional 1-2 days of learning. Errors and omissions are also prone to requiring rework. Furthermore, manual breakdown lacks unified standards, resulting in discrepancies among different personnel, making it difficult to establish a standardized process. As the number of reports increases or their content becomes more complex, time costs increase exponentially, easily delaying the overall compilation process.

[0003] Second, regarding data management, existing employee capability data (such as skill proficiency and task completion quality) across departments is stored in separate, independent systems, creating data silos. When assigning tasks across departments, the coordinating body struggles to access data from other departments, which can lead to a mismatch between tasks and employee capabilities. Furthermore, existing technology lacks a secure data sharing solution. Direct sharing can easily leak private and core data, while non-sharing can only lead to subjective recommendations, creating a dilemma between sharing and efficiency.

[0004] Third, regarding task allocation mechanisms, mainstream systems (such as the SAP task module) rely on pre-set rules, resulting in static allocation. These rules are incapable of responding to load fluctuations, leading to uneven workloads and wasted resources. Furthermore, they only support single-level allocation, failing to implement the multi-level linkage of "report breakdown → department matching → employee matching," requiring manual coordination. More critically, existing mechanisms fail to consider privacy protection, transmitting and storing sensitive information in plain text, which can easily lead to leaks.

[0005] In summary, the multiple defects of existing technologies in task decomposition efficiency, data sharing security, dynamic allocation adaptability and privacy protection have seriously restricted the efficiency and quality of cross-departmental collaborative tasks, and cannot meet the organization's needs for intelligent, secure and efficient task processing. An innovative technical solution is urgently needed to address the above pain points. Summary of the Invention

[0006] In view of this, in order to solve the technical problems of multiple defects in existing technologies in task decomposition efficiency, data sharing security, dynamic allocation adaptability and privacy protection when performing cross-departmental collaborative tasks, the embodiments of the present invention provide a task decomposition and federal allocation method and system based on a multi-level knowledge graph, which can realize intelligent and accurate decomposition of report content, dynamic matching of departmental responsibilities and employee capabilities, and real-time optimization of task allocation paths without centrally obtaining the privacy data of each department. In a first aspect, the present invention provides a task decomposition and federation allocation method based on a multi-level knowledge graph, comprising: Using semantic analysis methods, unstructured documents to be processed that require cross-departmental collaboration are parsed, key information related to task assignment is extracted from the documents, and the unstructured documents are converted into structured task carriers that can be used for task assignment; Construct a two-layer knowledge graph covering organizational unit dimensions and individual member dimensions, use a cross-dimensional association algorithm to model cross-hierarchical relationships, map the organizational unit dimension graph and the individual member dimension graph into a unified semantic space, and associate the two types of graphs; Based on the privacy protection mechanism combined with the federated learning mechanism, task allocation is performed on organizational units and individual members; Obtain feedback data on individual member execution tasks and dynamically adjust the task allocation weights of individual members to achieve overall system load balancing.

[0007] The multi-level knowledge graph-based task decomposition and federated allocation method provided by the embodiment of the present invention starts from the entire process of cross-departmental collaborative task processing, realizes the conversion of unstructured documents into structured task carriers through semantic analysis, solves the problems of inefficiency and inconsistent standards of traditional manual decomposition, and provides an accurate basis for subsequent allocation; the construction of a two-layer knowledge graph and cross-dimensional association break the barrier between organizational units and individual member data, and make up for the defects of single-level allocation in existing technologies; privacy protection is combined with federated learning to achieve cross-departmental collaborative allocation while ensuring data security, breaking the dilemma of data silos and privacy leakage; dynamic weight adjustment responds to load changes and avoids resource waste. The overall "analysis-modeling-allocation-optimization" closed loop is formed, which significantly improves the accuracy of task allocation and the efficiency of system response, and adapts to the efficient collaboration needs of cross-departmental reporting tasks of enterprises and institutions.

[0008] In an optional embodiment, the key information related to task allocation includes at least one of the document chapter structure, exclusive organizational unit identification, and core task expression terms; the structured task carrier is a structured task tree. When constructing the structured task tree, key fields and task nodes in the unstructured document to be processed are extracted, and a task tree structure is built based on the association relationship between the key fields and the task nodes.

[0009] The embodiment of the present invention clarifies the key information types and the form of structured task carriers, making the semantic parsing link more targeted and operational. The extracted key information such as the document chapter structure and exclusive organizational unit identification can be directly used for the rapid matching of subsequent tasks and organizational units, reducing the cost of manual identification; the structured task tree is built by associating key fields with task nodes, with clear hierarchy and complete information, which can intuitively present the task decomposition logic, making it easier for the subsequent federal allocation link to accurately locate the organizational unit and individual member corresponding to the task. At the same time, the unified task tree construction rules avoid the differences caused by different personnel analysis, forming a standardized disassembly process, further improving the efficiency and consistency of task disassembly, and laying a standardized data foundation for the entire allocation process.

[0010] In an optional embodiment, the organizational unit dimension map uses organizational units as nodes and inter-unit authority and responsibility relationships as edges, and the member individual dimension map uses member individuals as nodes and inter-individual collaboration relationships as edges; the inter-organizational unit authority and responsibility relationships include approval relationships and collaboration relationships, and the inter-member collaboration relationships use historical collaboration frequency as a quantitative indicator; The cross-dimensional association algorithm is the TransR algorithm, which is used to model cross-level relationships. Specifically, for the entities h, t and cross-level relationship vector r in the organizational unit dimension map, and the entities hr, tr and relationship vector r in the member individual dimension map, the entities are projected from the entity space to the relationship space through the relationship projection matrix Mr, and the association relationship satisfies ; Among them, h r = h M r, t r = t M r , Mr ∈ R k×d is the projection matrix corresponding to the relationship vector r, k is the entity vector dimension, and d is the relationship vector dimension.

[0011] The embodiment of the present invention defines the node and edge attributes of a two-layer knowledge graph, enabling the graph to accurately map the rights and responsibilities of organizational units and the collaboration of individual members, providing high-quality data support for cross-level associations. The clarification of approval and collaboration relationships between organizational units facilitates the rapid matching of tasks to the corresponding responsible departments; the quantification of the frequency of individual member collaboration allows for the prioritization of employee combinations that work well together, improving execution efficiency. The TransR algorithm achieves a unified semantic space mapping of the two types of graphs through a projection matrix, effectively resolving the problem of cross-level semantic misalignment and ensuring the precise alignment of departmental rights and responsibilities with employee capabilities, breaking the limitations of existing technologies that only have single-level allocation and no cross-level associations.

[0012] In an optional embodiment, the privacy protection mechanism is combined with the federated learning mechanism to use key information related to task allocation to allocate tasks to organizational units and individual members, including: Calculating the correlation between each task in the structured task carrier and the rights and responsibilities of the organizational unit, and screening matching organizational units based on the correlation; Each organizational unit quantifies the capabilities of its members locally and encrypts the quantification results through a privacy protection mechanism. Aggregate the encrypted individual capability data of members of each organizational unit to generate the final task assignment list.

[0013] The embodiment of the present invention integrates privacy protection and federated learning mechanisms into the entire task allocation process. First, it screens and matches organizational units through correlation calculation to ensure that task allocation is consistent with the scope of departmental rights and responsibilities, avoiding allocation deviations caused by unclear rights and responsibilities. Each organizational unit locally quantifies member capabilities and encrypts them, so that the original capability data is always retained locally, effectively preventing data leakage and protecting the privacy of departments and members. The federation aggregates encrypted data to generate a task list, eliminating the need to centrally obtain the original data of each department, thus breaking down data silos, achieving cross-departmental data collaboration, and ensuring data security. This process takes into account both allocation accuracy and data security, solving the problems of existing technologies with no basis for allocation or insecure shared data, and improving the security and reliability of cross-departmental task allocation.

[0014] In an optional embodiment, the cosine similarity is used to calculate the correlation between the task and the authority and responsibility of the organizational unit, and a correlation threshold is set. When the calculated correlation is greater than the threshold, it is determined that the task matches the corresponding organizational unit; For tasks with a unique organizational unit identifier, they are directly associated with the corresponding organizational unit; The privacy protection mechanism is a differential privacy mechanism, which implements privacy protection by adding Laplace noise to the quantification results of individual member capabilities, expressed as: F=E i +Laplace(0,β) Among them, F is the encrypted individual capability vector of the member; E i is the original ability vector of member individual i, Laplace(0,β) represents the random noise with location parameter 0 and scale parameter β Laplace distribution, which is used to calculate the original ability vector E i Perform perturbation encryption.

[0015] This embodiment of the present invention uses cosine similarity calculation to provide a quantitative standard for the correlation between tasks and organizational unit responsibilities. Combined with a correlation threshold, this makes organizational unit matching more objective and accurate, avoiding subjective judgment errors. The direct association of tasks with dedicated organizational unit identifiers eliminates the correlation calculation step, further improving allocation efficiency. The differential privacy mechanism achieves encryption by adding Laplace noise. While protecting the privacy of the original data of individual members' capabilities, it effectively maintains data availability, ensuring that employees can still be accurately screened and matched based on encrypted data during federation aggregation. This balances privacy protection with allocation accuracy, meeting data security compliance requirements while ensuring accurate and efficient task allocation. This addresses the issue of allocation failure caused by a lack of privacy protection or excessive encryption in existing technologies.

[0016] In an optional embodiment, obtaining feedback data on individual members' task execution and dynamically adjusting the task allocation weights of individual members includes: Obtain feedback data on individual members' task execution, including task completion progress, task completion quality, and task processing time, to update the individual member's ability quantification results and task backlog; Dynamically adjust the task allocation weight according to the backlog amount, and use the following formula to update the weight:

[0017] Among them, n represents the total number of members who can be assigned tasks in the current system; Q i is the amount of tasks to be processed by member individual i; It represents the sum of all members’ individual tasks to be processed, Q avg is the average number of tasks to be processed by all member individuals; η represents the learning rate, which is dynamically adjusted according to the system performance: η t =η t-1 exp(-0.1 |Q avg -Q i |); Among them, η t Represents the current learning rate, η t-1 represents the learning rate at the previous moment, and η is set with preset upper and lower limits, and exp( ) is an exponential function with the natural constant e as the base.

[0018] The embodiment of the present invention clearly covers the progress, quality and time consumption of task completion around the feedback data, which can fully reflect the individual task execution status of members, provide real and dynamic data basis for updating the ability quantification results and task backlog, and ensure that the subsequent weight adjustment fits the actual ability and load status of the members. The weight update formula based on the backlog and the dynamic learning rate reduces the weight of members with too high a load and increases the weight of members with too low a load, effectively achieving system load balancing and avoiding the uneven busyness and resource waste caused by the static allocation of the existing technology; the exponential decay adjustment and upper and lower limit settings of the learning rate prevent the weight adjustment from being too large and causing system oscillations, ensuring a smooth adjustment process. The overall dynamic optimization of task allocation is achieved, the system resource utilization and task response speed are improved, and the task volume fluctuation scenario is adapted.

[0019] In a second aspect, the present invention provides a task decomposition and federation allocation system based on a multi-level knowledge graph, the system comprising: The document structuring module is used to parse unstructured documents to be processed that require cross-departmental collaboration using semantic analysis, extract key information related to task allocation from the documents, and convert the unstructured documents into structured task carriers that can be used for task allocation; A two-layer knowledge graph construction module is used to construct a two-layer knowledge graph covering the organizational unit dimension and the individual member dimension. It uses a cross-dimensional association algorithm to model cross-hierarchical relationships, maps the organizational unit dimension graph and the individual member dimension graph into a unified semantic space, and associates the two types of graphs. The task allocation module is used to allocate tasks to organizational units and individual members based on the privacy protection mechanism combined with the federated learning mechanism; The dynamic adjustment module is used to obtain feedback data on the execution of tasks by individual members and dynamically adjust the task allocation weights of individual members to achieve overall load balancing of the system.

[0020] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to execute the task decomposition and federated allocation method based on the multi-level knowledge graph of the above-mentioned first aspect or any corresponding embodiment thereof.

[0021] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the task decomposition and federated allocation method based on a multi-level knowledge graph according to the first aspect or any corresponding embodiment thereof.

[0022] In a fifth aspect, the present invention provides a computer program product comprising computer instructions, which are used to enable a computer to execute the task decomposition and federated allocation method based on a multi-level knowledge graph according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 1 is a flowchart of a method for task decomposition and federation allocation based on a multi-level knowledge graph according to an embodiment of the present invention; Figure 2 is a schematic diagram of using various task processors to perform task allocation according to an embodiment of the present invention; Figure 3 2. It is a structural block diagram of a task decomposition and federated allocation system based on a multi-level knowledge graph according to an embodiment of the present invention; Figure 4 A schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0026] This embodiment provides a task decomposition and federation allocation method based on a multi-level knowledge graph. Figure 1 This is a flowchart of a method for task decomposition and federated assignment based on a multi-level knowledge graph according to an embodiment of the present invention. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer device such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here. Figure 1 As shown, the process includes the following steps: Step S1: Use semantic parsing to parse unstructured documents to be processed that require cross-departmental collaboration, extract key information related to task allocation in the documents, and convert the unstructured documents to be processed into structured task carriers that can be used for task allocation.

[0027] An embodiment of the present invention extracts key information related to task assignment from a document, including at least one of the document chapter structure, exclusive organizational unit identifier, and core task expression terms; the structured task carrier is a structured task tree. When constructing the structured task tree, key fields and task nodes in the unstructured document to be processed are extracted, and a task tree structure is constructed based on the association relationship between the key fields and the task nodes.

[0028] The document to be processed is the "2024 Annual Enterprise Work Report", which is stored in Word / PDF format and contains chapters such as "Summary of 2024 Annual Planning Completion", "2025 Annual Budget Management Plan", "Market Expansion Effectiveness Analysis", and "Internal Audit Issue Rectification". The content of each chapter is described in natural language without fixed task division identifiers, which is a typical unstructured document. After inputting it into the multi-level knowledge graph engine, Figure 2 As shown in Figure 2, the process of using a semantic parser to convert it into a structured task carrier that can be used for task assignment includes: 1. Semantic Parsing and Key Information Extraction: The semantic parser uses natural language processing technology (for example, using pre-trained models such as BERT for keyword recognition, entity linking, and chapter structure analysis) to extract three types of key information related to task assignment: (1) Document chapter structure information: Identify the hierarchical structure of "Summary of 2024 Annual Plan Completion → Sub-item 1: Statistics on the achievement rate of planning indicators of each business line; Sub-item 2: Analysis of reasons for non-compliance with indicators; Sub-item 3: Formulation of subsequent improvement measures" and clarify the hierarchical relationship of task decomposition; (2) Exclusive organizational unit identifier: From the “2025 Budget Management Plan” section, identify the statement “The Finance Department is exclusively responsible for budget preparation and cost accounting” and extract the exclusive identifier of the “Finance Department”; from the “Internal Audit Issue Rectification” section, identify the statement “The Audit Department takes the lead and all business departments cooperate in rectification” and extract the organizational unit association identifier of “Audit Department (lead) + Business Department (cooperation)”; (3) Core task expression terms: Extract core task terms such as “competitive product market share research”, “customer satisfaction statistics”, and “new area expansion effect evaluation” from the “market expansion effectiveness analysis” chapter to clarify the specific task content corresponding to each chapter.

[0029] The process of converting an unstructured report into a structured task carrier in this embodiment of the present invention involves converting the unstructured report into a "structured task tree," a carrier for task assignment, based on the extracted key information. The task tree begins with "Overall Tasks for the 2024 Annual Corporate Work Report," with each chapter corresponding to a first-level subnode (e.g., "Planning Completion Tasks" and "Budget Management Tasks"). Sub-items and core tasks within each chapter correspond to second- and third-level subnodes. Each task node is annotated with associated key information (e.g., "Budget Management Task - First-Level Subnode: Exclusive Organizational Unit = Finance Department; Core Task Terms = Budget Preparation, Cost Accounting"; "Competitive Product Market Share Research - Second-Level Subnode: Core Task Terms = Competitive Product Data Collection, Market Share Calculation; Associated Organizational Unit = Marketing Department"), creating a structured task assignment framework with a clear hierarchy and complete information.

[0030] This process addresses the pain point of inefficient manual task analysis, improving both efficiency and standardization. Semantic parsing can transform unstructured documents into structured task trees in minutes, significantly faster than the 2-5 days it typically takes for manual analysis. Furthermore, the parsing process, based on unified keyword recognition and structural analysis rules, avoids subjective differences in manual analysis and ensures consistent task analysis standards across reports and time periods, creating a standardized task analysis process suitable for scenarios with surging report volumes or increasing report complexity. This provides a precise basis for subsequent multi-level task allocation: Key information such as "unique organizational unit identifiers" and "core task terminology" annotated in the structured task tree directly supports the subsequent "graph construction" phase for matching departmental responsibilities and "federal allocation" for matching employee capabilities, avoiding allocation biases caused by missing information and improving task allocation accuracy.

[0031] Step S2: construct a two-layer knowledge graph covering the organizational unit dimension and the member individual dimension, use the cross-dimensional association algorithm to model the cross-level relationship, map the organizational unit dimension graph and the member individual dimension graph to a unified semantic space, and associate the two types of graphs.

[0032] Specifically, the organizational unit dimension map in the embodiment of the present invention uses organizational units as nodes and the rights and responsibilities between units as edges, and the member individual dimension map uses member individuals as nodes and the collaborative relationships between individuals as edges; the rights and responsibilities between organizational units include approval relationships and collaborative relationships, and the collaborative relationships between member individuals use historical collaboration frequency as a quantitative indicator.

[0033] Taking the four major organizational units (Finance, Planning, Marketing, and Audit) and their employees involved in a company's 2024 Annual Corporate Work Report as an example, the construction and association logic of the two-layer knowledge graph are explained in detail: 1. Example of organizational unit dimension map (1) Node definition: With "Finance Department, Planning Department, Marketing Department, Audit Department" as the core nodes, each node contains basic attributes such as the name of the organizational unit and the core scope of authority and responsibility (e.g., Finance Department attributes: Name = Finance Department, Core responsibilities = Budget preparation, cost accounting, fund management; Planning Department attributes: Name = Planning Department, Core responsibilities = Planning, indicator monitoring, progress tracking).

[0034] (2) Definition of edges (rights and responsibilities): Taking the approval relationship and collaboration relationship between organizational units as edges, clarify the relationship type and business meaning: Approval relationship: Finance Department → Procurement Department (in this scenario, the Procurement Department participates in the "Procurement Cost Analysis" subtask in the report). The edge attribute is labeled "Budget Approval Authority," indicating that the "Procurement Cost Analysis Data" submitted by the Procurement Department must be approved by the Finance Department before being included in the report. Finance Department → Planning Department. The edge attribute is labeled "Budget Adjustment Approval Authority," indicating that the "2025 Annual Planning Budget" formulated by the Planning Department must be approved by the Finance Department. Collaborative relationship: Planning Department → Marketing Department, with the edge attribute labeled "Planning Collaboration", indicates that the Planning Department needs to collaborate with the Marketing Department to obtain market expansion data in the "Plan Completion Summary" task; Audit Department → Finance Department / Planning Department / Marketing Department, with the edge attribute labeled "Audit Rectification Collaboration", indicates that the Audit Department needs the cooperation of various departments to provide rectification materials in the "Internal Audit Issue Rectification" task.

[0035] 2. Example of individual member dimension map (1) Node definition: The core employees in each organizational unit are taken as nodes. Each node contains attributes such as employee name, department, and core competency label (e.g., attributes of employee A in the Finance Department: Name = Employee A, Department = Finance Department, Core Competencies = Budget Preparation (Proficiency 0.9), Cost Accounting (Proficiency 0.8); attributes of employee B in the Marketing Department: Name = Employee B, Department = Marketing Department, Core Competencies = Competitive Product Analysis (Proficiency 0.9), Data Statistics (Proficiency 0.85)).

[0036] (2) Definition of edge (collaboration relationship): The edge is constructed based on the historical collaboration frequency between employees as a quantitative indicator. The frequency value range is 0-1 (the closer the value is to 1, the more frequent the collaboration and the smoother the cooperation): Finance employee A → Planning employee C, with the edge attribute labeled "Historical Collaboration Frequency = 0.8," indicating that both employees were core collaborators in the past three "Budget and Planning Collaboration" tasks, demonstrating high collaboration efficiency. Marketing Department Employee B → Marketing Department Employee D, with the edge attribute labeled "Historical Collaboration Frequency = 0.9," indicates that the two have long collaborated to complete the "Market Data Research and Analysis" task and have strong data connection and result integration capabilities. Audit Department Employee E → Finance Department Employee A, with the edge attribute labeled "Historical Collaboration Frequency = 0.75," indicates that the two have collaborated on two "Audit Rectification Data Verification" tasks and are familiar with each other's work processes.

[0037] The dual-layer graph constructed by this invention achieves a hierarchical association between "organizational units (departments) - individual members (employees)" for the first time, unlike existing technologies that only support single-level allocation. The organizational unit graph clarifies departmental responsibilities and collaboration relationships, providing a basis for "report task → department matching"; the individual member graph quantifies employee collaboration efficiency, supporting "department task → employee matching"; thus, forming a complete allocation chain from "report breakdown → department matching → employee matching", avoiding manual connection costs caused by broken task allocation chains and improving task execution efficiency. Simultaneously, the organizational unit graph provides a structured storage of "scopes of responsibilities and cross-departmental association rules" distributed across departments, while the individual member graph systematically integrates employee "competency tags and historical collaboration data". These two types of graphs break down departmental data barriers, preventing information asymmetry caused by data silos and providing a structured data carrier for subsequent privacy-preserving sharing under the federated learning mechanism.

[0038] The cross-dimensional association algorithm adopted in the embodiment of the present invention is the TransR algorithm, which is used to model cross-level relationships. Specifically, for the entities h, t and cross-level relationship vector r in the organizational unit dimension map, as well as the entities hr, tr and relationship vector r in the member individual dimension map, the entities are projected from the entity space to the relationship space through the relationship projection matrix Mr, and the association relationship satisfies ; Among them, h r = h M r, t r = t M r , Mr ∈ R k×d is the projection matrix corresponding to the relationship vector r, k is the entity vector dimension, and d is the relationship vector dimension.

[0039] In this embodiment of the present invention, the TransR algorithm is used to model cross-hierarchical relationships between the organizational unit dimension map (department) and the member individual dimension map (employee). Its core function is to map entity vectors in two different semantic spaces into a unified relational space through a relational projection matrix, achieving a precise association between "departmental responsibilities and employee capabilities." This paper uses the example of a company processing the cross-hierarchical association between "Finance Department (organizational unit) and Employee A (individual member)" in its "2024 Annual Corporate Work Report" to illustrate the specific application and value of the TransR algorithm. The specific steps are as follows: 1. Define modeling objects and vector dimensions (1) Entity and relationship definition: Organizational unit dimension entity h: Finance Department (organizational unit entity), its original entity vector (Assume (k = 5), representing the five core responsibilities of "budget preparation, cost accounting, fund management, tax declaration, and financial auditing", and the vector value is the proficiency of each responsibility, such as (h = [0.95, 0.9, 0.85, 0.8, 0.75])); (2) Member individual dimension entity t: Employee A (finance department employee), whose original entity vector (The same dimension (k=5), representing the proficiency of the corresponding responsibilities, such as (t = [0.92, 0.88, 0.8, 0.75, 0.7])); Cross-level relationship r: "Budget preparation authority-capability association" of "Finance Department-Employee A", relationship vector (Assume (d=3), representing the three related indicators of "rights and responsibilities matching, close collaboration, and task completion rate", such as (r =[0.9, 0.85, 0.88])).

[0040] (3) Definition of projection matrix: For the specific relationship r of “budget preparation authority-capability association”, the relationship projection matrix is ​​preset (5 3) Matrix, used to project entities from the 5-dimensional entity space to the 3-dimensional relational space. The matrix values ​​are trained by historical "department responsibilities-employee capabilities" association data. An example is as follows:

[0041] 2. Perform cross-space projection calculations According to the TransR algorithm rules, the organizational unit entity h and the member individual entity t are mapped to the relational space through the projection matrix, and the projected entity vectors hr and tr are obtained: Finance Department projection vector h r =h :Substitute h = [0.95, 0.9, 0.85, 0.8, 0.75] and the above Mr to calculate hr [0.91, 0.86, 0.89] (Specific calculation process: For example, the first column is 0.95 0.85 + 0.9 0.1 +0.85 0.05 + 0.8 0.03 + 0.75 0.02 0.91, the same applies to the other columns); Employee A's projection vector t r =t :Substituting t = [0.92, 0.88, 0.8, 0.75, 0.7] and the above Mr, we can calculate t r [0.88, 0.83,0.86].

[0042] 3. Verify the effectiveness of cross-level associations According to the core association formula h of the TransR algorithm r + t r , verify whether the cross-level association of "Finance Department-Employee A" matches: Calculate h r + : [0.91 + 0.9, 0.86 + 0.85, 0.89 + 0.88] = [1.81, 1.71, 1.77]; Calculate vector similarity (such as cosine similarity): h r + With t r The cosine similarity is ≈0.98 (close to 1), which meets the "approximately equal" condition. This shows that the "budget preparation responsibilities of the Finance Department" and the "budget preparation capabilities of employee A" are associated and matched. The "budget management task" can be further assigned from the Finance Department to employee A.

[0043] This embodiment of the present invention uses the TransR algorithm for cross-hierarchical relationship modeling to address the pain point of "cross-hierarchical semantic misalignment" and precisely align responsibilities and capabilities. Existing technologies lack cross-hierarchical association modeling. The TransR algorithm uses a projection matrix to map departments (entity space) and employees (entity space) into a unified relational space, eliminating the semantic dimensional differences between the two entities. For example, in the example above, the association similarity between "Finance Department Responsibilities" and "Employee A's Capabilities" reaches 0.98. This ensures consistent "responsibilities-capabilities" when assigning tasks from department to employee, improving task allocation accuracy.

[0044] The core goal of this invention is to achieve a three-level linkage of "report decomposition → department matching → employee matching". The TransR algorithm is the key technical support: by modeling the cross-level association of "department-employee", the result of "report task → department matching" (such as "budget management task matching the Finance Department") can be directly connected to "department → employee matching" (such as the Finance Department matching employee A), forming a complete allocation chain. This solves the defect of existing technologies that only support single-level allocation and require manual connection, and improves the degree of automation of the task allocation process.

[0045] In step S3, tasks are allocated to organizational units and individual members based on the privacy protection mechanism combined with the federated learning mechanism.

[0046] like Figure 2 As shown, the embodiment of the present invention generates a task list through a federated allocation controller and a differential privacy aggregator, which specifically includes the following steps: S31, calculating the correlation between each task in the structured task carrier and the responsibilities of the organizational unit, and screening matching organizational units based on the correlation. Specifically, cosine similarity is used to calculate the correlation between the tasks and the responsibilities of the organizational unit, and a correlation threshold is set. When the calculated correlation is greater than the threshold, the task is determined to match the corresponding organizational unit.

[0047] In step S32, each organizational unit quantifies the individual capabilities of its members locally and encrypts the quantification results using a privacy protection mechanism. Specifically, the privacy protection mechanism is a differential privacy mechanism that adds Laplace noise to the quantification results of individual member capabilities, which is expressed as: F=E i +Laplace(0,β) Among them, F is the encrypted individual capability vector of the member; E i is the original ability vector of member individual i, Laplace(0,β) represents the random noise with location parameter 0 and scale parameter β Laplace distribution, which is used to calculate the original ability vector E i Perform perturbation encryption.

[0048] S33, aggregate the encrypted individual capability data of members of each organizational unit to generate a final task assignment list.

[0049] In one example, the structured task carrier is the "2024 Annual Corporate Work Report" task tree. The "2025 Annual Budget Management" task node is labeled with the core terms "budget preparation, cost accounting," and the "Market Expansion Effectiveness Analysis" task node is labeled with the core terms "competitive product market share research, customer satisfaction statistics." In the organizational unit dimension map, the Finance Department's responsibility vector includes "budget preparation (weight 0.95), cost accounting (weight 0.9)," and the Marketing Department's responsibility vector includes "competitive product market share research (weight 0.92), customer satisfaction statistics (weight 0.88)." This includes the following processes: 1. Use cosine similarity to calculate the correlation between tasks and organizational unit responsibilities:

[0050] The correlation score between the budget management task and the Finance Department is approximately 0.98; the correlation score between the budget management task and the Marketing Department is approximately 0; and the correlation score between the market expansion analysis task and the Marketing Department is approximately 0.99. With a preset correlation threshold of 0.7, the budget management task is assigned to the Finance Department, and the market expansion analysis task is assigned to the Marketing Department. This allows for direct assignment of organizational units without manual intervention.

[0051] 2. Each department completes the employee capability vector calculation locally (without uploading the original data to the federal controller): Finance Department Local: Employee A Capability Vector E A = [Budgeting: 0.92, Cost Accounting: 0.88], Employee B's capability vector E B = [Budgeting:0.85, Costing:0.9]); Marketing Department Local: Employee C Capability Vector E C = [Competitive product research: 0.93, Satisfaction statistics: 0.89], Employee D's ability vector E D = [Competitive research: 0.87, satisfaction statistics: 0.91].

[0052] Differential privacy encryption processing: Each department adds noise to the capability vector based on the Laplace mechanism. The noise parameter β is set according to the privacy level of the department (for example, the Finance Department’s data is sensitive, β=0.1; the Marketing Department’s β=0.08): Finance Department: Employee A encrypted vector (F A = E A + Laplace(0,0.1) = [0.92+0.07, 0.88-0.05] = [0.99, 0.83]; Employee B’s encrypted vector F B = [0.85-0.06, 0.9+0.04] = [0.79,0.94]; Marketing Department: Employee C Encrypted Vector F C = [0.93+0.06, 0.89-0.03] = [0.99, 0.86]; employee D encrypted vector F D = [0.87-0.04, 0.91+0.05] = [0.83, 0.96].

[0053] 3. Aggregate encrypted data and generate the final task assignment list: Encrypted data upload and aggregation: Finance Department and Marketing Department only upload the encrypted F A 、F B 、F C 、F D Uploaded to the federated allocation controller, the controller does not need to obtain the original capability data, and directly calculates the "task-employee" matching degree based on the encrypted vector (matching degree = cosine similarity between the core terms of the task and the encrypted capability vector): Budget management task-employee A: fit ≈ 0.97; Budget management task-employee B: fit ≈ 0.92; Market expansion analysis task-employee C: matching degree ≈ 0.98; market expansion analysis task-employee D: matching degree ≈ 0.93.

[0054] Generate a task list: Sort by matching degree in descending order, filter the top 1 employee for assignment, and assign tasks to the following results: "Budget Management Task → Finance Department Employee A; Market Development Effectiveness Analysis Task → Marketing Department Employee C." This task list is then synchronized to the corresponding employee's task terminal. Furthermore, to protect task and employee information, a task list is generated that includes task and employee IDs, for example: {Task ID: T001, T001, Assigned Employees: {Employee ID: E023, Matching Degree: 0.97}, {Employee ID: E107, Matching Degree: 0.93}}.

[0055] This embodiment of the present invention uses a federated learning framework, allowing each department to upload only encrypted data, while the original capability data remains locally. This not only breaks down data silos (enabling cross-departmental data collaboration) but also ensures privacy through Laplace noise (for example, the original capability of a Finance Department employee, 0.92, is visible only to the department; only the encrypted 0.99 is visible to the outside world). The core advantage of the differential privacy mechanism is that encryption does not affect data availability. After adding Laplace noise in this invention, the deviation between the encrypted and original vectors is controlled within 5% (for example, employee A's budget preparation capability is originally 0.92, and after encryption, it is 0.99, resulting in a deviation of ≈7.6%, which can be further reduced by reducing β). This ensures that the matching degree calculated by the federated controller based on encrypted data still reflects the employee's true capability, avoids allocation failure due to over-encryption, and achieves a balance between privacy protection and allocation efficiency.

[0056] Step S4: Obtain feedback data on individual member tasks and dynamically adjust the task allocation weights of individual members to achieve overall system load balancing.

[0057] The embodiment of the present invention realizes the overall load balancing of the system through a dynamic adjuster. Specifically, the following steps are included: S41, obtaining feedback data on individual members' task execution, including task completion progress, task completion quality, and task processing time, for updating the member's individual capability quantification results and task backlog; S42 dynamically adjusts the task allocation weight based on the backlog, using the following formula to update the weight:

[0058] Among them, n represents the total number of members who can be assigned tasks in the current system; Q i is the amount of tasks to be processed by member individual i; It represents the sum of all members’ individual tasks to be processed, Q avg is the average number of tasks to be processed by all member individuals; η represents the learning rate, which is dynamically adjusted according to the system performance: η t =η t-1 exp(-0.1 |Q avg -Q i |); Among them, η t Represents the current learning rate, which is a key parameter used to adjust the weight of employee task allocation at time t. The learning rate determines the step size of the weight adjustment, and its value affects the amplitude of the task allocation weight adjustment; η t-1It represents the learning rate at the previous moment, that is, the learning rate value at the t-1 moment. It provides a basic reference value for the calculation of the learning rate at the current moment, reflecting the continuity and correlation of the learning rate in the time series. avg It is the average amount of tasks to be processed by all individual members, which is obtained by dividing the sum of the amount of tasks to be processed by all individual members in the system who can assign tasks by the total number of individual members. It is used to measure the average level of task load of the entire system. i It represents the amount of pending tasks of individual member i, reflecting the number of unfinished tasks currently accumulated by a single employee. By comparing the Q i With Q avg , can be used to determine whether the employee's task load is overloaded, average, or underloaded. exp() is an exponential function with the natural constant e as its base, which converts differences in employee task backlogs into learning rate adjustment coefficients. In this embodiment of the present invention, η is set with preset upper and lower limits to prevent oscillation.

[0059] This embodiment of the present invention uses the task load adjustment of "three employees in the marketing department (employees C, D, and E)" when a company processes the "2024 Annual Enterprise Work Report" as an example to illustrate the application and value of the weight update formula.

[0060] 1. Basic parameter definition The total number of individual members, n, is 3 (Marketing Department employees C, D, and E, all of whom are assigned to the subtasks related to "Market Expansion Effectiveness Analysis"); Amount of tasks to be processed: Q of employee C C =8, Q of employee D D =3, employee E's Q E =1; Initial learning rate η t-1 =0.5 (preset upper limit η max =0.8, lower limit η min =0.2).

[0061] 2. Calculation of key parameters The total number of tasks to be processed by all members = 8+3+1=12; Average number of tasks to be processed = 4.

[0062] 3. Learning rate calculation (based on η t =η t-1 xp(-0.1 |Q avg -Q i |)) Employee C (Q C=8, too high load): η t =0.5 exp(-0.1 ≈0.335, valid within the range of (0.2-0.8)); Employee D (Q D =3, load close to average): η t =0.5 exp(-0.1 ≈0.45; Employee E (Q E =1, load too low): η t =0.5 exp(-0.1 ≈0.37.

[0063] 4. Task allocation weight adjustment Assume that the weights of the three people before the adjustment are w t-1 = 0.3, the weight update logic is "the higher the load, the more the weight is reduced" (combined with η t Control adjustment range): Employee C (high load, η t ≈0.335)):w t = 0.33 - 0.335 0.2≈ 0.26 (weight reduction, reducing new task assignment); Employee D (moderate load, η t ≈0.45):w t = 0.33 - 0.45 0.05≈ 0.31) (weight fine-tuning); Employee E (low load, η t ≈0.37):w t =0.33 + 0.37 0.2≈ 0.39 weight increase, adding new task allocation).

[0064] The embodiment of the present invention dynamically calculates the learning rate and weight through a formula. The weight of employees with excessive load (such as C) is reduced, and the weight of employees with excessive load (such as E) is increased, ensuring that tasks are tilted towards idle members, avoiding uneven workload, reducing waste of human resources, and improving overall task processing efficiency. The technical effect is that the system response time is less than 30 seconds when the sudden task volume increases by 200%, avoiding adjustment shocks and ensuring stable system operation.

[0065] In this embodiment, a task decomposition and federated allocation system based on a multi-level knowledge graph is also provided. The system is used to implement the above-mentioned embodiments and preferred implementation methods, and the details that have been described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the system described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceivable.

[0066] This embodiment provides a task decomposition and federation allocation system based on a multi-level knowledge graph, such as Figure 3 Shown, including: The document structuring module 31 is used to parse unstructured documents to be processed that require cross-departmental collaboration using semantic analysis, extract key information related to task assignment from the documents, and convert the unstructured documents into structured task carriers that can be used for task assignment; A dual-layer knowledge graph construction module 32 is used to construct a dual-layer knowledge graph covering both organizational unit dimensions and individual member dimensions, use a cross-dimensional association algorithm to model cross-hierarchical relationships, map the organizational unit dimension graph and the individual member dimension graph into a unified semantic space, and associate the two types of graphs; The task allocation module 33 is used to allocate tasks to organizational units and individual members based on the privacy protection mechanism combined with the federated learning mechanism; The dynamic adjustment module 34 is used to obtain feedback data on individual member execution tasks and dynamically adjust the task allocation weights of individual members to achieve overall system load balancing.

[0067] In some optional embodiments, the key information related to task assignment in the document structuring module 31 includes at least one of the document chapter structure, exclusive organizational unit identification, and core task expression terms; the structured task carrier is a structured task tree. When constructing the structured task tree, the key fields and task nodes in the unstructured document to be processed are extracted, and the task tree structure is built based on the association relationship between the key fields and the task nodes.

[0068] In some optional implementations, the organizational unit dimension graph in the dual-layer knowledge graph construction module 32 uses organizational units as nodes and inter-unit authority and responsibility relationships as edges, while the member individual dimension graph uses member individuals as nodes and inter-individual collaboration relationships as edges; the inter-organizational authority and responsibility relationships include approval relationships and collaboration relationships, and the inter-member collaboration relationships use historical collaboration frequency as a quantitative indicator; The cross-dimensional association algorithm is the TransR algorithm, which is used to model cross-level relationships. Specifically, for the entities h, t and cross-level relationship vector r in the organizational unit dimension map, and the entities hr, tr and relationship vector r in the member individual dimension map, the entities are projected from the entity space to the relationship space through the relationship projection matrix Mr, and the association relationship satisfies h r + ;t r； Among them, h r = h M r, t r = t M r , Mr ∈ R k×d is the projection matrix corresponding to the relationship vector r, k is the entity vector dimension, and d is the relationship vector dimension.

[0069] In some optional implementations, the task assignment module 33 includes: The organizational unit matching submodule is used to calculate the correlation between each task in the structured task carrier and the rights and responsibilities of the organizational unit, and to screen the matching organizational units based on the correlation; the encryption processing submodule is used for each organizational unit to quantify the individual capabilities of its members locally and encrypt the quantification results through a privacy protection mechanism; the allocation list generation submodule is used to aggregate the encrypted individual capability data of each organizational unit to generate the final task allocation list.

[0070] In an optional embodiment, the cosine similarity is used to calculate the correlation between the task and the authority and responsibility of the organizational unit, and a correlation threshold is set. When the calculated correlation is greater than the threshold, it is determined that the task matches the corresponding organizational unit; For tasks with a unique organizational unit identifier, they are directly associated with the corresponding organizational unit; The privacy protection mechanism is a differential privacy mechanism, which implements privacy protection by adding Laplace noise to the quantification results of individual member capabilities, expressed as: F=E i +Laplace(0,β) Among them, F is the encrypted individual capability vector of the member; E i is the original ability vector of member individual i, Laplace(0,β) represents the random noise with location parameter 0 and scale parameter β Laplace distribution, which is used to calculate the original ability vector E i Perform perturbation encryption.

[0071] In an optional embodiment, the dynamic adjustment module 34 includes: The feedback data submodule is used to obtain feedback data on individual members' task execution, including task completion progress, task completion quality, and task processing time, which is used to update the individual member's ability quantification results and task backlog; The weight update submodule is used to dynamically adjust the task allocation weights based on the backlog. The weight update is performed using the following formula:

[0072] Among them, n represents the total number of members who can be assigned tasks in the current system; Q i is the amount of tasks to be processed by member individual i; It represents the sum of all members’ individual tasks to be processed, Q avg is the average number of tasks to be processed by all member individuals; η represents the learning rate, which is dynamically adjusted according to the system performance: η t =η t-1 exp(-0.1 |Q avg -Q i |); η t Represents the current learning rate, η t-1 Represents the learning rate at the previous moment, and η is set with preset upper and lower limits.

[0073] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here. The task decomposition and federated allocation system based on the multi-level knowledge graph in this embodiment is presented in the form of functional units, where the units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0074] The embodiment of the present invention also provides a computer device having the above Figure 3 The task decomposition and federated allocation system based on multi-level knowledge graph is shown.

[0075] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 4As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0076] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0077] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0078] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0079] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0080] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0081] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0082] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0083] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A task decomposition and federation allocation method based on multi-level knowledge graph, characterized by: include: Using semantic analysis methods, unstructured documents to be processed that require cross-departmental collaboration are parsed, key information related to task assignment is extracted from the documents, and the unstructured documents are converted into structured task carriers that can be used for task assignment; Construct a two-layer knowledge graph covering organizational unit dimensions and individual member dimensions, use a cross-dimensional association algorithm to model cross-hierarchical relationships, map the organizational unit dimension graph and the individual member dimension graph into a unified semantic space, and associate the two types of graphs; Based on the privacy protection mechanism combined with the federated learning mechanism, task allocation is performed on organizational units and individual members; Obtain feedback data on individual member execution tasks and dynamically adjust the task allocation weights of individual members to achieve overall system load balancing.

2. The method according to claim 1, characterized in that The key information related to task allocation includes at least one of the document chapter structure, exclusive organizational unit identification, and core task expression terms; the structured task carrier is a structured task tree. When constructing the structured task tree, key fields and task nodes in the unstructured document to be processed are extracted, and the task tree structure is built based on the association relationship between the key fields and the task nodes.

3. The method according to claim 1, characterized in that The organizational unit dimension map uses organizational units as nodes and the rights and responsibilities between units as edges. The member individual dimension map uses individual members as nodes and the collaborative relationships between individuals as edges. The rights and responsibilities between organizational units include approval relationships and collaborative relationships. The collaborative relationships between individual members are quantified using the historical collaboration frequency. The cross-dimensional association algorithm is the TransR algorithm, which is used to model cross-level relationships. Specifically, for the entities h, t and cross-level relationship vector r in the organizational unit dimension map, and the entities hr, tr and relationship vector r in the member individual dimension map, the entities are projected from the entity space to the relationship space through the relationship projection matrix Mr, and the association relationship satisfies h r + t r ; Among them, h r = h M r, t r = t M r , Mr ∈ R k×d is the projection matrix corresponding to the relationship vector r, k is the entity vector dimension, and d is the relationship vector dimension.

4. The method according to claim 1, wherein The privacy protection mechanism combined with the federated learning mechanism uses key information related to task allocation to allocate tasks to organizational units and individual members, including: The correlation between each task in the structured task carrier and the rights and responsibilities of the organizational unit is calculated, and matching organizational units are screened based on the correlation; each organizational unit quantifies the individual capabilities of its members locally and encrypts the quantification results through a privacy protection mechanism; the encrypted individual capability data of each organizational unit is aggregated to generate a final task assignment list.

5. The method according to claim 2, characterized in that The cosine similarity is used to calculate the correlation between the task and the responsibilities of the organizational unit, and a correlation threshold is set. When the calculated correlation is greater than the threshold, the task is determined to match the corresponding organizational unit; For tasks with a unique organizational unit identifier, they are directly associated with the corresponding organizational unit; The privacy protection mechanism is a differential privacy mechanism, which implements privacy protection by adding Laplace noise to the quantification results of individual member capabilities, expressed as: F=E i +Laplace(0,β) Among them, F is the encrypted individual capability vector of the member; E i is the original ability vector of member individual i, Laplace(0,β) represents the random noise with location parameter 0 and scale parameter β Laplace distribution, which is used to calculate the original ability vector E i Perform perturbation encryption.

6. The method according to claim 1, characterized in that The step of obtaining feedback data on individual members' task execution and dynamically adjusting the task allocation weights of individual members includes: Obtain feedback data on individual members' task execution, including task completion progress, task completion quality, and task processing time, to update the individual member's ability quantification results and task backlog; Dynamically adjust the task allocation weight according to the backlog amount, and use the following formula to update the weight: Among them, n represents the total number of members who can be assigned tasks in the current system; Q i is the amount of tasks to be processed by member individual i; It represents the sum of all members’ individual tasks to be processed, Q avg is the average number of tasks to be processed by all member individuals; η represents the learning rate, which is dynamically adjusted according to the system performance: the t =η t-1 exp(-0.1 |Q avg - Q i |); Among them, η t Represents the current learning rate, η t-1 represents the learning rate at the previous moment, and η is set with preset upper and lower limits, and exp( ) is an exponential function with the natural constant e as the base.

7. A task decomposition and federation allocation system based on multi-level knowledge graph, characterized by: include: The document structuring module is used to parse unstructured documents to be processed that require cross-departmental collaboration using semantic analysis, extract key information related to task allocation from the documents, and convert the unstructured documents into structured task carriers that can be used for task allocation; A two-layer knowledge graph construction module is used to construct a two-layer knowledge graph covering the organizational unit dimension and the individual member dimension. It uses a cross-dimensional association algorithm to model cross-hierarchical relationships, maps the organizational unit dimension graph and the individual member dimension graph into a unified semantic space, and associates the two types of graphs. The task allocation module is used to allocate tasks to organizational units and individual members based on the privacy protection mechanism combined with the federated learning mechanism; The dynamic adjustment module is used to obtain feedback data on the execution of tasks by individual members and dynamically adjust the task allocation weights of individual members to achieve overall load balancing of the system.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the task decomposition and federated allocation method based on a multi-level knowledge graph as described in any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, which are used to enable a computer to execute the task decomposition and federation allocation method based on a multi-level knowledge graph according to any one of claims 1 to 6.

10. A computer program product, characterized in that It includes computer instructions, which are used to enable a computer to execute the task decomposition and federation allocation method based on a multi-level knowledge graph as described in any one of claims 1 to 6.

Images