Virtual team management method based on enterprise information sharing

Abstract

The present application relates to the technical field of enterprise team time management, in particular, the present application relates to a virtual team management method based on enterprise information sharing, the present application sets a time stamp for key task cooperation link and collects multiple types of data on the information sharing platform, a dynamic reminding threshold model is constructed by using the method of combining transfer learning and adversarial generative network, the social network factor of the task is considered, the quantum annealing algorithm is used to optimize the model parameters, in the task reminding stage, the model is used to monitor the task processing time in real time, the reminding frequency is dynamically adjusted based on cognitive computing, the busy state of the members is obtained by using blockchain and edge computing, the reminding time is postponed by using time series prediction and multi-agent negotiation, emotion computing and virtual reality technology are also used to trigger the reminder, so as to avoid interfering with the work of the members, improve the acceptance and attention of the members to the reminder, and significantly improve the cooperation efficiency of the key task of the virtual team, and optimize the virtual team management effect.

Description

Technical Field

[0001] This invention relates to the field of enterprise team time management technology, and more specifically, to a virtual team management method based on enterprise information sharing. Background Technology

[0002] Enterprise team time management is an important technology with broad application prospects in the digital business environment. It can help enterprises rationally plan resources, improve work efficiency, gain an advantage in market competition, promote the efficient operation of various industries, and achieve sustainable development.

[0003] When virtual teams are working on software development projects, the geographically dispersed members in different time zones make it difficult to establish a unified work rhythm and communication standards. This leads to a lack of effective time management and information sharing mechanisms. When front-end developers complete page designs and pass on tasks to back-end developers, the back-end developers often struggle to understand the urgency and time requirements of the tasks due to these communication barriers, resulting in task backlog. Existing reminder mechanisms are often based on simple preset rules and do not fully consider the actual work situation of members, failing to accurately grasp their status. When back-end developers are busy handling high-priority urgent vulnerability fixing tasks, the existing reminder mechanism still sends new task reminders according to the established rules, which can easily cause members to resist the reminders. Furthermore, when the time schedule or urgency of the task itself changes, the existing mechanism cannot adjust the reminders in time, resulting in untimely task reminders, missed optimal processing opportunities, project delays, and low team collaboration efficiency. To solve this technical problem, we provide a virtual team management method based on enterprise information sharing. Summary of the Invention

[0004] The purpose of this invention is to provide a virtual team management method based on enterprise information sharing to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, one of the objectives of this invention is to provide a virtual team management method based on enterprise information sharing, comprising the following steps:

[0006] S1. On the information sharing platform, set timestamps for each collaborative stage of key tasks, record the time of task initiation, transmission and waiting for processing, and collect historical task data, member busy time data and task priority data.

[0007] S2. Analyze the collected historical task data, member busy time data and task priority data using machine learning algorithms, establish a dynamic reminder threshold model for different types of key task collaboration links, train the dynamic reminder threshold model using historical data, and regularly evaluate and adjust the model parameters based on actual application feedback.

[0008] S3. Based on the dynamic reminder threshold model, the time for the recipient to process the task is monitored in real time. When the time for the recipient to process the task reaches the dynamic threshold between the time for the task initiator to transmit the task, a reminder is triggered. If the recipient does not respond in time, as time reaches the upper limit of the normal processing time, the reminder frequency is dynamically adjusted according to the remaining time ratio and the urgency of the task. The busy status of members on the platform is obtained in real time. Combined with the urgency of the task, when a member is busy and the urgency of the task allows, the reminder time is postponed. The reminder is triggered immediately after the member's busy status is lifted.

[0009] S4. In accordance with the reminder strategy, send reminder information containing basic task information, remaining processing time, and urgency level to members through the information sharing platform, collect feedback from members on the reminder information, and provide the feedback information to the model for training and optimization.

[0010] As a further improvement to this technical solution, S2 employs a method that integrates transfer learning and generative adversarial networks to construct a dynamic reminder threshold model, as detailed below:

[0011] For historical task data, member busy period data, and task priority data, feature mapping and normalization techniques are used for preprocessing to transform them into a unified feature space.

[0012] A pre-trained base model in the same task management scenario is selected as the source model. The collected data is used to adjust the model. At the same time, a metric-based transfer learning method is used to adjust the model parameters by calculating the feature distance between the source domain data and the target domain data. The source domain data is the data that has been used to pre-train the base model in the same task management scenario, and the target domain data is the data corresponding to the specific task that the transfer learning model is currently adapting to.

[0013] Then, a generative adversarial network model is introduced, in which the generator is responsible for generating synthetic data with the same distribution as the real data, and the discriminator is used to distinguish between real data and synthetic data, and the real data and synthetic data are trained adversarially.

[0014] By fusing the above transfer learning model and generative adversarial network model, a dynamic reminder threshold model is obtained.

[0015] As a further improvement to this technical solution, the social network factors of the task are considered when establishing the dynamic reminder threshold model in S2, as follows:

[0016] Analyze the communication frequency, cooperation history and information sharing patterns among members, construct the social network of members, use graph theory to abstract members as nodes, abstract the relationships between members as edges, and assign weights to the edges to represent the closeness of the relationships;

[0017] By combining the collaborative aspects of the task with the social networks of its members, the propagation path and diffusion speed of the task in the network are analyzed. Through social network analysis algorithms, key members and key propagation paths are identified.

[0018] The centrality of members in the social network, the delay in task propagation, and information loss are transformed into features and incorporated into the dynamic reminder threshold model. As the relationships between members change and the task progresses, the social network structure is updated in real time, and the parameters of the dynamic reminder threshold model are dynamically adjusted according to the changes in the network structure.

[0019] As a further improvement to this technical solution, in step S2, when periodically evaluating the model and adjusting the model parameters based on actual application feedback, an optimization strategy based on the quantum annealing algorithm is adopted, as follows:

[0020] Define a task delay risk assessment index and construct an energy function that includes the model's performance evaluation index and parameter constraints. Transform the model's parameter adjustment problem into a quantum annealing problem. Simulate the optimization process of model parameters through the interaction of qubits. Use the quantum annealing algorithm to find the global minimum of the energy function, i.e., the optimal combination of model parameters.

[0021] During quantum annealing, the algorithm is guided to converge to the optimal solution by controlling the temperature and interaction strength of the qubits, and the parameters of the energy function are updated in real time based on the feedback of the members to the reminder information and the results of the task processing.

[0022] As a further improvement to this technical solution, when dynamically adjusting the reminder frequency based on the remaining time ratio and the urgency of the task in step S3, an adaptive reminder strategy based on cognitive computing is adopted, as follows:

[0023] Construct a cognitive model for members, which includes members' attention allocation mechanism and information processing speed. Combine the complexity, urgency and remaining time of the task to evaluate the cognitive load of the task on members and the cognitive resources required to compute the task. The cognitive load represents the amount of load that members bear when performing the task, and the cognitive resources represent the processing capacity of members when processing the task.

[0024] Based on the members' cognitive models and the cognitive load of the task, the reminder frequency is dynamically adjusted, and the members' feedback on the reminders is collected to update the members' cognitive models. Then, based on the updated cognitive models, the reminder strategy is dynamically adjusted.

[0025] As a further improvement to this technical solution, the S3 method for obtaining the member's busy status on the platform in real time adopts a distributed monitoring method based on blockchain and edge computing, as follows:

[0026] The operation data of members on the platform is stored on the blockchain. At the same time, each member is assigned a unique digital identity. Edge computing nodes are deployed on the edge devices of the information sharing platform to collect and process the operation data of members in real time. The edge computing nodes are used to analyze and process the data and extract key features.

[0027] Through the blockchain network, various edge computing nodes share data and perform collaborative computing, and use distributed machine learning algorithms to jointly train a busy status monitoring model. The busy status monitoring model is used to monitor the busy status of members in real time and feed the monitoring results back to the reminder system.

[0028] During distributed monitoring, anomaly detection algorithms are used to detect abnormal behavior in member operation data in real time, and an early warning is issued immediately when abnormal behavior is detected.

[0029] As a further improvement to this technical solution, in step S3, when a member is busy and the urgency of the task allows, the reminder time is postponed using a method based on time series prediction and multi-agent negotiation, as follows:

[0030] Deep learning models are used to model and predict members' historical work time series data, predicting members' busy status and free time in the future. Intelligent agents are built for each task and member. The task intelligent agent is responsible for managing the task's progress and reminder strategy, while the member intelligent agent is responsible for maintaining the member's status information.

[0031] When members are busy and the urgency of the task allows, the task agent and the member agent negotiate. During the negotiation, the task agent proposes a time to postpone the reminder based on the remaining time and urgency of the task. The member agent evaluates and provides feedback on the suggestion based on its own predicted busyness and work schedule. Through multiple negotiations, a time to postpone the reminder that is acceptable to both parties is reached.

[0032] During task execution, the work time sequence data of members and the status information of tasks are updated in real time. When the busy status of members or the urgency of tasks change, the agents renegotiate and dynamically adjust the reminder time.

[0033] As a further improvement to this technical solution, the reminder triggered in S3 adopts a reminder strategy based on emotion computing and virtual reality technology, as follows:

[0034] By analyzing members' text interactions, voice messages, and facial expressions on the platform, affective computing technology is used to identify members' emotional states. Based on the members' emotional states and the urgency of the tasks, virtual reality reminder scenarios are created.

[0035] By using virtual reality devices, members are brought into a created virtual reality reminder scenario, where basic task information, remaining processing time, and urgency level are displayed. Emotional feedback from members in the virtual reality reminder scenario is collected, and the virtual reality reminder scenario is optimized based on the feedback.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] In a virtual team management approach based on enterprise information sharing, a dynamic reminder threshold model is constructed by integrating transfer learning and generative adversarial networks. This model leverages pre-trained model knowledge and expands data to improve performance and generalization ability. It considers task social network factors and adjusts them in real time to make reminder thresholds more aligned with actual needs. Quantum annealing is used to optimize model parameters, enhancing performance and stability. An adaptive reminder strategy based on cognitive computing adjusts reminder frequency according to member cognition and task load, avoiding interference and improving effectiveness. Blockchain and edge computing are used for distributed monitoring of busy states, ensuring data security, real-time performance, and anomaly detection. Time series forecasting and multi-agent negotiation are used to postpone reminder times, making reminders more realistic. Finally, affective computing and virtual reality technology are combined to trigger reminders, increasing member acceptance and attention, and improving the collaborative efficiency of key tasks within the virtual team. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the overall workflow of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Please see Figure 1 As shown, this embodiment provides a virtual team management method based on enterprise information sharing, including the following steps:

[0041] S1. On the information sharing platform, timestamps are set for each collaborative step of the key task to record the time of task initiation, transmission and waiting for processing, and historical task data, member busy time data and task priority data are collected.

[0042] S2. Analyze the collected historical task data, member busy time data, and task priority data using machine learning algorithms to establish dynamic reminder threshold models for different types of key task collaboration links. Train the dynamic reminder threshold models using historical data and regularly evaluate and adjust the model parameters based on actual application feedback.

[0043] S2 employs a method that combines transfer learning and generative adversarial networks to construct a dynamic reminder threshold model, as detailed below:

[0044] Historical task data, member busy time period data, and task priority data have different feature representations and data distributions, which will affect the training effect of subsequent models. For historical task data, member busy time period data, and task priority data, feature mapping and normalization techniques are used for preprocessing to transform them into a unified feature space. The preprocessed data has a unified feature space and similar data distribution, which can accelerate the convergence speed of the model and improve the stability and generalization ability of the model.

[0045] In the same task management scenario, the pre-trained base model has already learned some common features and patterns. Select a pre-trained base model in the same task management scenario as the source model, and adjust it using the collected data. At the same time, use a metric-based transfer learning method to adjust the model parameters by calculating the feature distance between the source domain and target domain data. The source domain data is the data that has been used to pre-train the base model in the same task management scenario, while the target domain data is the data corresponding to the specific task that the transfer learning model is currently adapting to.

[0046] In practical applications, the target domain data may suffer from insufficient samples or imbalanced distribution, which can affect the model's generalization ability. By introducing a generative adversarial network (GAN) model, the generator is responsible for generating synthetic data with the same distribution as the real data, while the discriminator is used to distinguish between real and synthetic data. The real and synthetic data are then used for adversarial training. The GAN model can generate high-quality synthetic data, expand the scale and diversity of training data, alleviate the problems of insufficient data and imbalanced distribution, and improve the model's generalization ability.

[0047] By fusing the aforementioned transfer learning model and generative adversarial network model, a dynamic reminder threshold model is obtained. This model fusion combines the advantages of transfer learning and generative adversarial networks, utilizing the knowledge of the pre-trained model while expanding the training data, thereby improving the model's performance and generalization ability.

[0048] When establishing the dynamic reminder threshold model in S2, the social network factor of the task is considered, as follows:

[0049] In virtual teams, social relationships such as communication frequency, collaboration history, and information sharing patterns among members have a significant impact on task execution and progress. By analyzing the communication frequency, collaboration history, and information sharing patterns among members, a social network of members can be constructed. Using graph theory, members are abstracted as nodes, and the relationships between members are abstracted as edges, with weights assigned to the edges to represent the closeness of the relationships. By constructing a social network, we can better understand the interaction patterns among members, providing an intuitive basis for analyzing task propagation and identifying key members.

[0050] The propagation path and speed of a task within a team affect its execution efficiency and completion time. This study analyzes the propagation path and speed of a task within the network, considering the collaborative aspects of the task and the social networks of its members. Using social network analysis algorithms, key members and key propagation paths are identified. A shortest path algorithm is then used to find the shortest propagation path from the initiating node to the receiving node. Let the edge weights be ω. ij This indicates that the task is in member v i to v j The propagation time for a given starting node v s and target node v t The shortest path algorithm finds the shortest path by continuously updating the shortest distance between nodes.

[0051] The diffusion rate is calculated based on the propagation time of the task between different nodes. Let the task start from node v. i Propagate to node v j The time is t ij Then on edge (v) i v j diffusion rate on ) Centrality analysis algorithms are used to identify key members and determine their degree of centrality. Where a ij These are elements of the adjacency matrix, representing node v. i and v j Whether there are edges connecting them, betweenness centrality Where σ st σ is the number of shortest paths from node s to t. st (v i ) is through node v i The number of shortest paths and critical propagation paths can be determined by analyzing edges whose betweenness centrality values ​​exceed a preset threshold. Through social network analysis algorithms, critical paths and key members of task propagation can be accurately identified, providing targeted strategies for task management.

[0052] The centrality of members in a social network, the delay in task propagation, and information loss all affect task processing time and efficiency. These factors are transformed into features and incorporated into a dynamic reminder threshold model. As relationships between members change and the task progresses, the social network structure is updated in real time, and the parameters of the dynamic reminder threshold model are dynamically adjusted based on these changes. Over time, relationships between members may change, requiring real-time updates to the edge weights of the social network. Based on the updated social network, centrality, propagation delay, and information loss are recalculated, and the parameters of the dynamic reminder threshold model are adjusted using a gradient descent optimization algorithm to minimize the loss function. Where T k It is the alert threshold predicted by the model. The actual alert threshold is N, where N is the number of samples and θ is the model parameter. The alert threshold better matches the actual needs of the task, and can remind members to handle the task in a timely and effective manner, thereby improving the efficiency and quality of task completion.

[0053] In S2, when periodically evaluating the model and adjusting its parameters based on feedback from real-world applications, an optimization strategy based on the quantum annealing algorithm is employed, as detailed below:

[0054] During the optimization process of the model, there needs to be a clear evaluation index to measure the performance of the model. At the same time, the constraints of the parameters need to be considered to ensure that the obtained parameter combination is reasonable and effective. The task delay risk assessment index R is defined, and an energy function E(θ) is constructed. This function contains the model performance evaluation index and parameter constraints. The parameter adjustment problem of the model is transformed into a quantum annealing problem. In the quantum annealing process, the temperature and interaction strength of the qubits are controlled to guide the algorithm to converge to the optimal solution.

[0055] The core of the quantum annealing algorithm is to simulate the evolution of a quantum system. Let the Hamiltonian of the quantum system be H = H0 + s(t)H. p Where H0 is the initial Hamiltonian, H p The problem is the Hamiltonian, s(t) is the annealing parameter, and s(0) = 0, s(T) = 1, where T is the annealing time. Initially, the quantum system is in the ground state H0, and as time t increases, it gradually evolves to H0. p The ground state corresponds to the minimum value of the energy function E(θ).

[0056] The temperature T of a quantum bit q and interaction strength J ij The temperature T during annealing will affect the evolution of the system. q Gradually decrease, interaction strength J ijBy gradually adjusting the system, it can converge from a disordered state to an ordered ground state, enabling the global minimum of the energy function to be found more quickly, which is the optimal combination of parameters for the model, thereby improving the model's performance and stability.

[0057] Based on member feedback on reminders and task processing results, the parameters of the energy function are updated in real time to improve task management effectiveness and reduce the risk of task delays.

[0058] S3. Based on the dynamic reminder threshold model, the time for the recipient to process the task is monitored in real time. When the time between the recipient's processing time and the time sent by the task initiator reaches the dynamic threshold, a reminder is triggered. If the recipient does not respond in time, as time reaches the upper limit of the normal processing time, the reminder frequency is dynamically adjusted according to the remaining time ratio and the urgency of the task. The busy status of members on the platform is obtained in real time. Combined with the urgency of the task, when a member is busy and the urgency of the task allows, the reminder time is postponed. The reminder is triggered immediately after the member's busy status is lifted.

[0059] S3 employs an adaptive reminder strategy based on cognitive computing when dynamically adjusting the reminder frequency according to the remaining time and the urgency of the task, as follows:

[0060] Different members have different attention allocation mechanisms and information processing speeds. These individual differences affect their ability to handle tasks and their response to reminders. By constructing a cognitive model of the members and combining the complexity, urgency, and remaining time of the task, we can assess the cognitive load of the task on the members and the cognitive resources required to compute the task.

[0061] Let ω be the weight of attention allocation for member i on different types of task j. ij Define the information processing speed v of member i. i Let C be the number of information units processed per unit time. Let the complexity of task j be Cj. j The urgency level is U. j The remaining time percentage is R. j The cognitive load L of task j on member i ij It can be calculated using the following formula:

[0062] L ij =aC j +bU j +(1-ab)(1-R j ), where a and b are weight coefficients, and member i requires R cognitive resources to process task j. ij =L ij / v i Accurate assessment of the cognitive load and required cognitive resources of the task provides a scientific basis for subsequent adjustments to reminder strategies.

[0063] Dynamically adjusting the reminder frequency based on the member's cognitive model and the cognitive load of the task ensures that reminders neither overly interfere with the member nor fail to promptly remind them to handle the task when necessary. Let the initial reminder frequency be f0, and based on the cognitive load L... ij Adjust reminder frequency f ij Adjustments are made using piecewise functions, i.e. Where L low and L high The threshold for cognitive load is k1 > 1, k2 < 1. Dynamically adjusting the reminder frequency can flexibly provide reminder services based on the cognitive state and task requirements of members, thereby improving the effectiveness and relevance of reminders and increasing the efficiency of task processing.

[0064] Member feedback on reminders reflects the effectiveness of the current reminder strategy and changes in members' cognitive states. By collecting member satisfaction feedback, attention allocation weights are updated based on satisfaction levels. Information processing speed is then updated based on new task completion times to update members' cognitive models. Finally, L is readjusted based on the updated cognitive models. low L high As time goes on, the reminder strategies k1 and k2 will become more and more in line with the cognitive characteristics and task needs of members, further improving the efficiency and quality of task management.

[0065] S3 employs a distributed monitoring method based on blockchain and edge computing to obtain real-time member activity status on the platform, as detailed below:

[0066] Members' operational data on the platform is stored on the blockchain. At the same time, each member is assigned a unique digital identity. Edge computing nodes are deployed on the edge devices of the information sharing platform to collect and process members' operational data in real time. The edge computing nodes are used to analyze and process the data, extracting the time interval of the operation as a key feature. The blockchain ensures the security and traceability of the data, and the digital identity facilitates member management and data ownership confirmation. Edge computing reduces data transmission latency and cloud computing pressure, and improves the real-time performance and efficiency of data processing.

[0067] The data collected by each edge computing node is local. By sharing data and conducting collaborative computing through a blockchain network, the data from each node can be integrated to obtain more comprehensive member operation information. Through the blockchain network, the edge computing nodes share data and conduct collaborative computing, and use distributed machine learning algorithms to jointly train a busy status monitoring model. The busy status monitoring model is used to monitor the busy status of members in real time and feed the monitoring results back to the reminder system.

[0068] Edge computing node k will extract feature data f kPackaged into a transaction T k The data is then broadcast to the blockchain network, where other nodes obtain the data by verifying the legality of the transaction. Let the local model parameters on edge computing node k be θ. k The global model parameter is Θ, and each node updates its local model based on local data. Where α is the learning rate. It is the local loss function L(θ) k f k The gradient of the model is calculated, and then each node uploads the updated local model parameters to the blockchain network, where the global model parameters are updated using a weighted average. Where m is the number of nodes, ω k The weight of node k is the busy state monitoring model MΘ. The trained busy state monitoring model predicts the busy state of the member S = M(Θ, f) based on the real-time input feature data f, and feeds the result back to the reminder system to provide a reliable basis for the reminder system and improve the pertinence and effectiveness of the reminder.

[0069] During distributed monitoring, anomaly detection algorithms are used to detect abnormal behaviors in member operation data in real time. When abnormal behavior is detected, an early warning is issued immediately, reducing task delays and losses caused by abnormal behavior and improving the reliability of virtual team management.

[0070] In S3, when members are busy and the urgency of the task allows, the reminder time is postponed using a method based on time series prediction and multi-agent negotiation, as follows:

[0071] The work of team members exhibits certain regularity and cyclicality. Deep learning models are used to model and predict the historical work time series data of team members, forecasting their busy and idle periods in the future. This provides a basis for adjusting subsequent reminder times. Let the historical work time series data of team members be X = {x1, x2, ..., x...}. n The model uses a Long Short-Term Memory (LSTM) network. By training the LSM network model, the working state Y is predicted T time steps in the future, expressed as Y = {y}. n+1 y n+2 , ..., y n+T Task agent A ta Maintenance task progress P and reminder strategy S re Member agent A me Maintain member status information S me This includes predicting busy states and work schedules, providing a reliable basis for adjusting reminder times, while the modular design of the agents improves the system's scalability and flexibility.

[0072] When members are busy and the urgency of the task allows, the task agent and member agents negotiate. During the negotiation, the task agent proposes a time to postpone the reminder based on the remaining time and urgency of the task. The member agents evaluate and provide feedback on the proposal based on their predicted busyness and work schedule. Through multiple rounds of negotiation, a time to postpone the reminder that is acceptable to both parties is reached.

[0073] Let r be the remaining time percentage of the task, e be the urgency level, and t be the suggested postponement reminder time proposed by the task agent. su It can be done through the function t su = calculate f(r, e), t su = k1(1-r)+k2e, where k1 and k2 are weight coefficients, and the member agents determine the busy state S based on their own predictions. bu and work arrangements S sc Calculate the probability of accepting the suggestion, p = g(S). bu S sc , t su If p is greater than a certain threshold θ, the suggestion is accepted; otherwise, a new postponement reminder time t is proposed. new Feedback was provided, and through multiple iterative negotiations, until both parties reached an agreement on postponing the reminder time t. final This avoids member interference and task delays caused by unreasonable reminder times, making reminders more relevant to actual work situations and improving the collaboration efficiency of virtual teams.

[0074] During task execution, the work time sequence data of members and the status information of tasks are updated in real time. When the busy status of members or the urgency of tasks change, the agents renegotiate and dynamically adjust the reminder time.

[0075] Real-time acquisition of new work data from members, adding it to the historical time series data, resulting in an updated time series of X, = {x1, x2, ..., x...} n x new The system retrains the Long Short-Term Memory (LSTM) network model for prediction and updates the task's progress and urgency. When a member's busy status or the task's urgency changes beyond a certain threshold, the task agent and member agents renegotiate. Based on the new remaining time ratio, urgency, predicted busy status, and work schedule, the above negotiation process is repeated to obtain a new postponement reminder time. This ensures that the reminder time always matches the actual situation during task execution, reducing task management errors caused by status changes and improving the work efficiency and task completion quality of the virtual team.

[0076] In S3, when members are busy and the urgency of the task allows, the reminder time is postponed using a method based on time series prediction and multi-agent negotiation, as follows:

[0077] A member's emotional state affects their acceptance of reminders and their efficiency in handling tasks. By analyzing members' text interaction information, voice information, and facial expression information on the platform, affective computing technology is used to identify members' emotional states. Based on the members' emotional states and the urgency of the tasks, virtual reality reminder scenarios are created.

[0078] For text-based interactive information, a pre-trained sentiment analysis model is used to output the probability distribution of sentiment categories, representing the probabilities of positive, negative, and neutral sentiments, respectively. For speech information, language features are extracted, and a support vector machine is used for sentiment classification to obtain the sentiment category. For facial expression information, a convolutional neural network is used for sentiment recognition, and the output of the convolutional neural network is the probability distribution of the sentiment category.

[0079] By comprehensively considering the analysis results of text, voice, and facial expressions, the final emotional state is obtained through a weighted average. Based on the emotional state and urgency, virtual reality reminder scenarios are selected from a preset scenario library. This allows for a more accurate grasp of the member's emotional state, providing more suitable reminder scenarios, improving the member's acceptance and attention to reminders, and further enhancing the effectiveness of reminders and the member's work efficiency.

[0080] S4. In accordance with the reminder strategy, send reminder information containing basic task information, remaining processing time, and urgency level to members through the information sharing platform, collect feedback from members on the reminder information, and provide the feedback information to the model for training and optimization.

[0081] The aforementioned virtual team management method based on enterprise information sharing sets timestamps for key task collaboration stages and collects multiple types of data on an information sharing platform. It constructs a dynamic reminder threshold model using a combination of transfer learning and generative adversarial networks, while also considering the social network factors of the task. The model parameters are optimized using a quantum annealing algorithm. During the task reminder stage, the model monitors task processing time in real time, dynamically adjusts the reminder frequency based on cognitive computing, utilizes blockchain and edge computing to obtain members' busy status, postpones reminder times through time series prediction and multi-agent negotiation, and uses affective computing and virtual reality technology to trigger reminders, avoiding interference with members' work and improving members' acceptance and attention to reminders. This significantly improves the collaboration efficiency of key tasks in the virtual team and optimizes the management effect of the virtual team.

[0082] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A virtual team management method based on enterprise information sharing, characterized by: Includes the following steps: S1. On the information sharing platform, set timestamps for each collaborative stage of key tasks, record the time of task initiation, transmission and waiting for processing, and collect historical task data, member busy time data and task priority data. S2. Analyze the collected historical task data, member busy time data and task priority data using machine learning algorithms, establish a dynamic reminder threshold model for different types of key task collaboration links, train the dynamic reminder threshold model using historical data, and regularly evaluate and adjust the model parameters based on actual application feedback. S3. Based on the dynamic reminder threshold model, the time for the recipient to process the task is monitored in real time. When the time for the recipient to process the task reaches the dynamic threshold between the time for the task initiator to transmit the task, a reminder is triggered. If the recipient does not respond in time, as time reaches the upper limit of the normal processing time, the reminder frequency is dynamically adjusted according to the remaining time ratio and the urgency of the task. The busy status of members on the platform is obtained in real time. Combined with the urgency of the task, when a member is busy and the urgency of the task allows, the reminder time is postponed. The reminder is triggered immediately after the member's busy status is lifted. S4. In accordance with the reminder strategy, send reminder information containing basic task information, remaining processing time and urgency level to members through the information sharing platform, collect feedback from members on the reminder information, and provide the feedback information to the model for training and optimization. When establishing the dynamic reminder threshold model in S2, the social network factors of the task are considered, as follows: Analyze the communication frequency, cooperation history and information sharing patterns among members, construct the social network of members, use graph theory to abstract members as nodes, abstract the relationships between members as edges, and assign weights to the edges to represent the closeness of the relationships; By combining the collaborative aspects of the task with the social networks of its members, the propagation path and diffusion speed of the task in the network are analyzed. Through social network analysis algorithms, key members and key propagation paths are identified. The centrality of members in the social network, the delay in task propagation, and information loss are transformed into features and incorporated into the dynamic reminder threshold model. As the relationships between members change and the task progresses, the social network structure is updated in real time, and the parameters of the dynamic reminder threshold model are dynamically adjusted according to the changes in the network structure. In S2, when periodically evaluating the model and adjusting its parameters based on feedback from actual applications, an optimization strategy based on the quantum annealing algorithm is adopted, as follows: Define a task delay risk assessment index and construct an energy function that includes the model's performance evaluation index and parameter constraints. Transform the model's parameter adjustment problem into a quantum annealing problem. Simulate the optimization process of model parameters through the interaction of qubits. Use the quantum annealing algorithm to find the global minimum of the energy function, i.e., the optimal combination of model parameters. During quantum annealing, the algorithm is guided to converge to the optimal solution by controlling the temperature and interaction strength of the qubits, and the parameters of the energy function are updated in real time based on the feedback of the members to the reminder information and the results of the task processing.

2. The virtual team management method based on enterprise information sharing according to claim 1, characterized in that: In S2, a dynamic reminder threshold model is constructed using a method that combines transfer learning and generative adversarial networks, as detailed below: For historical task data, member busy period data, and task priority data, feature mapping and normalization techniques are used for preprocessing to transform them into a unified feature space. A pre-trained base model in the same task management scenario is selected as the source model. The collected data is used to adjust the model. At the same time, a metric-based transfer learning method is used to adjust the model parameters by calculating the feature distance between the source domain data and the target domain data. The source domain data is the data that has been used to pre-train the base model in the same task management scenario, and the target domain data is the data corresponding to the specific task that the transfer learning model is currently adapting to. Then, a generative adversarial network model is introduced, in which the generator is responsible for generating synthetic data with the same distribution as the real data, and the discriminator is used to distinguish between real data and synthetic data, and the real data and synthetic data are trained adversarially. By fusing the above transfer learning model and generative adversarial network model, a dynamic reminder threshold model is obtained.

3. The virtual team management method based on enterprise information sharing according to claim 1, characterized in that: In step S3, when dynamically adjusting the reminder frequency based on the remaining time and the urgency of the task, an adaptive reminder strategy based on cognitive computing is adopted, as follows: Construct a cognitive model for members, which includes members' attention allocation mechanism and information processing speed. Combine the complexity, urgency and remaining time of the task to evaluate the cognitive load of the task on members and the cognitive resources required to compute the task. The cognitive load represents the amount of load that members bear when performing the task, and the cognitive resources represent the processing capacity of members when processing the task. Based on the members' cognitive models and the cognitive load of the task, the reminder frequency is dynamically adjusted, and the members' feedback on the reminders is collected to update the members' cognitive models. Then, based on the updated cognitive models, the reminder strategy is dynamically adjusted.

4. The virtual team management method based on enterprise information sharing according to claim 3, characterized in that: The S3 method for obtaining members' busy status on the platform in real time adopts a distributed monitoring method based on blockchain and edge computing, as detailed below: The operation data of members on the platform is stored on the blockchain. At the same time, each member is assigned a unique digital identity. Edge computing nodes are deployed on the edge devices of the information sharing platform to collect and process the operation data of members in real time. The edge computing nodes are used to analyze and process the data and extract key features. Through the blockchain network, various edge computing nodes share data and perform collaborative computing, and use distributed machine learning algorithms to jointly train a busy status monitoring model. The busy status monitoring model is used to monitor the busy status of members in real time and feed the monitoring results back to the reminder system. During distributed monitoring, anomaly detection algorithms are used to detect abnormal behavior in member operation data in real time, and an early warning is issued immediately when abnormal behavior is detected.

5. The virtual team management method based on enterprise information sharing according to claim 4, characterized in that: In S3, when a member is busy and the urgency of the task allows, the reminder time is postponed using a method based on time series prediction and multi-agent negotiation, as follows: Deep learning models are used to model and predict members' historical work time series data, predicting members' busy status and free time in the future. Intelligent agents are built for each task and member. The task intelligent agent is responsible for managing the task's progress and reminder strategy, while the member intelligent agent is responsible for maintaining the member's status information. When members are busy and the urgency of the task allows, the task agent and the member agent negotiate. During the negotiation, the task agent proposes a time to postpone the reminder based on the remaining time and urgency of the task. The member agent evaluates and provides feedback on the suggestion based on its own predicted busyness and work schedule. Through multiple negotiations, a time to postpone the reminder that is acceptable to both parties is reached. During task execution, the work time sequence data of members and the status information of tasks are updated in real time. When the busy status of members or the urgency of tasks change, the agents renegotiate and dynamically adjust the reminder time.

6. The virtual team management method based on enterprise information sharing according to claim 5, characterized in that: The reminder in S3 is triggered using a reminder strategy based on emotion computing and virtual reality technology, as detailed below: By analyzing members' text interactions, voice messages, and facial expressions on the platform, affective computing technology is used to identify members' emotional states. Based on the members' emotional states and the urgency of the tasks, virtual reality reminder scenarios are created. By using virtual reality devices, members are brought into a created virtual reality reminder scenario, where basic task information, remaining processing time, and urgency level are displayed. Emotional feedback from members in the virtual reality reminder scenario is collected, and the virtual reality reminder scenario is optimized based on the feedback.

Images