Cross-department cooperation efficiency dynamic optimization system and method

Abstract

The invention discloses a cross-department cooperation efficiency dynamic optimization system and method, and relates to the technical field of cross-department collaboration, and the method comprises the following steps: constructing a task distribution network containing all tasks based on the basic information of cross-department cooperation tasks, and comprehensively describing the structural characteristics and interdependence of the tasks in the cooperation process; in a cross-department cooperation operation process, setting a fixed monitoring window for each task, and collecting conflict data of a target task and other tasks in a cooperation process in real time to construct a conflict data analysis set; and based on the conflict data analysis set, extracting indexes of serious conflicts between the target task and other tasks from the conflict data analysis set. According to the method, intelligent judgment and grading of task conflicts can be realized, rapid optimization is carried out for mild conflicts, sub-task splitting is carried out adaptively for severe conflicts, the bottleneck is relieved, the flexibility and cooperation efficiency of task scheduling are improved, and efficient and stable operation of cross-department cooperation in a complex environment is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of cross-departmental collaboration technology, and in particular to a cross-departmental collaboration efficiency dynamic optimization system and method. Background Art

[0002] The cross-departmental collaborative efficiency dynamic optimization system is a comprehensive management system that integrates information sharing, process collaboration, and intelligent optimization decision-making. It aims to solve problems such as poor communication, waste of resources, and inconsistent goals that occur in the collaboration process between different departments within an enterprise or organization. Based on real-time collection and analysis of multi-dimensional data, the system dynamically perceives the task progress, resource status, and collaboration bottlenecks of each department. Through process modeling, performance indicator evaluation, and intelligent optimization algorithms, it adjusts collaboration strategies and resource allocation in real time to improve overall collaboration efficiency. At the same time, the system supports information transparency, goal alignment, and intelligent early warning across departments. It can automatically generate optimization plans when task conflicts, bottlenecks occur, or the external environment changes, to ensure the efficiency, flexibility, and sustainability of the collaboration process.

[0003] The prior art has the following deficiencies: Existing technologies generally have the problem of insufficient optimization capabilities when dealing with task conflicts in cross-departmental collaboration, especially when there is a serious conflict between a key task and multiple related tasks (such as high resource dependence, overlapping time windows, and priority conflicts). Conventional optimization methods can only handle it based on static rules or simple priority adjustments. It is difficult to take into account the complex dependencies between tasks and the global collaborative efficiency, which can easily lead to some tasks being forced to be postponed or resource allocation being unbalanced, thereby affecting the overall collaboration efficiency and project execution stability. It is difficult to meet the actual needs in highly dynamic, multi-constrained complex collaboration scenarios.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a system and method for dynamically optimizing the performance of cross-departmental collaboration, which comprehensively characterizes the structural characteristics and dependencies of cross-departmental collaborative tasks based on a task allocation network, combines real-time conflict data, and quantifies and intelligently determines the degree of task conflict through a machine learning model, thereby achieving accurate classification and hierarchical processing of mild and severe conflicts. For mild conflicts, the system can quickly generate task optimization plans through an automatic optimization unit to improve collaboration flexibility and response speed; and for severe conflicts, the system can adaptively divide and reconstruct tasks into granularly controllable subtasks based on the severity of the conflict, significantly alleviating bottleneck problems caused by concentrated task conflicts, and improving the flexibility of task scheduling and the optimization capabilities of global collaboration, thereby ensuring that cross-departmental collaboration can still maintain efficient and stable operation in a highly dynamic, multi-constrained complex environment to solve the problems in the above-mentioned background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a method for dynamically optimizing cross-departmental collaboration efficiency, comprising the following steps: Based on the basic information of cross-departmental collaborative tasks, a task allocation network containing each task is constructed to comprehensively describe the structural characteristics and interdependencies of tasks in the collaborative process; During the cross-departmental collaborative operation, a fixed monitoring window is set for each task to collect conflict data between the target task and other tasks in the collaborative process in real time to build a conflict data analysis set; Based on the conflict data analysis set, the indicators of serious conflicts between the target task and other tasks are extracted. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. The extracted conflict feature vector is input into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm to evaluate the conflict severity of the target task and output the conflict degree score of the target task; According to the conflict degree judgment result output by the machine learning model, the decision branch processing is executed: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit is used to directly generate a collaborative optimization plan; when the conflict degree between the target task and other tasks is severe, that is, there is a high degree of coupling, concentrated resource dependence or unsolvable time conflict between the task and multiple other tasks, the target task is divided into multiple independently schedulable sub-task units. Through fine-grained task splitting, the dependence on the original conflict path is minimized to the greatest extent, and the number of sub-task divisions is adaptively adjusted according to the severity of the conflict. The intensity of concentrated conflict is reduced through disassembly, and efficient dynamic optimization of task collaboration is achieved.

[0007] Preferably, based on the basic information of the cross-departmental collaborative tasks, a task allocation network including each task is constructed, and the specific steps are as follows: It is necessary to collect basic descriptions of each task, execution entities, required resources, expected duration, priority, and other information based on the business needs and project goals of each department, and standardize this information; After obtaining the basic information of each task, define each task as a node in the graph, assign it a unique identifier, and record key information such as the executor, resource requirements, and planned time of the task in the node attributes; After the nodes are established, corresponding association edges are established for the task nodes that need to interact or connect according to constraints such as the order of precedence and follow-up, resource sharing, and time window conflicts; After integrating all node and edge information, a complete task allocation network is formed, and possible redundancy or circular dependencies in the network are checked and corrected to ensure the availability and correctness of the network topology.

[0008] Preferably, indicators showing serious conflicts between the target task and other tasks are extracted from the conflict data analysis set, wherein the extracted indicators include the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks and the intensity of the bidirectional dependency relationship and the conflict between the target task and other tasks. Within a fixed monitoring window, after an in-depth analysis of the extracted indicators, a conflict chain depth reference value and a bidirectional conflict dependency reference value are generated respectively. The conflict feature vector of the task is formed by the conflict chain depth reference value and the bidirectional conflict dependency reference value to characterize the conflict pressure of the target task in the entire task network.

[0009] Preferably, a conflict feature vector consisting of a conflict chain depth reference value and a bidirectional conflict dependency reference value is input into a machine learning model that has been pre-trained based on historical collaborative tasks, and a conflict pressure coefficient is output through the machine learning model. Based on the conflict pressure coefficient, an intelligent assessment is performed on the conflict severity of the target task.

[0010] Preferably, the conflict pressure coefficient generated when evaluating the conflict severity of the target task through the machine learning model pre-trained based on the historical collaborative task is compared and analyzed with the pre-set reference threshold of the conflict pressure coefficient, and the conflict degree of the target task is further divided. The specific division steps are as follows: If the conflict pressure coefficient is greater than the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as severe; if the conflict pressure coefficient is less than or equal to the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as mild.

[0011] Preferably, when the target task is determined to be a heavy conflict task, the target task is automatically split into subtasks, and the granularity of the split, that is, the number of subtasks, is dynamically adjusted according to the specific value of the conflict pressure coefficient, using the following subtask division formula: ,in, Indicates the final number of subtasks into which the target task is divided; The number of basic subtasks, i.e. the minimum number of tasks to be divided; is the conflict pressure coefficient of the target task; A reference threshold value of the conflict pressure coefficient set for the system; It is the control coefficient of the partition granularity, which is used to control the impact of the conflict severity on the number of subtask partitions; is the conflict sensitivity index, which is used to control the nonlinear enhancement of the number of partitions as the conflict pressure increases; Indicates a round-up operation to ensure that the number of divisions is an integer.

[0012] Preferably, within a fixed monitoring window, the specific steps of generating a reference value of the conflict chain depth after in-depth analysis of the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks are as follows: In a fixed monitoring window, starting from the target task, recursively search all tasks that have direct or indirect conflict relationships with it in the task allocation network, and build a conflict dependency chain graph of the target task. In the conflict dependency chain graph, the length of each path is defined as the number of nodes on the path, and the length of the longest path is taken as the maximum depth of the conflict dependency chain of the target task; Based on the maximum depth of the target task conflict dependency chain and the branch expansibility of the conflict chain, a reference value of the conflict chain depth is calculated.

[0013] Preferably, within a fixed monitoring window, the specific steps of generating a bidirectional conflict dependency reference value after in-depth analysis of the bidirectional dependency relationship between the target task and other tasks and the intensity of the conflict between them are as follows: In a fixed monitoring window, for the target task, first identify the task set that has bidirectional dependencies with other task sets. For each pair of bidirectional dependencies, calculate the dependency strength score of the pair of bidirectional dependencies, which indicates the actual dependency tightness and conflict intensity between the tasks. The dependency strength scores of the target task and all its bidirectional dependent task pairs are coupled and superimposed, and the conflict diffusivity coefficient is introduced to form the bidirectional conflict dependency reference value of the target task, which is used to characterize its conflict pressure level in the entire task allocation network.

[0014] A cross-departmental collaborative efficiency dynamic optimization system, including a task allocation network construction module, a conflict data monitoring and collection module, a conflict feature extraction and vectorization module, a conflict severity intelligent assessment module, and a conflict classification optimization and task reconstruction module; The task allocation network construction module builds a task allocation network containing each task based on the basic information of cross-departmental collaborative tasks, and comprehensively describes the structural characteristics and interdependencies of tasks in the collaborative process; The conflict data monitoring and collection module sets a fixed monitoring window for each task during the cross-departmental collaborative operation, collects conflict data between the target task and other tasks in the collaborative process in real time, and builds a conflict data analysis set; The conflict feature extraction and vectorization module extracts indicators of serious conflicts between the target task and other tasks based on the conflict data analysis set. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. The conflict severity intelligent assessment module inputs the extracted conflict feature vector into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm, evaluates the conflict severity of the target task, and outputs the conflict degree score of the target task; The conflict classification optimization and task reconstruction module executes decision branch processing based on the conflict degree judgment results output by the machine learning model: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit is used to directly generate a collaborative optimization plan; when the conflict degree between the target task and other tasks is severe, the target task is divided into multiple independently schedulable sub-task units, and the number of sub-task divisions is adaptively adjusted according to the severity of the conflict.

[0015] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention comprehensively depicts the structural characteristics and dependencies of cross-departmental collaborative tasks based on the task allocation network, combines real-time conflict data, and quantifies and intelligently determines the degree of task conflict through a machine learning model, thereby achieving accurate classification and hierarchical processing of mild and severe conflicts. For mild conflicts, the system can quickly generate task optimization plans through an automatic optimization unit to improve collaborative flexibility and response speed; and for severe conflicts, the system can adaptively divide and reconstruct tasks into subtasks with controllable granularity based on the severity of the conflict, significantly alleviating the bottleneck problem caused by concentrated task conflicts, improving the flexibility of task scheduling and the optimization capability of global collaboration, thereby ensuring that cross-departmental collaboration can still maintain efficient and stable operation in a highly dynamic, multi-constrained complex environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0017] Figure 1 The present invention is a method flow chart of a method for dynamically optimizing cross-departmental collaboration efficiency.

[0018] Figure 2 A module schematic diagram of a cross-departmental collaboration efficiency dynamic optimization system of the present invention. DETAILED DESCRIPTION

[0019] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0020] The present invention provides Figure 1 A cross-departmental collaboration efficiency dynamic optimization method shown includes the following steps: Based on the basic information of cross-departmental collaborative tasks, a task allocation network containing each task is constructed to comprehensively describe the structural characteristics and interdependencies of tasks in the collaborative process; Step 1: Collect and organize basic information on cross-departmental collaboration tasks; First, we need to collect basic descriptions of each task, execution entities, required resources, expected duration, priority, and other information based on the business needs and project goals of each department, and standardize this information. Function: Lay the data foundation for subsequent graph structure modeling, ensuring that the system has a full and accurate grasp of the core attributes and constraints of each task.

[0021] Standardization processing refers to the unified formatting and processing of the collected task information to make it conform to certain standard specifications, which is convenient for subsequent system processing and analysis. Specifically, it includes: Standardization of task descriptions: Unify the way task descriptions are expressed to ensure that task names, types, objectives, and other information are clear and consistent.

[0022] Standardization of execution entities: Clarify and unify the role definitions of task execution entities (such as departments, teams, individuals, etc.) to ensure clear assignment of task responsibilities.

[0023] Standardization of resource requirements: uniformly classify and quantify the resources required for a task (such as manpower, materials, equipment, etc.) to facilitate resource allocation and scheduling.

[0024] Standardization of expected duration: The estimated completion time and duration of each task are expressed in unified units (such as days, hours, etc.) to ensure consistency of time information.

[0025] Priority standardization: Uniformly define and label the priorities of tasks to ensure clear distinction between priorities of tasks.

[0026] Standardized processing can unify the format and expression of task information, eliminate information inconsistencies or ambiguities, facilitate efficient system processing and analysis, improve the accuracy and efficiency of task scheduling, and provide accurate and consistent input data for subsequent task allocation, resource optimization and scheduling.

[0027] Step 2: Determine graph nodes and assign task identifiers; After obtaining the basic information of each task, define each task as a node in the graph, assign it a unique identifier, and record key information such as the executor, resource requirements, and planned time of the task in the node attributes. Function: Through clear node representation, the subsequent analysis and visualization of task relationships are more intuitive, which facilitates the system to quickly retrieve and manage tasks of various departments.

[0028] Step 3: Clarify the dependencies between tasks and build edges; After the nodes are established, corresponding associated edges are established for the task nodes that need to interact or connect according to constraints such as the order of precedence and post-order, resource sharing, and time window conflicts. For example, if task B can only be started after task A is completed, a directed edge from A to B is set in the graph. Function: Through the establishment of these edges, the system can accurately present the hierarchical structure, resource competition relationship, and time mutual exclusion relationship between tasks, providing a basis for conflict identification and subsequent optimization.

[0029] Step 4: Generate and store the task allocation network and complete consistency check; After integrating all nodes and edge information, a complete task allocation network is formed, and possible redundancy or circular dependencies in the network are checked and corrected to ensure the availability and correctness of the network topology. Function: In the final available network structure, all tasks of each department and their dependencies are clearly presented, providing high-quality basic support for subsequent conflict analysis, progress management, dynamic scheduling and other functions.

[0030] Graph modeling can comprehensively describe the structural characteristics and interdependencies of tasks in the collaborative process, clarify key constraints such as resource sharing, time windows, priority relationships, etc. among tasks, and provide structured basic information for subsequent conflict perception and optimization.

[0031] During the cross-departmental collaborative operation, a fixed monitoring window is set for each task to collect conflict data between the target task and other tasks in the collaborative process in real time to build a conflict data analysis set; A fixed monitoring window refers to a task life cycle interval that is pre-set for each task during cross-departmental collaboration, which is used to continuously monitor conflicts between the target task and other tasks during the collaboration process. This fixed monitoring window usually corresponds to the planned execution cycle, key nodes, or dependency time period of the task, ensuring that relevant conflict behavior data (such as resource conflicts, time conflicts, dependency conflicts, etc.) can be collected in real time and completely during the critical stages of the task. By setting a fixed monitoring window for the task, data redundancy and computing pressure caused by ineffective full-cycle monitoring can be avoided, while ensuring that the system focuses on conflict data collection and analysis during the most intensive and critical time periods of task conflicts, providing accurate and effective conflict data support for subsequent conflict severity quantification and optimized decision-making.

[0032] Based on the conflict data analysis set, the indicators of serious conflicts between the target task and other tasks are extracted. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. Indicators of serious conflicts between the target task and other tasks are extracted from the conflict data analysis set, where the extracted indicators include the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks and the intensity of the two-way dependency relationship and conflict between the target task and other tasks. Within a fixed monitoring window, after in-depth analysis of the extracted indicators, a conflict chain depth reference value and a two-way conflict dependency reference value are generated respectively. The conflict feature vector of the task is formed by the conflict chain depth reference value and the two-way conflict dependency reference value to characterize the conflict pressure of the target task in the entire task network.

[0033] The deeper the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks, the more serious the conflict of the target task is. This is because in cross-departmental collaboration, the conflict dependency chain reflects the conflict transmission relationship between the target task and other tasks in terms of time, resources or logic. The deeper the path, the conflict not only exists between directly adjacent tasks, but also continues to propagate and overlap in the task network through multiple levels and multiple links, forming a cascading conflict effect. The propagation of this conflict will cause more tasks to be affected in a chain reaction, resulting in a significant decrease in the flexibility of resource scheduling, time scheduling and task execution, and may even cause the failure or paralysis of the local collaboration network. Compared with single or shallow conflicts, deep path conflict chains are often difficult to alleviate through simple priority adjustments or resource reallocation. Therefore, the path depth of the conflict dependency chain is an important implicit indicator for measuring the severity of task conflicts and has significant engineering guidance value.

[0034] Within a fixed monitoring window, the specific steps for generating a reference value of the conflict chain depth after in-depth analysis of the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks are as follows: In a fixed monitoring window, starting from the target task, recursively search all tasks that have direct or indirect conflict relationships with it in the task allocation network, and build a conflict dependency chain graph of the target task. In the conflict dependency chain graph, the set of all paths from the target task to other conflicting tasks is recorded as , where each path represents a conflicting conducting link, It represents the total number of conflict dependency chains formed by the target task within a fixed monitoring window. The length of each path is defined as the number of nodes (tasks) on the path. The length of the longest path is taken as the maximum depth of the target task conflict dependency chain. The expression of the maximum depth of the target task conflict dependency chain is: ,in: For path The number of nodes, that is, the path length, The maximum depth of the conflict dependency chain of the target task; This step accurately identifies the “depth range” of the target task affected by the conflict in the collaborative task network through dependency chain modeling and deepest conflict chain extraction, providing a basis for subsequent quantification.

[0035] Maximum depth of dependency chain based on target-task conflict , and further combined with the branch expansibility of the conflict chain (i.e., the expansion speed of the conflict branch during the propagation process), the reference value of the conflict chain depth is calculated, and the calculation expression is: ,in: is the reference value of the conflict chain depth of the target task, is the conflict branch expansion coefficient, which represents the average branching factor of the conflict dependency chain starting from the target task during the propagation process. It is calculated as the average number of branches of all non-leaf nodes in the conflict chain. is the conflict sensitivity coefficient, which controls the contribution of branch expansion to , and its value can be set according to the system scale or scheduler design experience (such as ); The average number of branches of non-leaf nodes refers to the sum of the actual number of branches of all task nodes with downstream branches (that is, with one or more subtasks) in the conflict dependency chain of the target task, divided by the total number of these nodes. The average value is used to describe the diffusion speed and branch density of the conflict chain during the propagation process.

[0036] The conflict chain depth reference value not only takes into account the propagation depth of the conflict chain (the deeper the depth, the wider the impact range), but also introduces branch expansibility. That is, if a conflict dependency chain presents multi-branch diffusion (non-linear propagation) during propagation, the conflict chain depth reference value will be further amplified, truly reflecting the conflict diffusion pressure of the task and the conflict load within the task network.

[0037] It can be seen from the conflict chain depth reference value that, within a fixed monitoring window, the larger the performance value of the conflict chain depth reference value generated by in-depth analysis of the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks, the more serious the conflict between the target task and other tasks. The reason is that the conflict chain depth reference value comprehensively reflects two core factors: one is the depth of the conflict dependency chain (that is, the more layers of the conflict influence in the task network, the longer the task path involved, and the larger the scope of conflict diffusion); the second is the branch expansibility of the conflict (that is, the more branches the conflict has in the process of propagation, the faster it spreads, and the more complex the conflict network structure formed). When the conflict chain depth reference value is larger, it means that the conflict of the target task is not only limited to the local area, but also passed to more downstream tasks in terms of time, resources or dependencies, forming a multi-level and multi-branch conflict chain, resulting in a sharp increase in the complexity and coordination difficulty faced by task scheduling, resource allocation and collaborative arrangements, which seriously affects the overall executability and stability of the collaborative system; on the contrary, if the conflict chain depth reference value is small, it means that the conflicts are mostly shallow and small-scale conflicts within a local range, which can be easily alleviated through conventional task rescheduling or resource fine-tuning.

[0038] The higher the intensity of the bidirectional dependency between the target task and other tasks and the conflict between them, the more serious the conflict between the target task and other tasks. The essential reason is that the bidirectional dependency itself will form a highly coupled mutual constraint effect. When two or more tasks are mutually dependent (i.e. A depends on the result of B, and B depends on the resources or progress of A), a closed loop or bidirectional deadlock will be formed in the flow of resources, time and tasks. At this time, even if the system tries to optimize through simple priority adjustment or resource allocation, it often cannot break this mutual constraint relationship, resulting in severe limitation of the optimization space and easy to fall into local deadlock or system-level bottleneck. Therefore, bidirectional dependency conflict not only increases the severity and complexity of the conflict, but also directly aggravates the difficulty of task scheduling and collaborative optimization, and is a high-risk conflict structure in the collaborative network.

[0039] In a fixed monitoring window, the specific steps for generating a bidirectional conflict dependency reference value after in-depth analysis of the bidirectional dependency relationship and conflict intensity between the target task and other tasks are as follows: Within a fixed monitoring window, for the target task First, identify the task set that has bidirectional dependencies with other task sets and record this task set as , , that is, the target task A collection of tasks that are interdependent. Representation and target tasks There are tasks with bidirectional dependencies. For each pair of bidirectional dependencies , calculate the dependency strength score of the bidirectional dependency relationship , which indicates the actual degree of dependency and conflict between tasks. The calculation expression is: ,in: is the resource interaction degree, indicating the task With the task The number of resource types shared between is the dependency coupling degree, indicating the task With the task The strength of the dependency relationship can be expressed as the inverse of the dependency path length (the shorter the dependency, the stronger the coupling). is the buffer margin, indicating the task With the task The buffer duration is adjustable in time. The less buffer, the more serious the conflict. The dependency path length refers to the number of task nodes passed through by the dependency relationship from the target task to its dependent tasks in the task allocation network, that is, the number of task levels spanned by the shortest dependency chain between two tasks. The shorter the path, the tighter the dependency relationship between the two and the higher the coupling degree.

[0040] Adjustable buffer duration refers to the time margin that can be flexibly adjusted between the target task and its dependent tasks without affecting the overall progress of the project or task dependencies. That is, the available idle time window between the two that can be freely allocated and optimized by the scheduler.

[0041] This step models the conflict intensity between the target task and each of its bidirectional dependent task pairs to form a basic intensity score, which is used to measure the severity of the bidirectional conflict between a single pair of tasks.

[0042] The target task T The dependency strength scores of all bidirectional dependent task pairs are coupled and superimposed, and the conflict diffusion coefficient is introduced to form the bidirectional conflict dependency reference value of the target task, which is used to characterize its conflict pressure level in the entire task allocation network. The generation expression of the bidirectional conflict dependency reference value is: ,in: For bidirectional conflict dependent reference values, is the conflict diffusion coefficient, which indicates the target task The impact factor of the conflict propagation in the task network (which can be represented by the product of the out-degree and in-degree of the target task in the network, reflecting the ease of conflict propagation of the task), It is the conflict sensitivity amplification index (generally greater than 1), which is used to emphasize the impact of high-intensity conflicts and non-linearly increase the contribution of conflict severity to the index; The out-degree and in-degree of the target task in the network can be directly obtained by analyzing the number of dependent edges (out-degree) starting from the task and the number of dependent edges (in-degree) ending at the task in the task allocation network, which respectively represent the number of downstream tasks that the task depends on and the number of upstream tasks that it depends on.

[0043] This step generates a global conflict pressure index for the target task by coupling and aggregating the strength of multiple bidirectional conflict pairs and combining the ability of conflict to spread in the network. , reflecting its conflict centrality and potential influence in the collaborative task network.

[0044] It can be seen from the bidirectional conflict dependency reference value that, within a fixed monitoring window, the larger the performance value of the bidirectional conflict dependency reference value generated after in-depth analysis of the intensity of the bidirectional dependency relationship between the target task and other tasks and the conflict between them, the more serious the conflict between the target task and other tasks is, and vice versa. The reason is that the bidirectional conflict dependency reference value comprehensively reflects the degree of resource competition, dependency coupling, insufficient buffer margin, and conflict diffusion capacity between the target task and other tasks in structure by accumulating and nonlinearly amplifying the dependency intensity scores between the target task and all bidirectional dependent tasks. When the bidirectional conflict dependency reference value is larger, it means that the target task not only forms a close and difficult-to-relieve bidirectional dependency with multiple tasks, but also the conflict is highly intense and highly propagable in terms of time, resources, and network structure, causing the task to become a serious conflict center in the collaboration network; on the contrary, if the bidirectional conflict dependency reference value is small, it means that although the task may have a dependency relationship, the degree of conflict, scope of influence, or buffer capacity is relatively controllable, and the threat of conflict to the overall collaboration is low. Therefore, the bidirectional conflict dependency reference value can be effectively used as a criterion for measuring the severity of the conflict.

[0045] The extracted conflict feature vector is input into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm to evaluate the conflict severity of the target task and output the conflict degree score of the target task; The conflict feature vector composed of the conflict chain depth reference value and the bidirectional conflict dependency reference value is input into a machine learning model that has been pre-trained based on historical collaborative tasks. The conflict pressure coefficient is output by the machine learning model, and the conflict severity of the target task is intelligently evaluated based on the conflict pressure coefficient.

[0046] In the cross-departmental collaborative conflict optimization method, the machine learning model pre-trained based on historical collaborative tasks refers to the use of a large amount of historical project or collaborative task data as training samples before the system is deployed or actually operated, and the conflict severity assessment model is supervised or semi-supervised to enable it to have the ability to identify and quantify the relationship between conflict characteristics and conflict pressure between different tasks. Specifically, the system first extracts the structural attributes of each task (such as task dependency hierarchy, department, resource allocation, etc.), conflict records (such as resource contention events, task delay frequency, actual coordination times, etc.), and the final execution results of the task (such as whether it is delayed, whether it causes downstream chain problems, and the cost of conflict coordination, etc.) from the historically executed cross-departmental projects as the basic training data set. On this data set, the system combines the key conflict features such as the conflict chain depth reference value and the two-way conflict dependency reference value of the task to construct the corresponding conflict feature vector, and uses these vectors as model input, while using the "severity of conflict consequences" in the task execution results as a label or output target, such as using quantitative indicators such as conflict consequence severity score, conflict handling cost, or conflict propagation range as supervision signals. Through model training, the machine learning system can learn to automatically identify which feature combinations often correspond to high conflict pressure and which combinations can usually be quickly coordinated in complex, high-dimensional conflict scenarios, thereby forming an intelligent assessment model that can be generalized and applied to conflict judgment in new tasks.

[0047] The model can be trained using a variety of machine learning methods, including but not limited to: Random Forest, gradient boosting decision tree (such as XGBoost), support vector machine (SVM), graph neural network (GNN, used to process task dependency network graph), or multi-layer perceptron neural network (MLP). In particular, since collaborative tasks usually have graph structure characteristics and task dependencies naturally constitute a directed graph, the use of graph neural networks can better capture the structural conflict characteristics between tasks, such as the conflict chain propagation path, node centrality influence, etc. After the model training is completed, its parameters are fixed and can be called in real time as a runtime model. During the operation of cross-departmental collaborative tasks, the system only needs to extract the conflict feature vector of the current target task (such as the conflict chain depth reference value and bidirectional conflict dependency reference value of the current task) and pass it as input to the trained machine learning model, and a conflict pressure coefficient can be predicted in real time. The conflict pressure coefficient is a continuous or graded value that reflects the conflict pressure level of the current task in the system. The introduction of this machine learning model breaks through the technical bottleneck of inaccurate classification and inflexible handling of conflicting tasks by traditional static rule judgment or single priority strategy, and realizes the intelligent identification and optimized decision-making capabilities for conflict characteristics in cross-departmental collaboration.

[0048] The deep learning model is not specifically limited here, and can achieve the conflict chain depth reference value of the target task and bidirectional conflict dependent reference values Conduct comprehensive analysis to generate conflict pressure coefficients In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; the conflict pressure coefficient The generated expression is: , where , are the conflict chain depth reference values ​​of the target tasks respectively. and bidirectional conflict dependent reference values The preset scaling factor of , The preset proportionality coefficient refers to the value that the system uses artificially or based on experience to calculate the conflict pressure coefficient for each conflict characteristic index (such as the conflict chain depth reference value). and bidirectional conflict dependent reference values ) is the weight coefficient set by and Since different indicators contribute to different degrees to conflict pressure, in order to reasonably reflect the importance of each feature when comprehensively calculating conflict pressure, the system needs to pre-assign corresponding proportional coefficients to each feature during the modeling stage. For example, when experience or historical data shows that the impact of conflict chain depth on conflict pressure is relatively greater, If the bidirectional conflict dependency can better reflect the actual conflict propagation capability, then Assign a higher weight. The purpose of the preset proportional coefficient is to adjust the contribution ratio of each conflict feature in the calculation of the conflict pressure coefficient, so that the final calculated conflict pressure coefficient is more in line with the conflict severity determination requirements in the actual business scenario. , It can be obtained through domain expert experience, data analysis or machine learning optimization.

[0049] It can be seen from the conflict pressure coefficient that, within a fixed monitoring window, the greater the performance value of the conflict chain depth reference value generated after in-depth analysis of the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks, the greater the performance value of the two-way conflict dependency reference value generated after in-depth analysis of the intensity of the two-way dependency relationship and conflict between the target task and other tasks, that is, the greater the performance value of the conflict pressure coefficient generated when the conflict severity of the target task is evaluated by the machine learning model pre-trained based on historical collaborative tasks, the more serious the conflict between the target task and other tasks, and vice versa.

[0050] The conflict pressure coefficient generated when evaluating the conflict severity of the target task through the machine learning model pre-trained based on historical collaborative tasks is compared and analyzed with the pre-set reference threshold of the conflict pressure coefficient, and the conflict degree of the target task is further divided. The specific division steps are as follows: If the conflict pressure coefficient is greater than the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as severe; if the conflict pressure coefficient is less than or equal to the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as mild.

[0051] According to the conflict degree judgment result output by the machine learning model, the decision branch processing is executed: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit (such as task order rescheduling, resource priority allocation, buffer period adjustment, etc.) is used to directly generate a collaborative optimization plan; when the conflict degree between the target task and other tasks is severe, that is, there is a high degree of coupling, concentrated resource dependence or unsolvable time conflict between the task and multiple other tasks, the target task is divided into multiple independently schedulable subtask units. Through fine-grained task splitting, the dependence on the original conflict path is reduced to the greatest extent, and the number of subtask divisions is adaptively adjusted according to the severity of the conflict. The intensity of concentrated conflict is reduced through disassembly, and efficient dynamic optimization of task collaboration is achieved; When the conflict between the target task and other tasks is mild, it means that although there are certain conflicts such as resource contention, time overlap or dependency tension in the task during the collaboration process, the conflict has a small impact range and low intensity, which is not enough to cause serious obstacles to the overall collaboration process. In this case, the system does not need to make complex structural adjustments or split the tasks. Instead, it uses the automatic optimization unit to intelligently adopt common optimization methods such as task reordering, resource priority allocation, and buffer period adjustment based on the scheduling information of the current task and its related tasks. Task reordering refers to adjusting the execution order of tasks to alleviate resource or time conflicts without violating the previous and subsequent dependencies; resource priority allocation refers to allocating idle or low-load resources to mildly conflicting tasks to reduce the risk of contention; buffer period adjustment refers to appropriately extending or shortening the execution buffer time of tasks to increase scheduling flexibility. Through the combined application of the above optimization strategies, mild conflict problems can be solved quickly and at low cost, so that tasks can be smoothly integrated into the collaboration plan without introducing major structural changes, ensuring the smooth progress of cross-departmental collaboration.

[0052] The division process considers the logical divisibility, resource independence and time parallelism within the task. This means that when a heavily conflicting target task is divided into multiple subtasks, the system needs to comprehensively analyze the structural characteristics within the task to ensure that the divided subtasks have relatively independent and schedulable capabilities. Logical divisibility refers to whether the task itself can be disassembled into several subtasks with clear functional divisions or stage divisions, and each subtask can jointly meet the goal of the original task after completion; resource independence means that during the execution of the subtask, it is best to avoid continuing to rely on the same limited or tight key resources, thereby reducing secondary conflicts caused by resource contention; time parallelism refers to whether there is a possibility that the divided subtasks can be executed in parallel or partially overlapped, so as to improve the overall scheduling freedom through flexible time scheduling. By comprehensively considering these three factors, the structure of the original task can be reasonably refined while ensuring the quality of task completion, so that the divided subtasks can be more easily optimized and scheduled independently in the collaborative network, thereby effectively reducing the local bottlenecks and global collaboration pressure caused by heavy conflicts.

[0053] When the target task is judged as a heavy conflict task, the target task is automatically split into subtasks, and the granularity of the split (i.e. the number of subtasks) is dynamically adjusted according to the specific value of the conflict pressure coefficient. The following subtask division formula is used: ,in, Indicates the final number of subtasks into which the target task is divided; The number of basic subtasks, i.e., the minimum number of tasks to be divided (for example, it can be initially set to 2-3); The conflict pressure coefficient of the target task; A reference threshold value of the conflict pressure coefficient set for the system; is the partition granularity control coefficient (positive value), which is used to control the impact of conflict severity on the number of subtasks. When the conflict is more serious, the partition granularity control coefficient can increase the sensitivity of the increase in the number of subtasks, thereby achieving adaptive enhancement of the task partition granularity. is the conflict sensitivity index (greater than 1), which is used to control the nonlinear enhancement of the number of partitions as the conflict pressure increases; Indicates rounding up operation to ensure that the number of divisions is an integer; The above adaptive partitioning mechanism can ensure that: when the conflict pressure coefficient is only slightly higher than the reference threshold of the conflict pressure coefficient, the number of task partitions is close to the basic number of partitions. ; When the conflict pressure coefficient is much higher than the reference threshold of the conflict pressure coefficient, the number of subtask divisions increases significantly, achieving refined splitting under high conflict intensity. This step can transform severe conflict pressure into multiple low-conflict, independently schedulable subtasks, reduce local bottleneck effects, and form higher scheduling flexibility and optimizability in the task allocation network, providing a good structural foundation for subsequent resource allocation and progress coordination.

[0054] The above solution can effectively solve the problem of insufficient optimization capability in the existing technology when there are serious conflicts between key tasks and multiple related tasks. Specifically, the system can comprehensively characterize the structural characteristics and dependencies of cross-departmental collaborative tasks based on the task allocation network, and quantify and intelligently determine the degree of task conflict through machine learning models in combination with real-time conflict data, thereby achieving accurate classification and hierarchical processing of mild and severe conflicts. For mild conflicts, the system can quickly generate task optimization plans through automatic optimization units to improve collaboration flexibility and response speed; and for severe conflicts, the system can adaptively divide and reconstruct tasks into granularly controllable subtasks based on the severity of the conflict, significantly alleviating the bottleneck problem caused by concentrated task conflicts, improving the flexibility of task scheduling and the optimization capability of global collaboration, thereby ensuring that cross-departmental collaboration can still maintain efficient and stable operation in a highly dynamic, multi-constrained complex environment.

[0055] The present invention provides Figure 2 A cross-departmental collaborative efficiency dynamic optimization system is shown, comprising a task allocation network construction module, a conflict data monitoring and collection module, a conflict feature extraction and vectorization module, a conflict severity intelligent assessment module, and a conflict classification optimization and task reconstruction module; The task allocation network construction module builds a task allocation network containing each task based on the basic information of cross-departmental collaborative tasks, and comprehensively describes the structural characteristics and interdependencies of tasks in the collaborative process; The conflict data monitoring and collection module sets a fixed monitoring window for each task during the cross-departmental collaborative operation, collects conflict data between the target task and other tasks in the collaborative process in real time, and builds a conflict data analysis set; The conflict feature extraction and vectorization module extracts indicators of serious conflicts between the target task and other tasks based on the conflict data analysis set. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. The conflict severity intelligent assessment module inputs the extracted conflict feature vector into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm, evaluates the conflict severity of the target task, and outputs the conflict degree score of the target task; The conflict classification optimization and task reconstruction module executes decision branch processing based on the conflict degree judgment results output by the machine learning model: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit is used to directly generate a collaborative optimization plan; when the conflict degree between the target task and other tasks is severe, the target task is divided into multiple independently schedulable sub-task units, and the number of sub-task divisions is adaptively adjusted according to the severity of the conflict.

[0056] A method for dynamically optimizing cross-departmental collaboration efficiency provided by an embodiment of the present invention is implemented through the above-mentioned system for dynamically optimizing cross-departmental collaboration efficiency. The specific methods and processes of a system for dynamically optimizing cross-departmental collaboration efficiency are detailed in the embodiment of the above-mentioned method for dynamically optimizing cross-departmental collaboration efficiency, which will not be repeated here.

[0057] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0058] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0059] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for dynamically optimizing cross-departmental collaboration efficiency, characterized in that: The following steps are involved: Based on the basic information of cross-departmental collaborative tasks, a task allocation network containing each task is constructed to comprehensively describe the structural characteristics and interdependencies of tasks in the collaborative process; During the cross-departmental collaborative operation, a fixed monitoring window is set for each task to collect conflict data between the target task and other tasks in the collaborative process in real time to build a conflict data analysis set; Based on the conflict data analysis set, the indicators of serious conflicts between the target task and other tasks are extracted. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. The extracted conflict feature vector is input into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm to evaluate the conflict severity of the target task and output the conflict degree score of the target task; According to the conflict degree judgment result output by the machine learning model, the decision branch processing is executed: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit is used to directly generate a collaborative optimization plan; When the conflict between the target task and other tasks is severe, the target task is divided into multiple independently schedulable subtask units, and the number of subtask divisions is adaptively adjusted according to the severity of the conflict.

2. A method for dynamic optimization of cross-departmental collaboration efficiency according to claim 1, characterized in that: Based on the basic information of cross-departmental collaborative tasks, a task allocation network containing various tasks is constructed. The specific steps are as follows: It is necessary to collect basic information for each task and standardize the information according to the business needs and project goals of each department; After obtaining the basic information of each task, define each task as a node in the graph, assign it a unique identifier, and record the key information of the task in the node attributes; After the nodes are established, corresponding associated edges are established for the task nodes that need to interact or connect according to the constraints; After integrating all node and edge information, a complete task allocation network is formed, and the redundancy or circular dependency in the network is checked and corrected.

3. The method for dynamic optimization of cross-departmental collaboration efficiency according to claim 1, characterized in that: Indicators of serious conflicts between the target task and other tasks are extracted from the conflict data analysis set, where the extracted indicators include the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks and the intensity of the two-way dependency relationship and conflict between the target task and other tasks. Within a fixed monitoring window, after in-depth analysis of the extracted indicators, a conflict chain depth reference value and a two-way conflict dependency reference value are generated respectively. The conflict feature vector of the task is formed by the conflict chain depth reference value and the two-way conflict dependency reference value to characterize the conflict pressure of the target task in the entire task network.

4. A method for dynamic optimization of cross-departmental collaboration efficiency according to claim 3, characterized in that: The conflict feature vector composed of the conflict chain depth reference value and the bidirectional conflict dependency reference value is input into a machine learning model that has been pre-trained based on historical collaborative tasks. The conflict pressure coefficient is output by the machine learning model, and the conflict severity of the target task is intelligently evaluated based on the conflict pressure coefficient.

5. A method for dynamically optimizing cross-departmental collaboration efficiency according to claim 4, characterized in that: The conflict pressure coefficient generated when evaluating the conflict severity of the target task through the machine learning model pre-trained based on historical collaborative tasks is compared and analyzed with the pre-set reference threshold of the conflict pressure coefficient, and the conflict degree of the target task is further divided. The specific division steps are as follows: If the conflict pressure coefficient is greater than the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as severe; if the conflict pressure coefficient is less than or equal to the conflict pressure coefficient reference threshold, the conflict degree between the target task and other tasks is classified as mild.

6. A method for dynamic optimization of cross-departmental collaboration efficiency according to claim 5, characterized in that: When the target task is judged as a heavy conflict task, the target task is automatically split into subtasks, and the granularity of the split, that is, the number of subtasks, is dynamically adjusted according to the specific value of the conflict pressure coefficient. The following subtask division formula is used: ,in, Indicates the final number of subtasks into which the target task is divided; The number of basic subtasks, i.e. the minimum number of tasks to be divided; is the conflict pressure coefficient of the target task; A reference threshold value of the conflict pressure coefficient set for the system; It is the control coefficient of the partition granularity, which is used to control the impact of the conflict severity on the number of subtask partitions; is the conflict sensitivity index, which is used to control the nonlinear enhancement of the number of partitions as the conflict pressure increases; Indicates a round-up operation to ensure that the number of divisions is an integer.

7. The method for dynamic optimization of cross-departmental collaboration efficiency according to claim 3, characterized in that: Within a fixed monitoring window, the specific steps for generating a reference value of the conflict chain depth after in-depth analysis of the longest path depth of the conflict dependency chain formed between the target task and other conflicting tasks are as follows: In a fixed monitoring window, starting from the target task, recursively search all tasks that have direct or indirect conflict relationships with it in the task allocation network, and build a conflict dependency chain graph of the target task. In the conflict dependency chain graph, the length of each path is defined as the number of nodes on the path, and the length of the longest path is taken as the maximum depth of the conflict dependency chain of the target task; Based on the maximum depth of the target task conflict dependency chain and the branch expansibility of the conflict chain, a reference value of the conflict chain depth is calculated.

8. The method for dynamic optimization of cross-departmental collaboration efficiency according to claim 3, characterized in that: In a fixed monitoring window, the specific steps for generating a bidirectional conflict dependency reference value after in-depth analysis of the bidirectional dependency relationship and conflict intensity between the target task and other tasks are as follows: In a fixed monitoring window, for the target task, first identify the task set that has bidirectional dependencies with other task sets. For each pair of bidirectional dependencies, calculate the dependency strength score of the pair of bidirectional dependencies, which indicates the actual dependency tightness and conflict intensity between the tasks. The dependency strength scores of the target task and all its bidirectional dependent task pairs are coupled and superimposed, and the conflict diffusivity coefficient is introduced to form the bidirectional conflict dependency reference value of the target task, which is used to characterize its conflict pressure level in the entire task allocation network.

9. A cross-departmental collaboration efficiency dynamic optimization system, used to implement the cross-departmental collaboration efficiency dynamic optimization method described in any one of claims 1 to 8, characterized in that: It includes task allocation network construction module, conflict data monitoring and collection module, conflict feature extraction and vectorization module, conflict severity intelligent assessment module, and conflict classification optimization and task reconstruction module; The task allocation network construction module builds a task allocation network containing each task based on the basic information of cross-departmental collaborative tasks, and comprehensively describes the structural characteristics and interdependencies of tasks in the collaborative process; The conflict data monitoring and collection module sets a fixed monitoring window for each task during the cross-departmental collaborative operation, collects conflict data between the target task and other tasks in the collaborative process in real time, and builds a conflict data analysis set; The conflict feature extraction and vectorization module extracts indicators of serious conflicts between the target task and other tasks based on the conflict data analysis set. After in-depth analysis of the extracted indicators within a fixed monitoring window, the conflict feature vector of the task is formed to characterize the conflict pressure of the target task in the entire task network. The conflict severity intelligent assessment module inputs the extracted conflict feature vector into a machine learning model that is pre-trained based on historical collaborative tasks and using a supervised learning algorithm, evaluates the conflict severity of the target task, and outputs the conflict degree score of the target task; The conflict classification optimization and task reconstruction module performs decision branch processing based on the conflict degree judgment results output by the machine learning model: when the conflict degree between the target task and other tasks is mild, the automatic optimization unit is used to directly generate a collaborative optimization plan; When the conflict between the target task and other tasks is severe, the target task is divided into multiple independently schedulable subtask units, and the number of subtask divisions is adaptively adjusted according to the severity of the conflict.

Images