Cooperative office management system and method based on big data

Abstract

The invention relates to the technical field of office management, in particular to a collaborative office management system and method based on big data, in which task allocation, resource use and approval behaviors are collected and fused, a graph network with a heterogeneous structure is established, node features are aggregated by using the graph neural network, and high-dimensional collaborative behavior representation is formed. According to the method, a structured data expression system is constructed, fine modeling of complex cooperative behaviors is realized, through an isolated forest algorithm, rare detection is performed on each node in a graph in time and authority dimensions, active discovery of a task conflict path and an authority conflict path is realized, the abnormal behavior identification efficiency in a task cooperative process is improved, and the task cooperative behavior identification efficiency is improved. Through combination of abnormal path and conflict node analysis, authority border crossing and process bypassing behaviors are dynamically identified, and an intervention mechanism is triggered, so that diffusion of illegal behaviors in a system is effectively prevented, and the automatic processing capability, behavior prediction capability and execution security of a collaborative office process are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of office management technology, and in particular to a collaborative office management system and method based on big data. Background Art

[0002] The field of office management technology aims to manage and coordinate office activities such as task processes, personnel division of labor, resource scheduling, authority control and transaction execution in an organization in an efficient, orderly and controllable manner through systematic information processing methods. The goal is to improve office efficiency, enhance the controllability and traceability of task execution, achieve orderly multi-role collaboration, promote information sharing and decision support, and reduce management costs.

[0003] A collaborative office management system based on big data collects, archives, analyzes and models various office data generated within the organization to achieve centralized management of office behaviors such as task allocation, personnel scheduling, progress control, and resource coordination. The purpose is to improve cross-departmental collaboration efficiency, reduce information island effects, enhance the controllability and traceability of task execution, and realize intelligent decision-making support for office behaviors.

[0004] Existing technologies are unable to break down the relationship between behaviors at the task execution level, resulting in the inability to structure the collaborative dependencies between tasks, making it difficult to warn of potential conflicts during the collaboration process. In addition, there is a lack of clear modeling of the interaction path between task authority configuration and resource scheduling. Task unauthorized operations and repeated resource configuration are easily overlooked in complex processes, resulting in the same approval process being repeatedly executed and omitted for review, affecting the integrity of the decision. In addition, in multi-task and high-concurrency scenarios, resource competition, path blockages, and extended task waiting times often occur, reducing office processing efficiency and weakening the ability to respond to emergencies. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a collaborative office management system and method based on big data.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solutions: A collaborative office management system based on big data includes:

[0007] Collaborative behavior modeling module: Based on historical office data and user interaction behaviors, a graph neural network is used to construct a heterogeneous graph structure of collaborative office behaviors. Task allocation, resource utilization, and approval behaviors are used as nodes to form graph edges. By aggregating historical behavior features, a collaborative behavior network diagram is obtained.

[0008] Conflict path identification module: Based on the collaborative behavior network diagram, the module uses the isolation forest to compare the time overlap of tasks and resources, check the cross-authority of roles, identify task execution conflicts and authority overlap conflicts through comparison between nodes, and generate potential conflict paths;

[0009] Behavior consistency check module: Based on the potential conflict path, it performs real-time comparison of task state transitions with global execution constraints. By checking whether the execution order of tasks complies with the specified constraints, it marks inconsistent tasks and generates an execution order anomaly result.

[0010] Task scheduling and path optimization module: Based on the execution order anomaly results, it dynamically schedules tasks, selects qualified paths according to the resource requirements and priorities of the tasks, detects path conflicts, rearranges the task order according to feasible paths, and generates optimized task paths;

[0011] Security control module: Based on the abnormal results of the execution sequence and potential conflict paths, it checks the unauthorized operations and task bypass behaviors, triggers the dynamic alarm and intervention mechanism, and generates security intervention results by identifying and preventing violations.

[0012] As a further solution of the present invention, the collaborative behavior modeling module includes:

[0013] Graph structure construction submodule: Based on historical office data and user interaction behavior, by analyzing task allocation, resource usage, and approval behavior, a graph neural network is used to construct task nodes, role nodes, and resource nodes, and connect them to form edges between nodes, generating a collaborative behavior graph structure.

[0014] Node feature aggregation submodule: Based on the collaborative behavior graph structure, by weighted aggregation of the adjacent node features of each node, the hidden state of each node is updated, the task execution frequency and resource occupancy time of the node are aggregated, and the node hidden state representation is formed;

[0015] Collaborative behavior network generation submodule: Based on the node hidden state representation, combined with the relationship between each node and the historical behavior characteristics, the entire collaborative network is constructed to generate a collaborative behavior network graph.

[0016] As a further solution of the present invention, the graph neural network is according to the formula:

[0017]

[0018] in: represents the feature vector of node v in the kth layer in the collaborative office management system, σ represents the activation function, α v represents the task priority coefficient of node v, W (k)represents the weight matrix of the kth layer, represents the set of adjacent nodes connected to node v, β vu represents the behavioral influence coefficient between node v and adjacent node u, represents the feature vector of the adjacent node u in the k-1 layer, γ v represents the resource availability adjustment item of node v, b (k) represents the bias term of the kth layer.

[0019] As a further solution of the present invention, the conflict path identification module includes:

[0020] Time overlap comparison submodule: Based on the collaborative behavior network diagram, the isolation forest is used to compare the execution time of each task with the time interval of the occupied resources, detect the time overlap between tasks, identify the task combinations with time conflicts, and generate a task time conflict set;

[0021] The authority overlap detection submodule analyzes the authority scope of each role based on the collaborative behavior network diagram, determines whether there is authority overlap between roles, identifies the overlapping parts of multiple role permissions, and generates a role authority conflict set.

[0022] Potential conflict path generation submodule: Based on the task time conflict set and the role authority conflict set, the task paths are combined, and the potential conflict paths are identified and generated by judging the execution order of the conflicting tasks and the resource competition situation.

[0023] As a further embodiment of the present invention, the isolation forest is based on the formula:

[0024]

[0025] Where: S(x) represents the anomaly score of sample x, H(x) represents the path depth of sample x being split in the isolation forest, T r represents the time overlap of task r, P represents the cross-authority score of role, α r The weight coefficient of task time overlap, β p The weight coefficient representing the cross-functionality of role permissions, c(n) represents the dataset size constant, and n represents the number of samples.

[0026] As a further solution of the present invention, the scope of authority of each role is analyzed. First, the scope of authority of each role is determined, including the type of resources accessed, the type of operations that can be performed, and the timeliness of the authority. Then, whether there is any overlap in the role permissions is analyzed. The overlap judgment is based on the complete overlap and partial overlap of the permission items, and is compared according to the resources and operations involved in the role. When duplicate roles have the same and overlapping operation permissions on the same resource, potential conflicts and mutual exclusion situations are identified.

[0027] As a further solution of the present invention, the behavior consistency checking module includes:

[0028] State transition comparison submodule: Based on the potential conflict path, by checking the current state and target state of each task one by one, comparing whether the state transition of the task complies with the predetermined process rules, determining whether there is any state inconsistency, and generating a state transition consistency result;

[0029] Sequence constraint check submodule: Based on the state transition consistency result, compare the order of task execution, check whether the task execution is performed in the set order, and generate a sequence consistency check result;

[0030] Execution order anomaly marking submodule: based on the sequence consistency check result, marks the tasks that violate the predetermined execution order, records the specific identification of each execution order anomaly task, and generates an execution order anomaly result.

[0031] As a further solution of the present invention, the task scheduling and path optimization module includes:

[0032] Task path selection submodule: Based on the execution order anomaly results, the module selects feasible task paths by evaluating the resource requirements and priorities of each task, and generates task path selection results by considering the resource allocation of each path;

[0033] Path conflict detection submodule: Based on the task path selection result, perform conflict detection on the selected path, check whether there is resource contention and execution order conflict in the path, record the conflicts found, and generate path conflict detection results;

[0034] Task path optimization submodule: Based on the path conflict detection result, the task path is optimized, the task execution order is rearranged, the resource conflicts in the path are eliminated, and the optimized task path is generated by rearranging the task path.

[0035] As a further solution of the present invention, the security control module includes:

[0036] Illegal behavior detection submodule: Based on the abnormal results of the execution sequence and potential conflict paths, the module compares the execution permissions of each task with the actual operation to check whether there are any unauthorized operations or task bypass behaviors, analyzes whether each task complies with the role permissions, and generates illegal behavior detection results;

[0037] Dynamic alarm triggering submodule: Based on the violation detection results, it marks the identified violation tasks and triggers the alarm mechanism, while recording the detailed information of the violation and generating a violation alarm signal;

[0038] Security intervention execution submodule: Based on the violation alarm signal, intervene in the violation operation, immediately stop the execution of the violation task, roll back the executed operation, and generate a security intervention result.

[0039] A collaborative office management method based on big data is implemented based on the above-mentioned collaborative office management system based on big data, and includes the following steps:

[0040] S1: Based on historical office data and user interaction behaviors, we extract each user's task assignment data, resource usage records, and approval behaviors to form a graph structure containing nodes. Nodes include tasks, resources, and roles, and edges represent task assignments, resource usage, and approval relationships. We then aggregate historical behavior features through a graph neural network, combining the time and resource usage in user interaction behaviors to generate a collaborative behavior network diagram.

[0041] S2: Based on the collaborative behavior network diagram, the isolation forest algorithm is used to compare the time overlap between tasks and resources. The task execution time intervals are extracted and compared with the time intervals of the occupied resources to identify time conflicts between tasks. At the same time, the authority scope of the roles is analyzed and the isolation forest algorithm is used to determine the overlap of authority between roles. The areas where multiple roles' authority overlap are identified and potential conflict paths are generated.

[0042] S3: Based on the potential conflict path, check the execution order of each task one by one, extract the state transition information of each task, compare the state change of task execution with the global constraint conditions, determine whether the task execution order meets the specified constraints, mark the tasks with inconsistent states, and generate an execution order abnormality result;

[0043] S4: Based on the execution order anomaly results, dynamically schedule the tasks marked as abnormal, select a feasible path based on the resource requirements, priority, and execution order of each task, detect possible conflicts in the task path, and if conflicts are found, rearrange the task execution order according to the feasible path to generate an optimized task path;

[0044] S5: Based on the optimized task path, perform security control on the task, analyze whether there are any unauthorized operations and bypass behaviors during the task execution process, and strengthen the authority verification of the task. At the same time, by comparing the role permissions and task content in the task path, identify and prevent non-compliant operations, and trigger a dynamic alarm mechanism based on the inspection results, and execute corresponding intervention measures to generate security intervention results.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] 1. This invention collects and integrates task allocation, resource utilization, and approval behaviors to establish a heterogeneous graph network. This graph clearly expresses the actual relationships between office activities. Graph neural networks are then used to aggregate node features to form a high-dimensional representation of collaborative behavior. This structured data representation system enables detailed modeling of complex collaborative behaviors.

[0047] 2. In this invention, the isolation forest algorithm is used to detect the rarity of each node in the graph in terms of time and authority, enabling the active discovery of task conflict paths and authority conflict paths, thereby improving the efficiency of identifying abnormal behavior during task collaboration.

[0048] 3. In the present invention, by combining abnormal path and conflict node analysis, it dynamically identifies authority violations and process bypass behaviors, and triggers intervention mechanisms to effectively prevent the spread of illegal behaviors in the system, thereby enhancing the automated processing capabilities, behavior prediction capabilities and execution security of the collaborative office process. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a system flow chart of the present invention;

[0050] Figure 2 Schematic diagram of the system framework of the present invention;

[0051] Figure 3 Schematic diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0053] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0054] See also Figure 1 The present invention provides a technical solution: a collaborative office management system based on big data, comprising:

[0055] Collaborative behavior modeling module: Based on historical office data and user interaction behaviors, a graph neural network is used to construct a heterogeneous graph structure of collaborative office behaviors. Task allocation, resource utilization, and approval behaviors are used as nodes to form graph edges. By aggregating historical behavior features, a collaborative behavior network diagram is obtained.

[0056] Conflict Path Identification Module: Based on the collaborative behavior network diagram, the module uses the isolation forest to compare the temporal overlap of tasks and resources, check the cross-authority of roles, identify task execution conflicts and authority overlap conflicts through node comparison, and generate potential conflict paths.

[0057] Behavior consistency check module: Based on potential conflict paths, it performs real-time comparison of task state transitions with global execution constraints. By checking whether the execution order of tasks complies with the specified constraints, it marks inconsistent tasks and generates execution order anomaly results.

[0058] Task Scheduling and Path Optimization Module: Dynamically schedules tasks based on execution order anomalies, selects eligible paths based on task resource requirements and priorities, detects path conflicts, rearranges task order based on feasible paths, and generates optimized task paths.

[0059] Security control module: Based on abnormal execution sequence results and potential conflict paths, it checks unauthorized operations and task bypass behaviors, triggers dynamic alarm and intervention mechanisms, and generates security intervention results by identifying and preventing violations.

[0060] See also Figure 2 , the collaborative behavior modeling module includes:

[0061] Graph structure construction submodule: Based on historical office data and user interaction behavior, by analyzing task allocation, resource usage, and approval behavior, a graph neural network is used to construct task nodes, role nodes, and resource nodes, and connect them to form edges between nodes, generating a collaborative behavior graph structure.

[0062] Node feature aggregation submodule: Based on the collaborative behavior graph structure, it updates the hidden state of each node by weighted aggregation of the adjacent node features of each node, aggregates the task execution frequency and resource occupancy time of the node, and forms the node hidden state representation;

[0063] Collaborative behavior network generation submodule: Based on the node hidden state representation, combined with the relationship between each node and historical behavior characteristics, the entire collaborative network is constructed to generate a collaborative behavior network graph;

[0064] Graph structure construction submodule: Based on office history data and user interaction behavior, it analyzes task allocation, resource usage, and approval behavior. Using graph neural networks, it first defines task nodes, role nodes, and resource nodes, and sets feature vectors for each node. The feature vector of the task node includes task ID, task category, and task duration. The feature vector of the role node includes role ID, role authority, and role task execution frequency. The feature vector of the resource node includes resource ID, resource type, and resource occupancy rate. Then, using a message passing mechanism, the feature vector of each node is weighted averaged with the feature vectors of adjacent nodes. The weight is determined by the similarity of the feature vectors of adjacent nodes. Cosine similarity is used for calculation, and the similarity calculation result is used as the weight of the edge. During the calculation process, the edge weight is adjusted according to the following parameters: the similarity threshold is 0.5, and the L2 norm of the feature vector is normalized. Finally, the edge is generated by weighted average calculation to construct the entire collaborative behavior graph structure.

[0065] Node feature aggregation submodule: Based on the collaborative behavior graph structure, the adjacent node features of each node are weighted and aggregated to aggregate the adjacent node features of the node and update the node's hidden state. In the aggregation process of each layer, the node's feature matrix is ​​calculated with the adjacency matrix. The adjacency matrix uses the normalized Laplace matrix. The matrix is ​​obtained by standardizing the original adjacency matrix, and the weights are normalized using the L2 norm. In the node aggregation process, the task execution frequency is counted by the number of executions in the historical task data, and the resource occupation time is aggregated using the duration field in the historical resource usage log. After the calculation is completed, the node's hidden state is updated by weighted average and the feature value of the adjacent node, ultimately forming the hidden state representation of each node;

[0066] Collaborative behavior network generation submodule: Based on the node hidden state representation, combined with the relationship between each node and the historical behavior characteristics, a graph attention network is used to perform weighted calculation on the hidden state representation of each node, and the attention mechanism is used to calculate the weight coefficient between each pair of nodes. First, the similarity between each pair of nodes is calculated, and LeakyReLU is used as the activation function to process the similarity. The activation coefficient is set to a negative threshold of 0.2, and then softmax is used to normalize the weight coefficients of all adjacent nodes; during the calculation process, the node's feature vector will take into account the node's historical behavior data. The historical task completion status and resource usage statistics of each node will be included in the weighted calculation as the node's behavioral characteristics to generate the final weight coefficient.

[0067] See also Figure 2 , graph neural network, according to the formula:

[0068]

[0069] in: represents the feature vector of node v in the kth layer in the collaborative office management system, σ represents the activation function, α v represents the task priority coefficient of node v, W (k) represents the weight matrix of the kth layer, represents the set of adjacent nodes connected to node v, β vu represents the behavioral influence coefficient between node v and adjacent node u, represents the feature vector of the adjacent node u in the k-1 layer, γ v represents the resource availability adjustment item of node v, b (k) represents the bias term of the kth layer;

[0070] Execution process: First, for each task node, role node and resource node, extract the task urgency, task completion time limit and task priority from the office history data, and calculate the task priority coefficient α after standardization. v ,Then, the interaction frequency between the node and the adjacent nodes in the historical office behavior is counted, and the behavior influence coefficient β is obtained by normalization method. vu , dynamically calculate the resource availability adjustment item γ based on the ratio between the total amount of resources currently available and the actual demand associated with the node v Then, the above parameters are combined with the preset k-th layer weight matrix W (k) , bias term b (k) and the feature vector of the previous layer of the adjacent node Substitute the formula to complete the current layer feature vector of each node Calculation and output node features.

[0071] See also Figure 2 , the conflict path identification module includes:

[0072] Time overlap comparison submodule: Based on the collaborative behavior network graph, the isolation forest is used to compare the execution time of each task with the time interval of the occupied resources, detect the time overlap between tasks, identify task combinations with time conflicts, and generate a task time conflict set;

[0073] The permission overlap detection submodule analyzes the permission scope of each role based on the collaborative behavior network diagram, determines whether there is permission overlap between roles, identifies the overlapping parts of multiple role permissions, and generates a role permission conflict set.

[0074] Potential conflict path generation submodule: Based on the task time conflict set and the role authority conflict set, the task paths are combined, and the execution order of the conflicting tasks and the resource competition situation are judged to identify and generate potential conflict paths;

[0075] Time overlap comparison submodule: Based on the collaborative behavior network diagram, the isolation forest algorithm is used to first extract the execution time of each task and the time interval of the occupied resources. The task execution time is then processed by time window division. The time window size is set to 5 minutes, and the time interval of each task is divided. Then, using the training set in the isolation forest algorithm, the training model is trained to learn the normal range of each task execution time and resource occupation time. The time information in the training data is used for modeling to obtain the abnormal pattern of task execution time. At the same time, the time overlap between tasks is detected and the time overlap degree of tasks is evaluated. If the overlap degree is greater than the set threshold, it is considered that there is a time conflict between the two tasks, and a task time conflict set is generated;

[0076] Permission intersection detection submodule: Based on the collaborative behavior network diagram, the permission scope of each role is analyzed. Using set operations, the permission scope of each role is first extracted. The permission scope is obtained through the role's task permission table, which contains the task ID, role ID, and permission identifier. Then, the permissions of all roles are compared using set intersection operations. The intersection threshold is set to 50%. If the permission intersection of two roles is greater than the threshold, it is considered that there is permission intersection between the two roles. The number of intersections is used as the weight to generate a role permission conflict set. The roles are further sorted according to the conflict weight, and role combinations with higher conflict levels are identified first.

[0077] Potential conflict path generation submodule: Based on the task time conflict set and the role authority conflict set, the task paths are combined and the depth-first search algorithm is used. First, a task path graph is constructed, each task is used as a node, and the time conflicts and role authority conflicts between tasks are connected as edges. The search depth is set to 5. When performing the search, the execution order of the conflicting tasks is determined. The task sorting rules are used to sort the tasks according to the start and end time. At the same time, the resource competition situation is considered. The resource occupancy of each task path is compared and the resource competition algorithm is used to calculate the resource occupancy overlap in the task path. If the overlap exceeds the set threshold, it is considered that there is a potential conflict in the path and a potential conflict path is generated.

[0078] See also Figure 2 , Isolation Forest, according to the formula:

[0079]

[0080] Where: S(x) represents the anomaly score of sample x, H(x) represents the path depth of sample x being split in the isolation forest, T r represents the time overlap of task r, P represents the cross-authority score of role, α r The weight coefficient of task time overlap, βp The weight coefficient representing the cross-functionality of role permissions, c(n) represents the dataset size constant, and n represents the number of samples;

[0081] Execution process: For sample x, calculate the path depth H(x) in the isolation forest, that is, the number of layers the sample is split into. The depth value is used to reflect the abnormality of the sample. Then, compare the time overlap T of task r. r , that is, the overlap between the execution time of the task and the available time of the resource, identify the possible time conflict, and then calculate the role authority intersection score P. By analyzing the intersection between the role authority and the task role, it is determined whether there is an authority conflict. In order to ensure a reasonable evaluation of different factors, the task overlap degree T r and the authority intersection P set the weight coefficient α r and β p , adjust the coefficient according to the actual situation to optimize the detection process of conflicting paths, and finally combine the path depth H(x), task overlap T r , permission intersection P and sample number n, calculate the anomaly score S(x) of sample x, identify potential conflict paths, and provide accurate conflict detection and optimization suggestions for the collaborative office management system.

[0082] Analyze the scope of each role's permissions. First, determine the scope of each role's permissions, including the types of resources accessed, the types of operations that can be performed, and the timeliness of the permissions. Then analyze whether there is any overlap in role permissions. The overlap judgment is based on the complete overlap and partial overlap of permission items, and is compared based on the resources and operations involved in the role. When duplicate roles have the same and overlapping operation permissions on the same resource, potential conflicts and mutual exclusion are identified.

[0083] See also Figure 2 ,The behavior consistency check module includes:

[0084] State transition comparison submodule: Based on potential conflict paths, the module checks the current state of each task against the target state one by one, compares the task's state transition to see if it complies with the predetermined process rules, determines if there are any state inconsistencies, and generates a state transition consistency result.

[0085] Sequential Constraint Check Submodule: Based on the state transition consistency results, the order of task execution is compared to check whether the tasks are executed in the set order and generate sequential consistency check results;

[0086] Execution order anomaly marking submodule: Based on the sequence consistency check results, it marks the tasks that violate the predetermined execution order, records the specific identifier of each execution order anomaly task, and generates the execution order anomaly result;

[0087] State transition comparison submodule: Based on the potential conflict path, the current state and target state of each task are checked one by one. The state transition verification algorithm is used to first extract the current state and target state of each task from the task path. The current state is determined by the stage of task execution and related data fields, and the target state is obtained from the target state field in the predetermined process rules. Then, the process rule comparison algorithm is used to compare according to the state transition logic in the predetermined rules. The rules include the transition rules from the starting state to the completed state of the task. The state changes of the tasks should be processed in timestamp order. If the current state is inconsistent with the target state, it is recorded as not meeting the predetermined process rules, and the state transition consistency result is generated;

[0088] Sequential Constraint Check Submodule: Based on the state transition consistency results, the order of task execution is compared. Using a sequence comparison algorithm, the task execution order is first sorted based on the task timestamps. Then, using the minimum time difference constraint check method, the time difference threshold is set to 5 minutes to check whether the time interval between tasks meets the sequential execution requirements. If the execution order between tasks violates the predetermined constraints, the sequential consistency check result is recorded and generated.

[0089] Execution order anomaly marking submodule: Based on the results of the sequential consistency check, tasks that violate the predetermined execution order are marked. Using the anomaly marking algorithm, all task identifiers that violate the order are first extracted from the sequential consistency check results. Then, according to the task execution order record, the execution order of each task is marked and compared using the task identifier and timestamp. If there is an anomaly in the task execution order, a corresponding anomaly mark is generated. Finally, the specific identifier of each task with an execution order anomaly is recorded to generate an execution order anomaly result.

[0090] See also Figure 2 , the task scheduling and path optimization module includes:

[0091] Task path selection submodule: Based on the execution order anomaly results, it evaluates the resource requirements and priority of each task, selects feasible task paths, considers the resource allocation of each path, and generates task path selection results;

[0092] Path conflict detection submodule: Based on the task path selection results, it performs conflict detection on the selected path, checks whether there are resource contention and execution order conflicts in the path, records the conflicts found, and generates path conflict detection results;

[0093] Task path optimization submodule: Based on the path conflict detection results, it optimizes the task path, rearranges the task execution order, eliminates resource conflicts in the path, and generates an optimized task path by rearranging the task path;

[0094] Task path selection submodule: Based on the abnormal results of the execution order, the resource requirements and priority of each task are evaluated and a task scheduling algorithm is adopted. First, the resource requirements and priority of each task are obtained. The resource requirements are obtained through the task resource requirement table. The priority is calculated based on the importance and time limit of the task. The priority score ranges from 1 to 10. Then, a greedy algorithm is used to select the task path by selecting the tasks with the highest priority and whose resource requirements do not exceed the current available resources. When selecting a path, the resource allocation of each task is checked. If the resource requirement of the current task exceeds the available resources, the path is skipped to generate the task path selection result.

[0095] Path conflict detection submodule: Based on the task path selection results, the selected paths are subjected to conflict detection. Using the conflict detection algorithm, all paths are first extracted from the task path selection results, and resource occupancy detection is performed on the tasks in each path. The resource allocation table, which includes task ID, resource type, and resource occupancy duration, is used to check the resource occupancy of each task in turn. If the resource occupancy time of two tasks overlaps, it is considered that there is a resource contention conflict. Then, the sequential constraint detection is used to check whether there is a conflict in the execution order between tasks. If the task execution order does not match the predetermined order, it is recorded as an execution order conflict. All detected conflicts will be recorded and a path conflict detection result will be generated.

[0096] Task path optimization submodule: Based on the path conflict detection results, the task path is optimized and the task scheduling optimization algorithm is used. First, the path conflict detection results are analyzed to identify task pairs with resource conflicts and execution order conflicts. Then, the task rescheduling algorithm is used to eliminate resource conflicts by rearranging the task execution order, and tasks with smaller resource occupancy are scheduled first for execution. During the task rescheduling process, the resource requirements of the task will be compared with the currently available resources. If the current resources are not available, the task order will be temporarily adjusted. Finally, the optimized task path is generated by rescheduling the task path.

[0097] See also Figure 2 , the security control module includes:

[0098] Illegal behavior detection submodule: Based on abnormal execution sequence results and potential conflict paths, the module compares the execution permissions of each task with the actual operation to check for unauthorized operations and task bypassing behaviors, analyzes whether each task complies with role permissions, and generates illegal behavior detection results.

[0099] Dynamic alarm triggering submodule: Based on the violation detection results, it marks the identified violation tasks and triggers the alarm mechanism. It also records the detailed information of the violation and generates a violation alarm signal.

[0100] Security intervention execution submodule: Based on violation alarm signals, it intervenes in illegal operations, immediately stops the execution of illegal tasks, rolls back the executed operations, and generates security intervention results;

[0101] Violation checking submodule: Based on the abnormal results of the execution sequence and the potential conflict path, the execution authority of each task is compared with the actual operation, and the permission verification algorithm is adopted. First, the execution information of the task is extracted from the abnormal results of the execution sequence and the potential conflict path, including the task ID, task execution time, task operation type, and role authority. Then, the permission comparison algorithm is used to determine the execution authority range of the task by searching the role authority table, which contains the role ID, task ID and permission identifier. Then, the actual operation of each task is compared with the preset permission. If the actual operation of the task exceeds the role authority range, it is determined to be an unauthorized operation. If the task bypasses the predetermined execution process, it is determined to be a task bypass behavior. The detected violation is recorded and a violation detection result is generated.

[0102] Dynamic Alarm Triggering Submodule: Based on the violation detection results, the identified violation tasks are marked and the alarm mechanism is triggered. The alarm triggering algorithm first extracts all violation tasks from the violation detection results. Then, combining the task identifier and execution information, a threshold judgment mechanism is used. When the task violation level exceeds the set threshold, the system triggers an alarm. Then, through the alarm message push mechanism, detailed information about the violation is pushed to the security monitoring platform, generating a violation alarm signal.

[0103] Security intervention execution submodule: Based on the violation alarm signal, it intervenes in the violation operation and adopts the task termination and rollback algorithm. First, it determines the violation task according to the task identifier in the violation alarm signal, and calls the task stop command through the system control interface to stop the execution of the violation task. Then, it uses the rollback algorithm to extract the task execution steps and related data from the task operation log, and gradually rolls back the executed operations. During the rollback process, the database operations involved will be undone through the transaction rollback mechanism. Finally, the security intervention result is generated, and the detailed information of the intervention operation is recorded, including the time of stopping the task, the rollback steps and the data status after recovery.

[0104] See also Figure 3 A collaborative office management method based on big data is provided. The collaborative office management method based on big data is performed based on the above-mentioned collaborative office management system based on big data, and includes the following steps:

[0105] S1: Based on historical office data and user interaction behaviors, we extract each user's task assignment data, resource usage records, and approval behaviors to form a graph structure containing nodes. Nodes include tasks, resources, and roles, and edges represent task assignments, resource usage, and approval relationships. We then aggregate historical behavior features through a graph neural network, combining the time and resource usage in user interaction behaviors to generate a collaborative behavior network diagram.

[0106] S2: Based on the collaborative behavior network diagram, the isolation forest algorithm is used to compare the temporal overlap between tasks and resources. The task execution time intervals are extracted and compared with the time intervals of the occupied resources to identify time conflicts between tasks. The permissions of the roles are analyzed and the isolation forest algorithm is used to determine the overlap between the permissions of the roles. This identifies areas where the permissions of multiple roles overlap and generates potential conflict paths.

[0107] S3: Based on the potential conflict paths, the execution order of tasks is checked one by one, the state transition information of each task is extracted, the state changes of task execution are compared with the global constraints, and the task execution order is judged to see whether it meets the specified constraints. Tasks with inconsistent states are marked and an execution order anomaly result is generated.

[0108] S4: Based on the execution order anomaly results, dynamically schedule the tasks marked as abnormal. According to the resource requirements, priority, and execution order of each task, a feasible path is selected. Possible conflicts in the task path are detected. If conflicts are found, the task execution order is rearranged according to the feasible path to generate an optimized task path.

[0109] S5: Based on the optimized task path, perform security control on the task, analyze whether there are any unauthorized operations and bypass behaviors during the task execution, and strengthen the task permission verification. At the same time, by comparing the role permissions and task content in the task path, identify and prevent non-compliant operations, and trigger a dynamic alarm mechanism based on the inspection results, and implement corresponding intervention measures to generate security intervention results.

[0110] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A collaborative office management system based on big data, characterized in that: The system comprises: Collaborative behavior modeling module: Based on historical office data and user interaction behaviors, a graph neural network is used to construct a heterogeneous graph structure of collaborative office behaviors. Task allocation, resource utilization, and approval behaviors are used as nodes to form graph edges. By aggregating historical behavior features, a collaborative behavior network diagram is obtained. Conflict path identification module: Based on the collaborative behavior network diagram, the module uses the isolation forest to compare the time overlap of tasks and resources, check the cross-authority of roles, identify task execution conflicts and authority overlap conflicts through comparison between nodes, and generate potential conflict paths; Behavior consistency check module: Based on the potential conflict path, it performs real-time comparison of task state transitions with global execution constraints. By checking whether the execution order of tasks complies with the specified constraints, it marks inconsistent tasks and generates an execution order anomaly result. Task scheduling and path optimization module: Based on the execution order anomaly results, it dynamically schedules tasks, selects qualified paths according to the resource requirements and priorities of the tasks, detects path conflicts, rearranges the task order according to feasible paths, and generates optimized task paths; Security control module: Based on the abnormal results of the execution sequence and potential conflict paths, it checks the unauthorized operations and task bypass behaviors, triggers the dynamic alarm and intervention mechanism, and generates security intervention results by identifying and preventing violations.

2. The collaborative office management system based on big data according to claim 1, characterized in that: The collaborative behavior modeling module includes: Graph structure construction submodule: Based on historical office data and user interaction behavior, by analyzing task allocation, resource usage, and approval behavior, a graph neural network is used to construct task nodes, role nodes, and resource nodes, and connect them to form edges between nodes, generating a collaborative behavior graph structure. Node feature aggregation submodule: Based on the collaborative behavior graph structure, by weighted aggregation of the adjacent node features of each node, the hidden state of each node is updated, the task execution frequency and resource occupancy time of the node are aggregated, and the node hidden state representation is formed; Collaborative behavior network generation submodule: Based on the node hidden state representation, combined with the relationship between each node and the historical behavior characteristics, the entire collaborative network is constructed to generate a collaborative behavior network graph.

3. The collaborative office management system based on big data according to claim 2 is characterized in that: The graph neural network is based on the formula: in: represents the feature vector of node v in the kth layer in the collaborative office management system, σ represents the activation function, α v represents the task priority coefficient of node v, W (k) represents the weight matrix of the kth layer, represents the set of adjacent nodes connected to node v, β vu represents the behavioral influence coefficient between node v and adjacent node u, represents the feature vector of the adjacent node u in the k-1 layer, γ v represents the resource availability adjustment item of node v, b (k) represents the bias term of the kth layer.

4. The collaborative office management system based on big data according to claim 1, characterized in that: The conflict path identification module includes: Time overlap comparison submodule: Based on the collaborative behavior network diagram, the isolation forest is used to compare the execution time of each task with the time interval of the occupied resources, detect the time overlap between tasks, identify the task combinations with time conflicts, and generate a task time conflict set; The authority overlap detection submodule analyzes the authority scope of each role based on the collaborative behavior network diagram, determines whether there is authority overlap between roles, identifies the overlapping parts of multiple role permissions, and generates a role authority conflict set. Potential conflict path generation submodule: Based on the task time conflict set and the role authority conflict set, the task paths are combined, and the potential conflict paths are identified and generated by judging the execution order of the conflicting tasks and the resource competition situation.

5. The collaborative office management system based on big data according to claim 4 is characterized in that: The isolation forest is based on the formula: Where: S(x) represents the anomaly score of sample x, H(x) represents the path depth of sample x being split in the isolation forest, T r represents the time overlap of task r, P represents the cross-authority score of role, α r The weight coefficient of task time overlap, β p The weight coefficient representing the cross-functionality of role permissions, c(n) represents the dataset size constant, and n represents the number of samples.

6. The collaborative office management system based on big data according to claim 4 is characterized in that: The scope of authority of each role is analyzed. First, the scope of authority of each role is determined, including the type of resources accessed, the type of operations that can be performed, and the timeliness of the authority. Then, whether there is overlap in the role authority is analyzed. The overlap judgment is based on the complete overlap and partial overlap of the authority items. The resources and operations involved in the role are compared. When duplicate roles have the same and overlapping operation permissions on the same resource, potential conflicts and mutual exclusion situations are identified.

7. The collaborative office management system based on big data according to claim 1, characterized in that: The behavior consistency checking module includes: State transition comparison submodule: Based on the potential conflict path, by checking the current state and target state of each task one by one, comparing whether the state transition of the task complies with the predetermined process rules, determining whether there is any state inconsistency, and generating a state transition consistency result; Sequence constraint check submodule: Based on the state transition consistency result, compare the order of task execution, check whether the task execution is performed in the set order, and generate a sequence consistency check result; Execution order anomaly marking submodule: based on the sequence consistency check result, marks the tasks that violate the predetermined execution order, records the specific identification of each execution order anomaly task, and generates an execution order anomaly result.

8. The collaborative office management system based on big data according to claim 1, characterized in that: The task scheduling and path optimization module includes: Task path selection submodule: Based on the execution order anomaly results, the module selects feasible task paths by evaluating the resource requirements and priorities of each task, and generates task path selection results by considering the resource allocation of each path; Path conflict detection submodule: Based on the task path selection result, perform conflict detection on the selected path, check whether there is resource contention and execution order conflict in the path, record the conflicts found, and generate path conflict detection results; Task path optimization submodule: Based on the path conflict detection result, the task path is optimized, the task execution order is rearranged, the resource conflicts in the path are eliminated, and the optimized task path is generated by rearranging the task path.

9. The collaborative office management system based on big data according to claim 1, characterized in that: The security control module includes: Illegal behavior detection submodule: Based on the abnormal results of the execution sequence and potential conflict paths, the module compares the execution permissions of each task with the actual operation to check whether there are any unauthorized operations or task bypass behaviors, analyzes whether each task complies with the role permissions, and generates illegal behavior detection results; Dynamic alarm triggering submodule: Based on the violation detection results, it marks the identified violation tasks and triggers the alarm mechanism, while recording the detailed information of the violation and generating a violation alarm signal; Security intervention execution submodule: Based on the violation alarm signal, intervene in the violation operation, immediately stop the execution of the violation task, roll back the executed operation, and generate a security intervention result.

10. A collaborative office management method based on big data, characterized in that: The collaborative office management system based on big data according to any one of claims 1 to 9 comprises the following steps: S1: Based on historical office data and user interaction behaviors, we extract each user's task assignment data, resource usage records, and approval behaviors to form a graph structure containing nodes. Nodes include tasks, resources, and roles, and edges represent task assignments, resource usage, and approval relationships. We then aggregate historical behavior features through a graph neural network, combining the time and resource usage in user interaction behaviors to generate a collaborative behavior network diagram. S2: Based on the collaborative behavior network diagram, the isolation forest algorithm is used to compare the time overlap between tasks and resources. The task execution time intervals are extracted and compared with the time intervals of the occupied resources to identify time conflicts between tasks. At the same time, the authority scope of the roles is analyzed and the isolation forest algorithm is used to determine the overlap of authority between roles. The areas where multiple roles' authority overlap are identified and potential conflict paths are generated. S3: Based on the potential conflict path, check the execution order of each task one by one, extract the state transition information of each task, compare the state change of task execution with the global constraint conditions, determine whether the task execution order meets the specified constraints, mark the tasks with inconsistent states, and generate an execution order abnormality result; S4: Based on the execution order anomaly results, dynamically schedule the tasks marked as abnormal, select a feasible path based on the resource requirements, priority, and execution order of each task, detect possible conflicts in the task path, and if conflicts are found, rearrange the task execution order according to the feasible path to generate an optimized task path; S5: Based on the optimized task path, perform security control on the task, analyze whether there are any unauthorized operations and bypass behaviors during the task execution process, and strengthen the authority verification of the task. At the same time, by comparing the role permissions and task content in the task path, identify and prevent non-compliant operations, and trigger a dynamic alarm mechanism based on the inspection results, and execute corresponding intervention measures to generate security intervention results.

Images