Trans-department collaborative schedule management method and system

Abstract

The invention relates to the technical field of information processing, in particular to a cross-department collaboration schedule management method and system, and the method comprises the steps: obtaining department schedule data based on a cross-department collaboration framework; performing cross validation on the department schedule data through a spatio-temporal characteristic cross validation algorithm; inputting the structured conflict report into a deep reinforcement learning framework for analysis; performing constraint satisfaction analysis on the multi-modal plan set according to a distributed constraint optimization algorithm; determining a dynamic priority instruction packet according to the decision tree model, and converting the dynamic priority instruction packet into a multi-terminal real-time interactive visual interface rendering protocol; and performing multi-terminal synchronous rendering on the target rendering request in the edge computing node based on a visual interface rendering protocol, and converting real-time interactive interface data into a block chain anchored digital footprint. According to the method, cross-department schedule conflicts can be effectively detected and dynamically optimized, and the task execution efficiency and the resource utilization rate are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a cross-departmental collaborative schedule management method and system. Background Technology

[0002] In the process of enterprise informatization, schedule management is a key link to ensure efficient operation. However, existing schedule management methods and systems have many shortcomings in cross-departmental collaboration scenarios. Traditional schedule management is mostly department- or individual-centric, lacking a cross-departmental perspective. Each department uses independent schedule management tools, resulting in fragmented information that is difficult to share. This leads to difficulties in coordinating cross-departmental meeting arrangements, project progress, and other collaborative tasks, often causing time conflicts and wasting significant communication costs. In cross-departmental collaboration, tasks are highly interconnected; a schedule change in one department can affect other departments. However, existing systems lack effective linkage and coordination mechanisms, failing to synchronize schedule change information with relevant collaborating departments in a timely manner. This prevents other departments from adjusting their plans promptly, impacting overall work progress.

[0003] Chinese Patent Publication No. CN117893151A discloses a digital communication and office platform based on IM technology, including a system application integration unit, a platform building unit, a unified deployment unit, a distributed deployment unit, a user access unit, and an instant messaging unit. The system application integration unit is used to effectively centralize information resources from multiple application systems, providing a unified entry point. The platform building unit is used to construct a multi-system application office platform. The unified deployment unit is used for hierarchical authorization management, enabling subsidiaries to manage their respective portals and functions, while simultaneously allowing them to use the complete set of collaborative office functions of the unified deployment platform. Yes; the distributed deployment unit is used to enable independent processing of data across platforms while maintaining data interoperability, and to deploy internal data services using existing server resources of subsidiaries; the user access unit is used to provide login access for company employees, partners, and the public; the instant messaging unit is used to enable users to communicate instantly, initiate group chats at any time, and achieve barrier-free communication across regions and departments, helping to focus on to-do items and subscribe to the company's public account information, ensuring that important matters are not missed. Simultaneously, the message interface connects to basic mobile office applications such as work reports, approvals, voting, and voice conferencing, driving efficient business collaboration through applications. While this solution can integrate and manage multiple systems in a unified manner, preserving the advantages of the original distributed system and solving problems related to troubleshooting, software support, network issues, and security risks in multiple systems, it cannot effectively detect and dynamically optimize cross-departmental schedule conflicts, leading to task execution delays and low resource utilization. Summary of the Invention

[0004] To address this, the present invention provides a cross-departmental collaborative schedule management method and system to overcome the problems in the prior art that cannot effectively detect and dynamically optimize cross-departmental schedule conflicts, leading to task execution delays and low resource utilization.

[0005] To achieve the above objectives, on the one hand, the present invention provides a cross-departmental collaborative schedule management method, comprising the following steps: S1. Build a cross-departmental collaboration framework in the cloud, and based on the cross-departmental collaboration framework, parse the task dependency relationship of the cross-departmental collaboration schedule project nodes to obtain the task dependency relationship of the schedule project nodes, and obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project nodes. S2. Cross-validate the departmental schedule data using a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report containing schedule conflict type codes, associated task metadata, and impact scores. S3. Input the structured conflict report into the deep reinforcement learning framework for analysis to obtain a multimodal contingency plan set including time window elasticity coefficients, resource alternative matrix and priority adjustment weights; S4. Perform constraint satisfaction analysis on the multimodal contingency plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and construct a decision tree model based on the constraint satisfaction analysis results and real-time task status. S5. Based on the decision tree model, determine the dynamic priority instruction package, which includes the execution sequence, resource consumption threshold and schedule conflict rollback strategy, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. S6. Based on the visualization interface rendering protocol, the target rendering request is synchronously rendered on multiple terminals in the edge computing node to obtain real-time interactive interface data including task progress heatmap and resource usage analysis dashboard, and the real-time interactive interface data is transformed into a blockchain-anchored digital footprint.

[0006] Compared with the prior art, the beneficial effects of this application are as follows: 1. While existing technologies support cross-regional communication, they only connect basic office applications via messaging and lack a systematic detection mechanism for time conflicts in cross-departmental schedules. Existing technologies only achieve data interoperability and hierarchical management, but cannot parse cross-departmental task dependencies, let alone perform spatiotemporal feature cross-validation on departmental schedule data. In contrast, this application constructs a cross-departmental collaboration framework, parses task dependencies, and obtains departmental schedule data. By employing a spatiotemporal feature cross-validation algorithm, it detects time conflicts in departmental schedule data and generates a structured conflict report containing schedule conflict type codes, associated task metadata, and impact scores. This overcomes the problems of missed or misjudged conflicts caused by existing technologies that rely solely on manual coordination or simple time comparison. It achieves accurate location and quantitative assessment of cross-departmental schedule conflicts, avoiding task execution delays caused by undetected conflicts.

[0007] 2. While existing technologies connect to basic applications such as work reporting and approval, they only serve as message transmission channels and are not deeply integrated with schedule management. They cannot dynamically generate optimization solutions based on conflict detection results. Existing technologies only build multi-system office platforms and do not design deep reinforcement learning frameworks or distributed constraint optimization algorithms for schedule conflicts. In contrast, this application inputs structured conflict reports into a deep reinforcement learning framework to analyze and obtain a multimodal set of contingency plans, including time window elasticity coefficients, resource alternative matrix, and priority adjustment weights. The contingency plan set is then analyzed for constraint satisfaction using distributed constraint optimization algorithms and conflict resolution models. A decision tree model is constructed in conjunction with real-time task status, ultimately generating a dynamic priority instruction package containing execution sequences, resource occupancy thresholds, and schedule conflict rollback strategies. This solves the problems of existing technologies that statically handle conflicts and cannot dynamically adjust resources and priorities, significantly improving resource utilization and task execution efficiency.

[0008] 3. Existing technologies only provide login access functionality and do not design a real-time interactive visualization interface for schedule management across multiple terminals. Furthermore, the message interface of existing technologies does not integrate task progress and resource analysis data. In contrast, this application converts dynamic priority instruction packages into a visualization interface rendering protocol, performs multi-terminal synchronous rendering of target rendering requests based on edge computing nodes, and generates real-time interactive interface data including task progress heatmaps and resource usage analysis dashboards. It also anchors digital footprints through blockchain to ensure data immutability and traceability, overcoming the problems of delayed information synchronization and limited collaborative interface functionality in existing technologies. This enables real-time perception and collaborative decision-making regarding schedule status and resource usage across departments. Simultaneously, the digital footprint provides data support for subsequent auditing and optimization, further enhancing the reliability of cross-departmental collaboration.

[0009] Furthermore, S1 includes the following steps: S11. Construct a cross-departmental collaboration framework in the cloud based on preset departmental collaboration rules, data interaction interfaces, and multi-level permission management mechanisms; S12. Based on the cross-departmental collaboration framework, identify the logical relationships and execution order of cross-departmental collaboration schedule project nodes to obtain the task dependencies of schedule project nodes; S13. Based on the task dependencies of the schedule project nodes and the data interaction interface, obtain the department schedule data that matches the task dependencies of the schedule project nodes from the corresponding departments in the cross-departmental collaboration framework.

[0010] In this solution, the preset departmental collaboration rules refer to the pre-defined guidelines used to standardize the collaborative work methods, processes, and standards between different departments. The multi-level access control mechanism refers to the system that hierarchically manages the access and operation permissions of personnel and departments at different levels in the system. For example, ordinary employees can only view data of their own department, department heads can view data of their own department and subordinate groups, and senior leaders can view data of the entire company. Cross-departmental collaboration schedule project nodes refer to the stages with clear tasks and timelines in the cross-departmental collaboration schedule. For example, in a new product development project, the market research completion node and the design finalization node are such stages. The schedule project node task dependency relationship refers to the order and correlation of task execution between different schedule project nodes.

[0011] By building a cross-departmental collaboration framework in the cloud based on departmental collaboration rules, data interaction interfaces, and multi-level permission management mechanisms, the logical relationships and execution order of cross-departmental collaborative schedule project nodes can be accurately identified, and task dependencies can be clearly defined. Then, matching departmental schedule data can be retrieved from the corresponding departments based on these relationships, avoiding collaboration delays caused by poor information flow and conflicting permissions, thus improving the efficiency and accuracy of cross-departmental collaboration.

[0012] Furthermore, S2 includes the following steps: S21. Perform spatiotemporal feature analysis on the department's schedule data to obtain a structured spatiotemporal feature dataset containing timestamp sequences and spatial coordinate information; S22. Perform spatiotemporal feature cross-validation on the structured spatiotemporal feature dataset using a spatiotemporal feature cross-validation algorithm. When the spatiotemporal feature cross-validation meets the preset schedule conflict conditions, encode each conflict type in the conflict schedule list to obtain preliminary conflict data containing schedule conflict type codes and associated task metadata. S23. Calculate the impact of each conflict type in the preliminary conflict data to obtain the corresponding impact score, and integrate the schedule conflict type code, related task metadata and impact score to obtain a structured conflict report.

[0013] In this scheme, the timestamp sequence refers to the sequence formed by arranging the occurrence times of each event in the department's schedule data in chronological order, which can reflect the distribution of the schedule in the time dimension. Spatial coordinate information refers to the coordinate data related to the location of the department's schedule, which reflects the spatial location of the schedule. The spatiotemporal feature cross-validation algorithm is an algorithm used to validate the structured spatiotemporal feature dataset containing timestamp sequences and spatial coordinate information to determine whether different schedules conflict in the spatiotemporal dimension. The preset schedule conflict condition refers to the pre-set standard for judging whether schedules conflict. If the spatiotemporal feature cross-validation result meets the condition, it is determined to be a conflict. The schedule conflict type encoding refers to encoding and identifying each conflict type in the conflict schedule list. The associated task metadata refers to the basic information of the task related to the conflict schedule, such as the task name and person in charge.

[0014] By performing spatiotemporal feature analysis on departmental schedule data, a structured spatiotemporal feature dataset containing timestamp sequences and spatial coordinates is obtained. This dataset is then cross-validated using a spatiotemporal feature cross-validation algorithm. Compared to existing technologies that only verify time or spatial conflicts from a single dimension, this application achieves more accurate identification of schedule conflicts caused by spatiotemporal overlap by simultaneously associating timestamps and spatial coordinates for cross-validation. This effectively captures spatiotemporally coupled conflicts. Conflict types that meet preset conditions are coded to obtain preliminary data. Impact scores are then calculated and integrated into a structured conflict report. This avoids the problems of incomplete conflict identification and lack of spatiotemporal correlation analysis in existing technologies. The conflict report simultaneously includes conflict type codes, associated task metadata, and impact scores, more accurately reflecting the nature and impact of the conflict and improving the targeting and effectiveness of departmental schedule conflict handling.

[0015] Furthermore, S3 includes the following steps: S31. Input the structured conflict report into the feature extraction module of the deep reinforcement learning framework for feature extraction and analysis to obtain the conflict feature tensor; S32. Input the conflict feature tensor into the time series prediction network model and backpropagate the standard deviation of the resource supply and demand difference in each time window to obtain the time window elasticity coefficient. S33. Generate a resource alternative solution matrix based on the time window elasticity coefficient and resource availability data set; S34. Input the time window elasticity coefficient and resource alternative matrix into the policy optimization module of the deep reinforcement learning framework for policy optimization, and obtain a multimodal plan set containing the time window elasticity coefficient, resource alternative matrix and priority adjustment weight.

[0016] In this scheme, the deep reinforcement learning framework refers to a framework that combines deep learning and reinforcement learning, enabling agents to interact with the environment and continuously optimize strategies to achieve specific goals. The feature extraction module is a component in the deep reinforcement learning framework used to extract and analyze features from the input structured conflict reports, transforming the reports into conflict feature tensors. The temporal prediction network model is a model used to backpropagate and calculate the standard deviation of the resource supply and demand difference within each time window, thereby obtaining the time window elasticity coefficient. The standard deviation of the resource supply and demand difference within each time window is an indicator that measures the degree of fluctuation of the resource supply and demand difference within each time window. The time window elasticity coefficient is calculated by the temporal prediction network model and reflects the time window's adaptability to changes in resource supply and demand. The resource availability dataset is a dataset containing the availability of various resources under different time windows. The resource alternative matrix is ​​a matrix generated based on the time window elasticity coefficient and the resource availability dataset, displaying different resource alternatives. The policy optimization module is a module in the deep reinforcement learning framework used to optimize policies based on the time window elasticity coefficient and the resource alternative matrix, generating a multimodal plan set.

[0017] By using the feature extraction module and time-series prediction network model of the deep reinforcement learning framework, conflict characteristics are accurately analyzed and time window elasticity coefficients are calculated. Combined with the resource availability dataset, a resource alternative solution matrix is ​​generated. After optimization by the policy optimization module, a multimodal contingency plan set including time window elasticity coefficients, resource alternative solution matrix and priority adjustment weights is obtained. This helps to comprehensively and scientifically address resource conflicts, improve the flexibility and rationality of resource allocation, effectively reduce the negative impact of resource conflicts, and ensure the smooth progress of various tasks.

[0018] Furthermore, S4 includes the following steps: S41. Based on the distributed constraint optimization algorithm, the discrete plan modes in the multimodal plan set are converted into distributed variable nodes, and the hard constraint rules and soft constraint rules between the distributed variable nodes are determined according to the preset constraint propagation mechanism. S42. By analyzing hard constraint rules, soft constraint rules and historical collaboration data through statistical association rule mining methods, a conflict resolution model is constructed. The conflict resolution model includes a constraint weight adjustment mechanism based on a time prediction module. The time prediction module is trained using task completion time series data from historical collaboration data. S43. Construct an initial constraint satisfaction model based on hard constraint rules, soft constraint rules, and conflict resolution model. Perform constraint satisfaction analysis on the multimodal plan set according to the initial constraint satisfaction model to obtain the constraint satisfaction analysis results. Then, transform the plan subset sequence in the constraint satisfaction analysis results into a structured feature vector. S44. Obtain real-time task status data and construct a decision tree model with structured feature vectors as leaf nodes and real-time task status data as branch nodes.

[0019] In this scheme, discrete contingency plan mode refers to contingency plans existing in discrete form within a multimodal contingency plan set. Each contingency plan represents a specific resource allocation and task arrangement method. Distributed variable nodes are nodes formed after converting discrete contingency plan modes, used in the distributed system to represent relevant variables of different contingency plans. Pre-defined constraint propagation mechanism refers to a pre-set rule propagation method used to determine the constraint relationships between distributed variable nodes, ensuring the effective transmission of constraint rules between nodes. Hard constraint rules are constraints that must be strictly satisfied; if not satisfied, the relevant contingency plan is infeasible. Soft constraint rules are constraints with a certain degree of flexibility; under the premise of satisfying hard constraints, they can be appropriately relaxed to meet other objectives. Statistical association rule mining method refers to finding association rules that meet certain support and confidence thresholds by analyzing the frequency and patterns of simultaneous occurrences of items in the dataset. Historical collaboration data refers to a data set that records relevant information about collaboration between various departments during the execution of multimodal contingency plans in the past. Conflict resolution model refers to a model designed to resolve conflicts between various constraint rules during distributed constraint optimization. Constraint weight adjustment mechanism refers to a mechanism used to dynamically adjust the relative importance of different constraint rules in constraint satisfaction problems. Initial constraint satisfaction model refers to a model built based on hard and soft constraint rules, used to analyze whether the multimodal contingency plan set satisfies the constraint conditions. Contingency plan subset sequence refers to the order in which the contingency plan subsets that satisfy the constraint conditions are arranged in the constraint satisfaction analysis results. Structured feature vector refers to a vector with a specific structure and meaning transformed from the contingency plan subset sequence. Real-time task status data refers to the actual running status data of the task at the current moment, reflecting the real-time progress and resource usage of the task.

[0020] By using a distributed constraint optimization algorithm, discrete contingency plans are transformed into distributed variable nodes. Hard and soft constraint rules are determined based on a pre-defined constraint propagation mechanism. An initial constraint satisfaction model is constructed to perform constraint satisfaction analysis on the multimodal contingency plan set. The sequence of contingency plan subsets is transformed into structured feature vectors. Then, a decision tree model is constructed by combining real-time task status data. This helps to quickly select contingency plans that meet the constraints in complex environments and dynamically adjust decisions based on real-time task status, thereby improving the feasibility and adaptability of contingency plans and ensuring efficient and orderly task execution.

[0021] Furthermore, S5 includes the following steps: S51. Traverse all leaf nodes of the decision tree model and use the cluster centers of the structured feature vectors as the execution sequence. S52. Backtrack the branch nodes along the decision path of the decision tree model and convert the real-time task status data into resource consumption thresholds. S53. Insert conflict detection markers at each branch node of the decision tree model, and define a schedule conflict rollback strategy based on the conflict detection markers. S54. Construct a dynamic priority instruction package based on the execution sequence, resource consumption threshold, and schedule conflict rollback strategy; S55. Parse the dynamic priority instruction packet into protocol fields, and analyze the protocol fields through preset standardized mapping rules to generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

[0022] In this scheme, the cluster center of the structured feature vector refers to the center point of all vectors in each cluster after cluster analysis of the structured feature vector, representing the typical features of the cluster. The execution sequence refers to the sequence formed by arranging the cluster centers of the structured feature vector in a certain order after traversing all leaf nodes of the decision tree model. The decision path refers to the path from the root node to the leaf node of the decision tree model, reflecting the process of making decisions based on real-time task status data. The resource usage threshold refers to the resource usage limit value obtained by tracing back the branch nodes along the decision path and converting the real-time task status data, used to judge whether the resource usage is reasonable. The conflict detection marker refers to the marker inserted at each branch node of the decision tree model, used to detect whether a conflict has occurred at that node. The schedule conflict rollback strategy refers to the strategy defined based on the schedule conflict detection marker, which restores the system to the state before the conflict occurred when a conflict is detected. The protocol field refers to the data unit with specific meaning and format obtained after parsing the dynamic priority instruction packet. The preset standardization mapping rule refers to the pre-set rule used to analyze the protocol field and convert it into a format that meets the requirements of the rendering pipeline.

[0023] By using the cluster centers of structured feature vectors as the execution sequence, backtracking branch nodes transforms real-time task status data into resource consumption thresholds, inserting conflict detection markers and defining schedule conflict rollback strategies, constructing dynamic priority instruction packages, parsing them into protocol fields, and analyzing them through preset standardized mapping rules, a multi-terminal real-time interactive visual interface rendering protocol is generated. This protocol can dynamically adjust the rendering process according to the real-time task status, rationally allocate resources, handle conflicts in a timely manner, improve rendering efficiency and flexibility, and ensure the smooth completion of rendering tasks.

[0024] Furthermore, S55 includes the following steps: S551. The parser performs a structured split on the input dynamic priority instruction packet to obtain the protocol fields; S552. The protocol fields are matched and transformed item by item through preset standardized mapping rules to obtain the intermediate data structure; S553. Analyze the intermediate data structure and generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

[0025] In this solution, the visual interface rendering protocol refers to the protocol generated by analyzing the intermediate data structure obtained after converting protocol fields through preset standardized mapping rules.

[0026] The parser performs a structured decomposition of dynamic priority instruction packets to obtain protocol fields. Then, it uses preset standardized mapping rules to match and transform each protocol field to obtain an intermediate data structure. Finally, it generates a rendering protocol for a real-time interactive visual interface for multiple terminals. This protocol can accurately parse instruction packet information, convert data of different formats into a unified and standardized intermediate data structure, and generate a protocol that adapts to the rendering pipeline. This can improve the flexibility and efficiency of rendering, reduce manual intervention, ensure that the rendering results meet expectations, and improve the overall rendering quality and stability.

[0027] Furthermore, S6 includes the following steps: S61. In the edge computing node, analyze the terminal characteristics and network environment based on the visual interface rendering protocol, and preprocess the target rendering request based on the terminal characteristics and network environment to obtain the preprocessed target rendering request. S62. Distribute the preprocessed target rendering request to multiple rendering modules for parallel computing to obtain real-time interactive interface data for different terminals, including task progress heatmaps and resource usage analysis dashboards. S63. Integrate and unify the format of the real-time interactive interface data to obtain a standardized real-time interactive interface data set. S64. Perform a hash operation on the standardized real-time interactive interface data set to obtain the hash operation result, and write the hash operation result into the blockchain according to the smart contract to obtain the on-chain anchored digital footprint certificate.

[0028] In this solution, edge computing nodes refer to computing units deployed at the network edge close to terminal devices. Target rendering requests refer to requests issued by users or the system that require rendering processing, containing the basic information needed for rendering. Preprocessed target rendering requests refer to requests obtained by edge computing nodes after performing preprocessing operations such as format adjustment and parameter adaptation based on terminal characteristics and network environment. Real-time interactive interface data refers to real-time interactive interface-related data generated for different terminals after multiple rendering modules perform parallel computation on the preprocessed target rendering requests, such as interface layout and element display. Standardized real-time interactive interface data set refers to the data set obtained by integrating real-time interactive interface data from different terminals and unifying the format. Smart contracts refer to automatically executed contract terms deployed on the blockchain in the form of code. On-chain anchored digital footprint credentials are formed by hashing the standardized real-time interactive interface data set on the blockchain in the form of code, and then writing the result into the blockchain through smart contracts.

[0029] By analyzing terminal characteristics and network environment based on a visual interface rendering protocol at edge computing nodes, the target rendering request is preprocessed and distributed to multiple rendering modules for parallel computation to obtain real-time interactive interface data. After integration and format unification, a standardized real-time interactive interface data set is formed. Then, through hash operation and using smart contracts, it is written to the blockchain to obtain on-chain anchored digital footprint credentials. This can effectively improve rendering efficiency, reduce latency, ensure that different terminals obtain adapted interactive interfaces, and at the same time guarantee data security and traceability.

[0030] Furthermore, S64 includes the following steps: S641. Perform a one-way hash calculation on the standardized real-time interactive interface data set according to the cryptographic hash algorithm to obtain the hash operation result; S642. A smart contract deployed through the blockchain calls the write function of the smart contract with the hash operation result as an input parameter; S643. Trigger the consensus mechanism of the blockchain according to the write function, and store the hash value in the hash operation result into the blockchain according to the consensus mechanism to obtain the on-chain anchored digital footprint certificate.

[0031] In this scheme, the cryptographic hash algorithm refers to an algorithm that converts a standardized real-time interactive interface data set of arbitrary length into a fixed-length hash value through calculation rules. The write function refers to a function defined in a smart contract deployed on the blockchain, which is used to take the hash calculation result as an input parameter to trigger the operation of writing data to the blockchain. The consensus mechanism of the blockchain refers to the rules and processes for achieving data consistency among nodes in the blockchain network. Through the consensus mechanism, multiple nodes reach a consensus on the writing and storage of data, ensuring that the data on the blockchain is immutable and reliable.

[0032] By performing a one-way hash calculation on a standardized real-time interactive interface data set using a cryptographic hash algorithm, the hash result is obtained. The smart contract deployed on the blockchain uses the hash result as an input parameter to call the write function, triggering the blockchain's consensus mechanism and storing the hash value in the blockchain. This results in an on-chain anchored digital footprint credential, ensuring data integrity and immutability, guaranteeing data security and credibility, and effectively preventing data from being maliciously tampered with or forged.

[0033] On the other hand, the present invention also provides a cross-departmental collaborative schedule management system, comprising: The departmental schedule data acquisition module is used to build a cross-departmental collaboration framework in the cloud, and to parse the task dependency relationship of the cross-departmental collaborative schedule project nodes based on the cross-departmental collaboration framework to obtain the task dependency relationship of the schedule project node, and to obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project node. The structured conflict report generation module is used to cross-validate departmental schedule data through a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report that includes schedule conflict type codes, associated task metadata, and impact scores. The multimodal contingency plan generation module is used to input structured conflict reports into a deep reinforcement learning framework for analysis, and obtain a multimodal contingency plan that includes time window resilience coefficients, resource alternative matrix and priority adjustment weights. The decision tree model building module is used to perform constraint satisfaction analysis on the multimodal plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and build a decision tree model based on the constraint satisfaction analysis results and real-time task status. The rendering protocol conversion module is used to determine the dynamic priority instruction package containing the execution sequence, resource consumption threshold and schedule conflict rollback strategy based on the decision tree model, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. The multi-terminal synchronous rendering module is used to perform multi-terminal synchronous rendering of target rendering requests in edge computing nodes based on a visual interface rendering protocol, to obtain real-time interactive interface data including task progress heatmaps and resource usage analysis dashboards, and to transform the real-time interactive interface data into a blockchain-anchored digital footprint. Attached Figure Description

[0034] Figure 1 This is a flowchart illustrating a cross-departmental collaborative schedule management method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a cross-departmental collaborative schedule management system according to an embodiment of the present invention. Detailed Implementation

[0035] The following detailed description illustrates the specific implementation method: Please see Figure 1 The diagram shown is a flowchart illustrating a cross-departmental collaborative schedule management method according to an embodiment of the present invention, including the following steps: S1. Build a cross-departmental collaboration framework in the cloud, and based on the cross-departmental collaboration framework, parse the task dependency relationship of the cross-departmental collaboration schedule project nodes to obtain the task dependency relationship of the schedule project nodes, and obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project nodes. S2. Cross-validate the departmental schedule data using a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report containing schedule conflict type codes, associated task metadata, and impact scores. S3. Input the structured conflict report into the deep reinforcement learning framework for analysis to obtain a multimodal contingency plan set including time window elasticity coefficients, resource alternative matrix and priority adjustment weights; S4. Perform constraint satisfaction analysis on the multimodal contingency plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and construct a decision tree model based on the constraint satisfaction analysis results and real-time task status. S5. Based on the decision tree model, determine the dynamic priority instruction package, which includes the execution sequence, resource consumption threshold and schedule conflict rollback strategy, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. S6. Based on the visualization interface rendering protocol, the target rendering request is synchronously rendered on multiple terminals in the edge computing node to obtain real-time interactive interface data including task progress heatmap and resource usage analysis dashboard, and the real-time interactive interface data is transformed into a blockchain-anchored digital footprint.

[0036] Specifically, S1 includes the following steps: S11. Construct a cross-departmental collaboration framework in the cloud based on preset departmental collaboration rules, data interaction interfaces, and multi-level permission management mechanisms; S12. Based on the cross-departmental collaboration framework, identify the logical relationships and execution order of cross-departmental collaboration schedule project nodes to obtain the task dependencies of schedule project nodes; S13. Based on the task dependencies of the schedule project nodes and the data interaction interface, obtain the department schedule data that matches the task dependencies of the schedule project nodes from the corresponding departments in the cross-departmental collaboration framework.

[0037] In this embodiment, the multi-level permission management mechanism is meticulously designed based on factors such as departmental roles and job functions. Through role allocation and permission configuration, it ensures the security of data access and operations. In identifying the logical connections and execution order of cross-departmental collaborative schedule project nodes, it is necessary to deeply analyze the specific tasks, time nodes, and dependencies between tasks in each department's collaborative project. This can be achieved through a comprehensive judgment based on project plan documents, historical collaboration records, and expert experience. Using process modeling tools, the identified logical connections and execution order are visualized, thereby obtaining the task dependencies of the schedule project nodes and clarifying the sequence and constraints between tasks. Based on the departments to which the tasks involved in the task dependencies belong, data requests are initiated to the corresponding departmental systems through data interaction interfaces. During the data request process, identity verification and permission checks are strictly performed according to the permission management mechanism to ensure that only requests with the appropriate permissions can be processed. After receiving a legitimate request, each departmental system filters out schedule data matching the task dependencies from its local schedule database based on the request content, such as task start time, end time, and responsible person information, and feeds this data back to the cross-departmental collaboration framework through the data interaction interface.

[0038] Specifically, S2 includes the following steps: S21. Perform spatiotemporal feature analysis on the department's schedule data to obtain a structured spatiotemporal feature dataset containing timestamp sequences and spatial coordinate information; S22. Perform spatiotemporal feature cross-validation on the structured spatiotemporal feature dataset using a spatiotemporal feature cross-validation algorithm. When the spatiotemporal feature cross-validation meets the preset schedule conflict conditions, encode each conflict type in the conflict schedule list to obtain preliminary conflict data containing schedule conflict type codes and associated task metadata. S23. Calculate the impact of each conflict type in the preliminary conflict data to obtain the corresponding impact score, and integrate the schedule conflict type code, related task metadata and impact score to obtain a structured conflict report.

[0039] In this embodiment, spatiotemporal feature analysis of departmental schedule data includes extracting key time features from timestamp sequences, such as the start time, end time, duration, time period (morning / afternoon / evening), and weekday / weekend identifiers. For periodic schedules (such as weekly meetings), their periodic patterns are extracted. Based on spatial coordinate information, the geographical location features of the schedule are extracted, such as the department's office area, meeting room floor, and project site location. For mobile schedules (such as visiting clients), the coordinates of the starting point and destination, as well as the movement path, are recorded.

[0040] Check for time overlap between different tasks at the same spatial coordinates. For example, if the timestamp sequences of two tasks overlap (i.e., the end time of one task is later than the start time of another), it is considered a time conflict. Check for spatial coordinate conflicts between different tasks within the same time period. For example, if the time periods of two tasks overlap and their spatial coordinates are the same (e.g., the same meeting room is booked repeatedly), it is considered a spatial conflict. Combine the time and spatial dimensions to verify compound spatiotemporal conflicts of tasks. For example, an employee may be assigned tasks to different locations within the same time period, or two tasks may overlap in both time and space. Based on preset schedule conflict conditions, run a cross-validation algorithm to detect conflicts. When a conflict is detected, record the detailed information of the conflicting task (such as task name, time, location, participants, etc.). The preset schedule conflict conditions include checking whether the schedule time intervals overlap in the time dimension and determining whether the resource occupation range overlaps in the spatial dimension.

[0041] Detected conflicts are classified and coded. For example, code T1 represents a time conflict (same location, overlapping times); code B1 represents a spatial conflict (same time, conflicting locations); and code TB1 represents a combined spatiotemporal conflict (conflicts in both time and space). The schedule conflict type codes are integrated with associated task metadata (such as task ID, participants, priority, etc.) to form preliminary conflict data.

[0042] The impact calculation for each conflict type in the preliminary conflict data includes calculating the impact of the time dimension based on factors such as the duration of the conflict task, the number of participants, and task priority. For example, time-related conflicts involving high-priority tasks have a higher impact. The spatial dimension impact is calculated based on factors such as the capacity limitations and frequency of use of the conflict location. For example, the smaller the meeting room capacity, the higher the spatial impact. Combining the time and spatial impacts, a comprehensive impact score is calculated using a weighted average or other multi-attribute decision-making methods (such as the analytic hierarchy process). For example, the time impact might have a weight of 60%, and the spatial impact might have a weight of 40%.

[0043] The spatiotemporal feature cross-validation algorithm takes a structured spatiotemporal feature dataset as input: Meeting A (timestamp 9:00-11:00, spatial coordinates: 3rd floor meeting room A, participant A) and Meeting B (timestamp 10:30-12:00, spatial coordinates: 3rd floor meeting room A, participant B). Through cross-validation, the algorithm detects an intersection of timestamp sequences at the same spatial coordinates (10:30-11:00), satisfying the time conflict condition. The spatiotemporal feature cross-validation algorithm outputs preliminary conflict data: conflict type code T1, associated task metadata (task ID: A01 / B02, participant A / B, both high priority).

[0044] The determination of pre-defined schedule conflict conditions is based on departmental resource constraints and task characteristics. Time dimension: Based on the principle that "the same resource cannot be repeatedly occupied in the same time period," rules for determining overlapping time intervals are set (e.g., task A's end time > task B's start time and task A's start time < task B's end time constitutes an overlap). Spatial dimension: Combining resource capacity limitations, it is stipulated that if the same spatial coordinate is occupied by different tasks during overlapping time periods, it constitutes a conflict (e.g., a meeting room can only accommodate one task at a time). Composite conflict conditions combine both temporal and spatial dimensions; if the same person is assigned to tasks at different spatial coordinates during overlapping time periods, it is determined to be a temporal and spatial composite conflict.

[0045] Specifically, S3 includes the following steps: S31. Input the structured conflict report into the feature extraction module of the deep reinforcement learning framework for feature extraction and analysis to obtain the conflict feature tensor; S32. Input the conflict feature tensor into the time series prediction network model and backpropagate the standard deviation of the resource supply and demand difference in each time window to obtain the time window elasticity coefficient. S33. Generate a resource alternative solution matrix based on the time window elasticity coefficient and resource availability data set; S34. Input the time window elasticity coefficient and resource alternative matrix into the policy optimization module of the deep reinforcement learning framework for policy optimization, and obtain a multimodal plan set containing the time window elasticity coefficient, resource alternative matrix and priority adjustment weight.

[0046] In this embodiment, when the structured conflict report is input into the feature extraction module of the deep reinforcement learning framework, the report data is preprocessed first. The text data in the report (such as task description) is converted into a numerical vector using natural language processing techniques, such as the bag-of-words model or the TF-IDF algorithm; the numerical data (such as impact score and timestamp) is directly standardized. The feature extraction module uses a convolutional neural network (CNN) to perform sliding convolution operations on the data using convolution kernels to extract local features, such as the correlation pattern between conflict type and task priority. Then, the feature dimensionality is reduced by pooling layers to retain key information. After multiple layers of convolution and pooling operations, the features at different levels are fused to finally obtain a conflict feature tensor, which stores various feature information of the conflict in the form of a multi-dimensional array. The input of the CNN is the preprocessed structured conflict data vector, which includes conflict type encoding (T1 / S1, etc. converted to numerical values ​​[1,0,0]), impact score (standardized to 0.8), number of participants (standardized to 0.6), task priority (standardized to 0.9), and other features, forming a 1×10 input vector. The structure adopts a two-layer convolutional pooling architecture: the first layer has a kernel size of 3×1, which extracts local correlations (such as the coupling features of conflict type and priority) and outputs 6 feature maps; after dimensionality reduction by a max pooling layer, the second layer has a kernel size of 2×1 to deepen the features and outputs 4 feature maps; finally, the features are fused through a fully connected layer to output an 8×8×4 conflict feature tensor, which stores multi-dimensional correlation information such as conflict type and impact.

[0047] The conflict feature tensor is input into a temporal prediction network model, which uses a Long Short-Term Memory (LSTM) network. LSTM can process temporal data and capture the dynamic changes in resource supply and demand within a time window. During training, the standard deviation of the resource supply and demand difference within each time window is used as part of the loss function. The model predicts the resource supply and demand difference based on the current feature tensor and calculates the standard deviation between the predicted and actual values. Through the backpropagation algorithm, gradients are calculated layer by layer starting from the output layer, and the error information is propagated back to each layer of neurons. The model parameters, such as weights and biases, are adjusted based on the gradient information so that the standard deviation of the predicted resource supply and demand difference continuously approaches the true value, ultimately obtaining the time window elasticity coefficient that reflects the sensitivity of the time window to changes in resource supply and demand. The input to the LSTM is an 8×8×4 conflict feature tensor output by the CNN (flattened to a 256-dimensional temporal vector), containing conflict data from three consecutive time windows. The architecture employs a single-hidden-layer LSTM: the input layer receives 256-dimensional features, and the hidden layer has 32 memory units. A gating mechanism captures the dynamics of resource supply and demand within a time window (e.g., changes in the supply-demand difference in a meeting room from 9:00 to 12:00). The standard deviation of the supply-demand difference is used as the loss function, and the weights are adjusted via backpropagation. The output layer outputs a 1-dimensional time window resilience coefficient (e.g., 0.72) through a fully connected layer, reflecting the sensitivity of resource supply and demand to conflicts during that period, thus achieving dynamic fluctuation quantification.

[0048] The time window resilience coefficient and resource alternative matrix are input into the policy optimization module of the deep reinforcement learning framework, which employs the policy gradient algorithm. The policy gradient algorithm directly optimizes the policy function to maximize the cumulative reward of the agent (i.e., the decision-making system) in the environment. During optimization, the time window resilience coefficient and resource alternative matrix are used as state information input to the policy function, which outputs an action (e.g., choosing a resource alternative) based on the current state. After executing an action, the agent receives a reward based on environmental feedback (e.g., conflict resolution effectiveness, resource utilization efficiency). By continuously sampling the state-action-reward sequence, calculating the policy gradient, and updating the policy function parameters, the agent is encouraged to choose actions that yield higher rewards. Ultimately, a multimodal set of solutions is obtained, including the time window resilience coefficient, the resource alternative matrix, and priority adjustment weights, achieving optimal resource allocation and effective conflict resolution.

[0049] Specifically, S4 includes the following steps: S41. Based on the distributed constraint optimization algorithm, the discrete plan modes in the multimodal plan set are converted into distributed variable nodes, and the hard constraint rules and soft constraint rules between the distributed variable nodes are determined according to the preset constraint propagation mechanism. S42. By analyzing hard constraint rules, soft constraint rules and historical collaboration data through statistical association rule mining methods, a conflict resolution model is constructed. The conflict resolution model includes a constraint weight adjustment mechanism based on the time prediction module. The time prediction module is trained by the task completion time series in the historical collaboration data. S43. Construct an initial constraint satisfaction model based on hard constraint rules, soft constraint rules, and conflict resolution model. Perform constraint satisfaction analysis on the multimodal plan set according to the initial constraint satisfaction model to obtain the constraint satisfaction analysis results. Then, transform the plan subset sequence in the constraint satisfaction analysis results into a structured feature vector. S44. Obtain real-time task status data and construct a decision tree model with structured feature vectors as leaf nodes and real-time task status data as branch nodes.

[0050] In this embodiment, an initial constraint satisfaction model (CSP) is constructed based on hard constraint rules (such as "Plan A and Plan B cannot be used simultaneously") and soft constraint rules (such as "the cost of Plan C does not exceed 10% of the budget"). The constraint weight adjustment mechanism based on the time prediction module first collects historical collaboration data, covering task completion time series of various project types to ensure data diversity. The data is cleaned, outliers and missing values ​​are removed, the time format is standardized, and initial constraint weights are labeled according to task attributes. Next, a suitable model is selected to build the time prediction module, such as ARIMA or LSTM. The module is trained using the processed task completion time series, with training and validation sets divided, and parameter tuning to allow the model to understand time patterns. After training, the model is integrated into the conflict resolution model. When a new schedule conflict occurs, the model calls the time prediction module to predict the normal completion time based on the current conflicting task information. Combined with actual constraints, if a long predicted completion time affects progress, the constraint weight of that task is increased to prioritize its processing; conversely, the weight is decreased to avoid overly restricting other tasks, thereby achieving effective conflict resolution. Constraint propagation algorithms (such as AC-3) are used to infer CSP (Constraint Strategies), eliminating combinations of plans that do not meet hard constraints, and adjusting the satisfaction level of soft constraints through local search (such as the minimum conflict method). The final output is the constraint satisfaction analysis results, which select a subset of feasible plans and transform them into structured feature vectors (such as the binary encoding of the plan activation order). Real-time task status data (such as current system load and environmental sensor readings) is dynamically acquired and used as branch node conditions for the decision tree model (such as "triggering a degradation plan when CPU utilization > 80%)". Using the structured feature vectors (such as the encoding of the plan sequence) as the output of the leaf nodes, branch nodes are recursively divided using information gain or Gini impurity criteria to construct an interpretable decision tree model. For example, if real-time data indicates a sudden increase in network latency, the decision tree will select a subset of plans with low bandwidth consumption as the leaf node output.

[0051] The conflict resolution model's inputs include: historical collaboration data (such as the completion time series of meeting tasks over the past 3 months and the number of participants) and current conflict task metadata (task priority, conflict type code TS1). The model's structure includes a data preprocessing layer (cleaning abnormal time data), an LSTM time prediction module (predicting the normal completion time of tasks), and a weight adjustment logic layer (dynamically adjusting time / space constraint weights based on prediction results). For example, if a high-priority meeting is delayed due to time and space conflicts, the model predicts its reasonable duration and adjusts the time constraint weight from 60% to 75%. The conflict resolution model's output is the adjusted constraint weights (e.g., increasing the time constraint weight of high-priority tasks to 70%).

[0052] The inputs to the initial constraint satisfaction model calculation are: hard constraint rules (e.g., "Meeting room A and B cannot be occupied simultaneously between 9:00-11:00"), soft constraint rules (e.g., "conflict resolution cost ≤ 500 yuan"), and candidate solutions from the multimodal contingency plan set. The structure of the initial constraint satisfaction model calculation is as follows: variables represent the contingency plan selection state, and the constraint layer includes hard constraint checks (AC-3 algorithm to eliminate conflicting solutions) and soft constraint optimization (minimum conflict method to adjust costs). For example, it excludes meeting room booking solutions with overlapping times and retains alternative contingency plans that meet the cost requirements. The outputs of the initial constraint satisfaction model calculation are: a sequence of feasible contingency plan subsets (e.g., [Contingency Plan 2, Contingency Plan 5]) and structured feature vectors (e.g., [1,0,1] binary encoding).

[0053] The inputs to a decision tree model example are: real-time task status data (e.g., "network latency = 200ms", "CPU utilization = 85%") and structured feature vectors (plan encoding). The structure of the decision tree model example is: branch nodes are based on real-time data (e.g., "network latency > 150ms"), and leaf nodes are based on feature vectors. Nodes are partitioned using information gain; for example, when "CPU utilization > 80%", the branch points to the leaf node of the plan with low resource consumption. For instance, if a sudden increase in network latency is detected, the decision tree triggers a low-bandwidth plan as output. The output of the decision tree model example is: a subset of optimal plans (e.g., low-bandwidth consumption plans).

[0054] The pre-defined constraint propagation mechanism is determined based on a distributed constraint optimization algorithm, combined with historical resource conflict patterns. Hard constraints are defined by inviolable resource attributes (e.g., "only one task can occupy the same meeting room at the same time," stemming from historical data showing 70% of space conflicts were due to duplicate bookings); soft constraints refer to the resource elasticity range (e.g., "meeting room usage costs fluctuate by no more than 20%," based on historical cost fluctuation data). Iterative verification is performed using a constraint propagation algorithm (e.g., AC-3): first checking for temporal overlap, then verifying the spatial resource occupancy range, ultimately forming a core constraint chain of "no temporal overlap + no spatial overlap."

[0055] Specifically, S5 includes the following steps: S51. Traverse all leaf nodes of the decision tree model and use the cluster centers of the structured feature vectors as the execution sequence. S52. Backtrack the branch nodes along the decision path of the decision tree model and convert the real-time task status data into resource consumption thresholds. S53. Insert conflict detection markers at each branch node of the decision tree model, and define a schedule conflict rollback strategy based on the conflict detection markers. S54. Construct a dynamic priority instruction package based on the execution sequence, resource consumption threshold, and schedule conflict rollback strategy; S55. Parse the dynamic priority instruction packet into protocol fields, and analyze the protocol fields through preset standardized mapping rules to generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

[0056] Specifically, S55 includes the following steps: S551. The parser performs a structured split on the input dynamic priority instruction packet to obtain the protocol fields; S552. The protocol fields are matched and transformed item by item through preset standardized mapping rules to obtain the intermediate data structure; S553. Analyze the intermediate data structure and generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

[0057] In this embodiment, the intermediate data structure is validated and optimized for legality based on the hardware characteristics (such as GPU architecture and driver limitations) and software requirements (such as frame buffer format and synchronization mechanism) of the rendering pipeline. For example, if the execution sequence contains texture sampling times supported by the pipeline, it is automatically split into multiple sub-sequences; if the resource threshold exceeds the video memory capacity, the texture resolution is downgraded according to priority. The adapted intermediate data structure is encapsulated into an instruction sequence executable by the rendering pipeline. The visualization interface rendering protocol needs to include modules such as state initialization (such as binding shader programs), resource allocation (such as dynamically adjusting video memory blocks), and conflict handling (such as triggering a rollback strategy when a rendering target conflict is detected) to ensure that the visualization interface rendering protocol is compatible with the pipeline driver and can respond to task state changes in real time. The generated visualization interface rendering protocol is executed through the pipeline simulator or the actual GPU to monitor the rendering results (such as frame rate and image correctness) and resource usage (such as video memory and computing unit utilization). If performance abnormalities are detected (such as stuttering caused by frequent rollbacks), the mapping rule weights are dynamically adjusted (such as reducing the sensitivity of conflict detection), and the protocol generation logic is iteratively optimized.

[0058] The determination of pre-defined standardized mapping rules requires dynamic iteration based on hardware characteristics, software requirements, and performance feedback. First, considering GPU architecture and driver limitations, hardware parameters such as texture sampling capabilities and memory capacity are analyzed. For example, the mapping threshold from execution sequence to sampling instructions is determined based on the GPU's maximum sampling count per cycle. Second, based on software requirements such as frame buffer format and synchronization mechanisms, the conversion format between protocol fields and intermediate data structures is defined, such as pixel format conversion rules matching the rendering target. Finally, by monitoring metrics such as frame rate and memory usage during actual rendering, if stuttering occurs due to overly sensitive conflict detection, the sensitivity weight of the corresponding rule is reduced, forming a mapping logic that adapts to the hardware and responds in real time.

[0059] Specifically, S6 includes the following steps: S61. In the edge computing node, analyze the terminal characteristics and network environment based on the visual interface rendering protocol, and preprocess the target rendering request based on the terminal characteristics and network environment to obtain the preprocessed target rendering request. S62. Distribute the preprocessed target rendering request to multiple rendering modules for parallel computing to obtain real-time interactive interface data for different terminals, including task progress heatmaps and resource usage analysis dashboards. S63. Integrate and unify the format of the real-time interactive interface data to obtain a standardized real-time interactive interface data set. S64. Perform a hash operation on the standardized real-time interactive interface data set to obtain the hash operation result, and write the hash operation result into the blockchain according to the smart contract to obtain the on-chain anchored digital footprint certificate.

[0060] Specifically, S64 includes the following steps: S641. Perform a one-way hash calculation on the standardized real-time interactive interface data set according to the cryptographic hash algorithm to obtain the hash operation result; S642. A smart contract deployed through the blockchain calls the write function of the smart contract with the hash operation result as an input parameter; S643. Trigger the consensus mechanism of the blockchain according to the write function, and store the hash value in the hash operation result into the blockchain according to the consensus mechanism to obtain the on-chain anchored digital footprint certificate.

[0061] In this embodiment, at the edge computing node, device fingerprinting technology (such as User-Agent parsing and WebGL rendering capability detection) is used to obtain the terminal's hardware configuration (such as GPU model and screen resolution) and software environment (such as operating system version and browser support features). Simultaneously, network speed testing tools (such as RTT measurement and bandwidth estimation) are used to assess the current network conditions (such as latency, jitter, and packet loss rate). Based on this data, the target rendering request is preprocessed, for example, by adjusting the texture compression format to adapt to low-end devices or reducing the number of model polygons to cope with high-latency networks. The preprocessed rendering request is split into multiple sub-tasks (such as scene layering or view-based clipping regions), and dynamically allocated to multiple rendering modules within the edge node by a task scheduler based on the real-time load of the rendering modules (such as GPU utilization and queue depth). Each module uses the same rendering pipeline configuration (such as shader programs and lighting parameters) for parallel computation, ensuring that the real-time interactive interface data (such as video stream frames and UI element positions) corresponding to different terminals, including task progress heatmaps and resource usage analysis dashboards, are logically consistent. A collision-resistant cryptographic hash algorithm (such as SHA-256) is used to perform a full-content hash calculation on a standardized dataset to generate a unique hash value. The hash value is then passed as a parameter to the write function (such as mintFootprint) of a smart contract deployed on the blockchain via a blockchain node SDK (such as Web3.js). Upon contract execution, a consensus mechanism (such as PoS or PBFT) is triggered. After verification by a majority of nodes, the hash value is written to the blockchain ledger in the form of a transaction, generating an immutable digital footprint credential (such as a transaction hash or NFT Token ID).

[0062] Please see Figure 2 As shown, it is a schematic diagram of the structure of a cross-departmental collaborative schedule management system according to an embodiment of the present invention, including: The departmental schedule data acquisition module is used to build a cross-departmental collaboration framework in the cloud, and to parse the task dependency relationship of the cross-departmental collaborative schedule project nodes based on the cross-departmental collaboration framework to obtain the task dependency relationship of the schedule project node, and to obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project node. The structured conflict report generation module is used to cross-validate departmental schedule data through a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report that includes schedule conflict type codes, associated task metadata, and impact scores. The multimodal contingency plan generation module is used to input structured conflict reports into a deep reinforcement learning framework for analysis, and obtain a multimodal contingency plan that includes time window resilience coefficients, resource alternative matrix and priority adjustment weights. The decision tree model building module is used to perform constraint satisfaction analysis on the multimodal plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and build a decision tree model based on the constraint satisfaction analysis results and real-time task status. The rendering protocol conversion module is used to determine the dynamic priority instruction package containing the execution sequence, resource consumption threshold and schedule conflict rollback strategy based on the decision tree model, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. The multi-terminal synchronous rendering module is used to perform multi-terminal synchronous rendering of target rendering requests in edge computing nodes based on a visual interface rendering protocol, to obtain real-time interactive interface data including task progress heatmaps and resource usage analysis dashboards, and to transform the real-time interactive interface data into a blockchain-anchored digital footprint.

[0063] Based on the above content, we will provide examples and analysis: Suppose that in a certain company, the administrator has set a default priority for the leader's schedule (priority for high-level meetings) and protected against busy / idle schedule queries (visible only to assistants and core department heads); the assistant has created a "Group Strategy Seminar from 10:00 AM to 12:00 PM on Wednesdays" for the leader (belonging to the leader's schedule and only the assistant can modify it); a regular user plans to create a "Department Business Review Meeting from 11:00 AM to 12:00 PM on Wednesdays" and invite the leader to participate (belonging to the regular meeting schedule).

[0064] First, the constructed cross-departmental collaboration framework integrates the leader's strategic seminar schedule (from the leader's schedule data managed by the assistant) and the business review meeting plans of ordinary users (from the meeting schedule data created by ordinary users), and analyzes the task dependency relationship that "business review meetings depend on the leader's available time", breaking down the information barrier between the leader's schedule and the ordinary user's schedule, and ensuring that all related schedules are included in unified analysis.

[0065] Subsequently, through the spatiotemporal feature cross-validation algorithm, it was detected that the two schedules overlapped in time between "11:00 AM and 12:00 PM on Wednesday", and a structured conflict report was generated: the conflict type was coded as "time overlap - key personnel conflict", and the associated task metadata clearly defined the attributes of the strategic seminar (high priority, involving group strategic decision-making) and the business review meeting (requiring leadership decision support). The impact score was determined to be 8 / 10 (if the conflict continues, the business review meeting will be delayed in subsequent execution due to the lack of leadership decision).

[0066] Based on this report, the deep reinforcement learning framework generates a multimodal contingency plan set: the time window flexibility coefficient shows that the business review meeting can be adjusted to Wednesday afternoon 2-3 pm (flexibility coefficient 0.9, high feasibility of adjustment); the resource alternative solution matrix suggests that if the leader cannot make the adjustment, the deputy leader in charge can be substituted (matching degree 92%); the priority adjustment weight is clear that the priority of the strategic seminar (1.2) is higher than that of the business review meeting (0.8), providing a systematic solution for dynamic optimization and avoiding the lag of traditional experience-based adjustments.

[0067] Based on the default high-priority attribute of the leader's schedule set by the administrator and its real-time status (strategic seminars involve urgent projects and cannot be adjusted), the distributed constraint optimization algorithm and conflict resolution model determine "adjusting the business review meeting time" as the optimal solution through constraint satisfaction analysis, and construct a decision tree model: if the leader's schedule has a higher priority and cannot be adjusted, then meetings that depend on its time will be adjusted first; if there is a suitable replacement, it can be used as an alternative.

[0068] The decision tree model generates a dynamic priority instruction package: the execution sequence is "first confirm that the strategic seminar cannot be adjusted, then notify ordinary users to reschedule the business review meeting to Wednesday afternoon from 2 pm to 3 pm"; resource usage thresholds limit the new time to prevent conflicts with the leader's existing afternoon schedule; the schedule conflict rollback strategy stipulates that if ordinary users cannot adjust the schedule, the deputy leader's alternative will be activated. Simultaneously, this dynamic priority instruction package is converted into a visual interface rendering protocol, ensuring that assistants see the complete adjustment logic in the interface managing the leader's schedule, while ordinary users only see the prompt "The leader is free on Wednesday afternoon, adjustment recommended" in the meeting creation interface (consistent with the setting where the leader's availability information is partially visible), thus protecting information security while achieving efficient cross-role collaboration and breaking down information silos.

[0069] Finally, real-time interactive interfaces are rendered synchronously across multiple terminals via edge computing nodes: the schedule creation interface for ordinary users displays a task progress heatmap (conflicting periods in the morning are marked in red, and idle periods in the afternoon are marked in green), while the interfaces for administrators and assistants display resource usage analysis dashboards (the availability status of afternoon time for deputy leaders), and all adjustment records are converted into digital footprints anchored to the blockchain (such as the trajectory of meeting time changes). This achieves real-time response throughout the entire process from conflict detection to dynamic optimization, avoiding task delays and resource waste caused by information asynchrony and lack of basis for adjustments in traditional management, and effectively solving the problems of information silos and delayed response.

[0070] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A cross-departmental collaborative schedule management method, characterized in that: Includes the following steps: S1. Build a cross-departmental collaboration framework in the cloud, and based on the cross-departmental collaboration framework, parse the task dependency relationship of the cross-departmental collaboration schedule project nodes to obtain the task dependency relationship of the schedule project nodes, and obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project nodes. S2. Cross-validate the departmental schedule data using a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report containing schedule conflict type codes, associated task metadata, and impact scores. S3. Input the structured conflict report into the deep reinforcement learning framework for analysis to obtain a multimodal contingency plan set including time window elasticity coefficients, resource alternative matrix and priority adjustment weights; S4. Perform constraint satisfaction analysis on the multimodal contingency plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and construct a decision tree model based on the constraint satisfaction analysis results and real-time task status. S5. Based on the decision tree model, determine the dynamic priority instruction package, which includes the execution sequence, resource consumption threshold and schedule conflict rollback strategy, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. S6. Based on the visualization interface rendering protocol, the target rendering request is synchronously rendered on multiple terminals in the edge computing node to obtain real-time interactive interface data including task progress heatmap and resource usage analysis dashboard, and the real-time interactive interface data is transformed into a blockchain-anchored digital footprint.

2. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S1 includes the following steps: S11. Construct a cross-departmental collaboration framework in the cloud based on preset departmental collaboration rules, data interaction interfaces, and multi-level permission management mechanisms; S12. Based on the cross-departmental collaboration framework, identify the logical relationships and execution order of cross-departmental collaboration schedule project nodes to obtain the task dependencies of schedule project nodes; S13. Based on the task dependencies of the schedule project nodes and the data interaction interface, obtain the department schedule data that matches the task dependencies of the schedule project nodes from the corresponding departments in the cross-departmental collaboration framework.

3. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S2 includes the following steps: S21. Perform spatiotemporal feature analysis on the department's schedule data to obtain a structured spatiotemporal feature dataset containing timestamp sequences and spatial coordinate information; S22. Perform spatiotemporal feature cross-validation on the structured spatiotemporal feature dataset using a spatiotemporal feature cross-validation algorithm. When the spatiotemporal feature cross-validation meets the preset schedule conflict conditions, encode each conflict type in the conflict schedule list to obtain preliminary conflict data containing schedule conflict type codes and associated task metadata. S23. Calculate the impact of each conflict type in the preliminary conflict data to obtain the corresponding impact score, and integrate the schedule conflict type code, related task metadata and impact score to obtain a structured conflict report.

4. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S3 includes the following steps: S31. Input the structured conflict report into the feature extraction module of the deep reinforcement learning framework for feature extraction and analysis to obtain the conflict feature tensor; S32. Input the conflict feature tensor into the time series prediction network model and backpropagate the standard deviation of the resource supply and demand difference in each time window to obtain the time window elasticity coefficient. S33. Generate a resource alternative solution matrix based on the time window elasticity coefficient and resource availability data set; S34. Input the time window elasticity coefficient and resource alternative matrix into the policy optimization module of the deep reinforcement learning framework for policy optimization, and obtain a multimodal plan set containing the time window elasticity coefficient, resource alternative matrix and priority adjustment weight.

5. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S4 includes the following steps: S41. Based on the distributed constraint optimization algorithm, the discrete plan modes in the multimodal plan set are converted into distributed variable nodes, and the hard constraint rules and soft constraint rules between the distributed variable nodes are determined according to the preset constraint propagation mechanism. S42. By analyzing hard constraint rules, soft constraint rules and historical collaboration data through statistical association rule mining methods, a conflict resolution model is constructed. The conflict resolution model includes a constraint weight adjustment mechanism based on a time prediction module. The time prediction module is trained using task completion time series data from historical collaboration data. S43. Construct an initial constraint satisfaction model based on hard constraint rules, soft constraint rules, and conflict resolution model. Perform constraint satisfaction analysis on the multimodal plan set according to the initial constraint satisfaction model to obtain the constraint satisfaction analysis results. Then, transform the plan subset sequence in the constraint satisfaction analysis results into a structured feature vector. S44. Obtain real-time task status data and construct a decision tree model with structured feature vectors as leaf nodes and real-time task status data as branch nodes.

6. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S5 includes the following steps: S51. Traverse all leaf nodes of the decision tree model and use the cluster centers of the structured feature vectors as the execution sequence. S52. Backtrack the branch nodes along the decision path of the decision tree model and convert the real-time task status data into resource consumption thresholds. S53. Insert conflict detection markers at each branch node of the decision tree model, and define a schedule conflict rollback strategy based on the conflict detection markers. S54. Construct a dynamic priority instruction package based on the execution sequence, resource consumption threshold, and schedule conflict rollback strategy; S55. Parse the dynamic priority instruction packet into protocol fields, and analyze the protocol fields through preset standardized mapping rules to generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

7. A cross-departmental collaborative schedule management method according to claim 6, characterized in that: S55 includes the following steps: S551. The parser performs a structured split on the input dynamic priority instruction packet to obtain the protocol fields; S552. The protocol fields are matched and transformed item by item through preset standardized mapping rules to obtain the intermediate data structure; S553. Analyze the intermediate data structure and generate a visualization interface rendering protocol for real-time interaction across multiple terminals.

8. The cross-departmental collaborative schedule management method according to claim 1, characterized in that: S6 includes the following steps: S61. In the edge computing node, analyze the terminal characteristics and network environment based on the visual interface rendering protocol, and preprocess the target rendering request based on the terminal characteristics and network environment to obtain the preprocessed target rendering request. S62. Distribute the preprocessed target rendering request to multiple rendering modules for parallel computing to obtain real-time interactive interface data for different terminals, including task progress heatmaps and resource usage analysis dashboards. S63. Integrate and unify the format of the real-time interactive interface data to obtain a standardized real-time interactive interface data set. S64. Perform a hash operation on the standardized real-time interactive interface data set to obtain the hash operation result, and write the hash operation result into the blockchain according to the smart contract to obtain the on-chain anchored digital footprint certificate.

9. A cross-departmental collaborative schedule management method according to claim 8, characterized in that: S64 includes the following steps: S641. Perform a one-way hash calculation on the standardized real-time interactive interface data set according to the cryptographic hash algorithm to obtain the hash operation result; S642. A smart contract deployed through the blockchain calls the write function of the smart contract with the hash operation result as an input parameter; S643. Trigger the consensus mechanism of the blockchain according to the write function, and store the hash value in the hash operation result into the blockchain according to the consensus mechanism to obtain the on-chain anchored digital footprint certificate.

10. A cross-departmental collaborative schedule management system, characterized in that: include: The departmental schedule data acquisition module is used to build a cross-departmental collaboration framework in the cloud, and to parse the task dependency relationship of the cross-departmental collaborative schedule project nodes based on the cross-departmental collaboration framework to obtain the task dependency relationship of the schedule project node, and to obtain the departmental schedule data from the cross-departmental collaboration framework according to the task dependency relationship of the schedule project node. The structured conflict report generation module is used to cross-validate departmental schedule data through a spatiotemporal feature cross-validation algorithm to obtain a structured conflict report that includes schedule conflict type codes, associated task metadata, and impact scores. The multimodal contingency plan generation module is used to input structured conflict reports into a deep reinforcement learning framework for analysis, and obtain a multimodal contingency plan that includes time window resilience coefficients, resource alternative matrix and priority adjustment weights. The decision tree model building module is used to perform constraint satisfaction analysis on the multimodal plan set based on the distributed constraint optimization algorithm and conflict resolution model, obtain the constraint satisfaction analysis results, and build a decision tree model based on the constraint satisfaction analysis results and real-time task status. The rendering protocol conversion module is used to determine the dynamic priority instruction package containing the execution sequence, resource consumption threshold and schedule conflict rollback strategy based on the decision tree model, and convert the dynamic priority instruction package into a visualization interface rendering protocol for real-time interaction across multiple terminals. The multi-terminal synchronous rendering module is used to perform multi-terminal synchronous rendering of target rendering requests in edge computing nodes based on a visual interface rendering protocol, to obtain real-time interactive interface data including task progress heatmaps and resource usage analysis dashboards, and to transform the real-time interactive interface data into a blockchain-anchored digital footprint.

Images