Enterprise task allocation method and device

Abstract

The invention discloses an enterprise task allocation method and device. The method comprises the following steps: acquiring a plurality of pieces of data related to enterprise tasks from an enterprise management system to obtain a data set; generating a task relation graph according to the data set; iteratively updating the feature vectors corresponding to the nodes to obtain a first feature vector set output by the graph neural network; determining a first allocation scheme for the enterprise task according to the first feature vector set; updating the task relation graph to obtain a target task relation graph, and iteratively updating feature vectors corresponding to target nodes in the target task relation graph to obtain a second feature vector set output by the graph neural network; and determining a second allocation scheme for the enterprise task according to the second feature vector set. The technical problem that the enterprise task management efficiency and accuracy are low due to the fact that the enterprise task management technology mainly depends on a static rule and a preset algorithm to carry out task allocation and cannot flexibly cope with enterprise requirements changing in real time is solved.

Description

Technical Field

[0001] The present application relates to the field of machine learning, and more specifically, to a method and device for allocating enterprise tasks. Background Art

[0002] As an important part of the transformation and upgrading of the manufacturing industry, intelligent manufacturing capability maturity diagnosis has received extensive attention and application in recent years. Intelligent manufacturing capability maturity diagnosis currently lacks effective information-based auxiliary tools, and the corresponding professional service companies still use traditional methods to implement diagnosis and arrange tasks, lacking digital management of the overall diagnosis process. By building information-based tools for intelligent manufacturing capability maturity diagnosis, various professional companies can achieve the goals of customer satisfaction, service leadership, technological innovation capabilities and scale revenue generation. In modern enterprise management, task allocation and scheduling are key factors in improving efficiency and the ability to respond to market changes. Traditional task management systems usually rely on manual decision-making to assign tasks, which is not only time-consuming and labor-intensive, but may also affect the accuracy and efficiency of task allocation due to human errors. Especially in an enterprise environment involving a large number of employees and complex project management, how to effectively assign appropriate tasks to appropriate execution objects, while considering the urgency of the task, the availability of the execution object and the professional capabilities, is an important challenge to improve the efficiency of enterprise operations.

[0003] Related task management technologies mainly rely on static rules and preset algorithms to handle task allocation problems, but these methods are often unable to flexibly respond to emergencies or real-time changing needs. For example, when the priority of a task suddenly increases, or the person who was originally scheduled to execute the task has something to do, the relevant system often needs to be re-adjusted manually, which is not only inefficient but also prone to errors.

[0004] In addition, related technologies have limitations when dealing with complex data relationships. For example, in an enterprise, there may be dependencies between tasks, and the capabilities of the execution objects and the needs of the enterprise need to be accurately matched. If these complex data relationships are not handled effectively, it may lead to unreasonable task allocation, affecting the overall execution efficiency and quality of results.

[0005] In summary, how to provide a method for intelligent pre-allocation of enterprise diagnostic tasks is an urgent problem that those skilled in the art need to solve. Summary of the invention

[0006] The embodiments of the present application provide a method and device for allocating enterprise tasks, so as to at least solve the technical problem of low efficiency and accuracy of enterprise task management caused by the fact that enterprise task management technology mainly relies on static rules and preset algorithms to allocate tasks and cannot flexibly respond to real-time changing enterprise needs.

[0007] According to one aspect of an embodiment of the present application, a method for allocating enterprise tasks is provided, including: acquiring multiple data related to enterprise tasks from an enterprise management system to obtain a data set; generating a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent enterprise tasks and execution objects of enterprise tasks, and the edges are used to represent correlation indicators between execution objects and enterprise tasks, priority differences between enterprise tasks, and correlation indicators between enterprise tasks; using a graph neural network, iteratively updating feature vectors corresponding to nodes to obtain a first feature vector set output by the graph neural network; determining a first allocation plan for enterprise tasks based on the first feature vector set; when any data in the data set changes, updating the task relationship graph to obtain a target task relationship graph, using the graph neural network to iteratively update feature vectors corresponding to target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; determining a second allocation plan for enterprise tasks based on the second feature vector set.

[0008] Optionally, multiple data related to enterprise tasks are obtained from the enterprise management system to obtain a data set, including: determining a data collection protocol based on the data types of different data related to the enterprise tasks and the verification rules corresponding to the different data types; extracting enterprise demand information in the enterprise management system based on the data collection protocol, wherein the enterprise demand information includes at least one of the following: demand description information, start time, end time, business field and demand priority information; and extracting enterprise scale information related to the enterprise demand information, wherein the enterprise scale information includes at least one of the following: the number of employees of the enterprise, the annual revenue of the enterprise and the market share of the enterprise; and extracting execution object information of the execution object, wherein the execution object is used to process the enterprise task, and the execution object information includes at least one of the following: educational background, number of historical processing cases and execution field; determining the agreed time based on the start time and the end time; formatting and standardizing the enterprise demand information, enterprise scale information, execution object information and agreed time to obtain a data set.

[0009] Optionally, a task relationship graph is generated based on the data set, including: extracting task information corresponding to each enterprise task from the data set, and encoding the task information into a first feature vector, wherein the task information includes at least one of the following: enterprise requirements, task time, and task priority; extracting execution object information corresponding to each execution object from the data set, and encoding the execution object information into a second feature vector; determining a first edge weight corresponding to a first edge based on the execution field and enterprise requirements, wherein the first edge is an edge formed by the execution object and the enterprise task, and the first edge weight includes: a correlation index between the execution object and the enterprise task; determining a second edge weight corresponding to a second edge based on a priority difference evaluation function and a correlation index evaluation function, wherein the second edge is an edge formed between enterprise tasks, and the second edge weight includes: a priority difference and a correlation index between enterprise tasks; generating a task relationship graph based on the first feature vector, the second feature vector, the first edge weight, and the second edge weight.

[0010] Optionally, when any data in the data set changes, the task relationship diagram is updated, including: determining whether the data set has changed according to a preset time interval, and when any data in the data set changes, updating the task relationship diagram by the following method: when the task information changes, encoding the changed task information into a third feature vector, and determining the first target feature vector corresponding to the task information based on the first feature vector and the third feature vector corresponding to the task information before the change; when the execution object information changes, encoding the changed execution object information into a fourth feature vector, and determining the second target feature vector corresponding to the execution object information based on the second feature vector and the fourth feature vector corresponding to the execution object information before the change; when the execution field and / or enterprise needs change, updating the first edge weight corresponding to the first edge; updating the task relationship diagram based on the first target feature vector, the second target feature vector, the updated first edge weight, and the second edge weight.

[0011] Optionally, a graph neural network is used to iteratively update the feature vectors corresponding to the target nodes in the target task relationship graph to obtain a second set of feature vectors output by the graph neural network, including: step S1, configuring the initialization parameters of the graph neural network, wherein the initialization parameters include: an adaptive learning rate and network hierarchy parameters; step S2, the configured graph neural network iteratively updates the feature vectors corresponding to the nodes according to the feature vector corresponding to the target node in the task relationship graph at time t, the adaptive learning rate at time t, the set of neighbor nodes of the target node, the edge weight of the edge where the target node is located, and the feature vector corresponding to the target node at time t+1, wherein t is a positive number and the target node is any node in the task relationship graph; step S3, after each iteration, adjusting the adaptive learning rate according to a preset learning rate adjustment factor and the network training error rate of each iteration; step S4, repeating steps S2 and S3 until the maximum number of iterations is reached or the adaptive learning rate reaches a preset convergence condition to obtain the first set of feature vectors output by the graph neural network.

[0012] Optionally, the graph neural network includes: multiple network layers; configuring the initialization parameters of the graph neural network, including: determining the adaptive learning rates of different network layers based on the baseline learning rate and each network layer in the graph neural network; determining the initial network weights of the graph neural network based on a preset weight distribution pattern and parameters of the preset weight distribution pattern, wherein the preset weight distribution pattern includes at least: normal distribution or uniform distribution, and the parameters of the preset weight distribution pattern include: mean and standard deviation; determining the target activation function of each network layer based on a preset activation function and the adaptive adjustment parameters of each network layer, wherein the adaptive adjustment parameters are correlated with the statistical characteristics of the data in the data set.

[0013] Optionally, based on the first feature vector set, a first allocation plan for enterprise tasks is determined, including: according to a preset feature vector conversion function, standardizing the first feature vector set to obtain a target vector set, wherein the target vector set includes: an execution object vector and an enterprise task vector; determining a scoring function based on a similarity coefficient between the execution object vector and the enterprise task vector; according to the scoring function, determining the execution object corresponding to each enterprise task to obtain a first initial allocation plan; processing the first initial allocation plan according to preset constraints to obtain a first allocation plan, wherein the preset constraints include at least: there is no conflict in the time when the same execution object processes different enterprise tasks.

[0014] Optionally, after determining the second allocation plan for enterprise tasks based on the second set of feature vectors, the method also includes: using a preset scheduling algorithm to convert the second allocation plan into a task schedule, wherein the task schedule includes: multiple enterprise tasks, the execution object assigned to each enterprise task, and the start time and end time of each enterprise task; using a graph neural network to adjust the start time and end time of each enterprise task until preset conditions are met to obtain a target task schedule, wherein the preset conditions at least include: the total completion time of all enterprise tasks is the shortest; uploading the target task schedule to the enterprise management system; converting the task schedule into a target graphic, and displaying the target graphic on a graphic display interface, wherein the target graphic includes at least one of the following: an enterprise task processing flow chart, an enterprise task processing timetable, and an enterprise task allocation chart.

[0015] According to another aspect of the embodiment of the present application, there is also provided a device for allocating enterprise tasks, including: an acquisition module, used to acquire multiple data related to enterprise tasks from an enterprise management system to obtain a data set; a generation module, used to generate a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent enterprise tasks and execution objects of enterprise tasks, and the edges are used to represent correlation indicators between execution objects and enterprise tasks, priority differences between enterprise tasks and correlation indicators between enterprise tasks; an iteration module, used to use a graph neural network to iteratively update feature vectors corresponding to nodes to obtain a first feature vector set output by the graph neural network; a first determination module, used to determine a first allocation scheme for enterprise tasks based on the first feature vector set; an update module, used to update the task relationship graph when any data in the data set changes to obtain a target task relationship graph, and use the graph neural network to iteratively update the feature vectors corresponding to target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; a second determination module, used to determine a second allocation scheme for enterprise tasks based on the second feature vector set.

[0016] According to another aspect of the embodiment of the present application, a non-volatile storage medium is provided, the storage medium includes a stored program, wherein when the program runs, the device where the storage medium is located is controlled to execute the above enterprise task allocation method.

[0017] According to another aspect of the embodiment of the present application, there is also provided an electronic device, including: a memory and a processor, the processor being used to run a program stored in the memory, wherein the above enterprise task allocation method is executed when the program is run.

[0018] According to yet another aspect of the embodiments of the present application, a computer program is further provided, wherein when the computer program is executed by a processor, the above enterprise task allocation method is implemented.

[0019] According to another aspect of the embodiment of the present application, a computer program product is also provided, which includes a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for allocating enterprise tasks is implemented.

[0020] In an embodiment of the present application, a plurality of data related to enterprise tasks are obtained from an enterprise management system to obtain a data set; a task relationship graph is generated based on the data set, wherein the task relationship graph includes a plurality of nodes and edges, the nodes are used to represent enterprise tasks and execution objects of enterprise tasks, and the edges are used to represent correlation indicators between execution objects and enterprise tasks, priority differences between enterprise tasks, and correlation indicators between enterprise tasks; a graph neural network is used to iteratively update feature vectors corresponding to nodes to obtain a first feature vector set output by the graph neural network; a first allocation scheme for enterprise tasks is determined based on the first feature vector set; when any data in the data set changes, the task relationship graph is updated to obtain a target task relationship graph, and a graph neural network is used to iteratively update feature vectors corresponding to target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; a second allocation scheme for enterprise tasks is determined based on the second feature vector set, thereby achieving the purpose of flexibly responding to real-time changing enterprise needs, thereby achieving the technical effect of improving the management efficiency and management accuracy of enterprise tasks, and further solving the technical problem of low efficiency and accuracy of enterprise task management caused by the fact that enterprise task management technology mainly relies on static rules and preset algorithms to perform task allocation and cannot flexibly respond to real-time changing enterprise needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0022] Figure 1 is a flow chart of a method for allocating enterprise tasks according to an embodiment of the present application;

[0023] Figure 2 is a flow chart of another method for allocating enterprise tasks according to an embodiment of the present application;

[0024] Figure 3 is a structural diagram of a device for allocating enterprise tasks according to an embodiment of the present application;

[0025] Figure 4 It is a hardware structure block diagram of a computer terminal according to a method for allocating enterprise tasks in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] According to an embodiment of the present application, a method embodiment of a method for allocating enterprise tasks is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Figure 1 is a flow chart of a method for allocating enterprise tasks according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0030] Step S101, acquiring multiple data related to enterprise tasks from the enterprise management system to obtain a data set.

[0031] The data related to the enterprise tasks include, but are not limited to: enterprise demand information, wherein the enterprise demand information includes: demand description information, start time, end time, business field and demand priority information.

[0032] Step S102, generating a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent the enterprise tasks and the execution objects of the enterprise tasks, and the edges are used to represent the correlation indicators between the execution objects and the enterprise tasks, the priority differences between the enterprise tasks, and the correlation indicators between the enterprise tasks.

[0033] The object of execution of enterprise tasks, for example: experts.

[0034] For example, a task relationship graph is: G = (V, E). V represents a node set, including all experts and tasks, and E represents an edge set, representing the relationship between nodes: V = {v1, v2, ..., v n}, represents the node set containing experts and tasks; E = {e ij |v i ,v j ∈V,i≠j}, which represents the set of edges between nodes.

[0035] Step S103, using the graph neural network, iteratively update the feature vectors corresponding to the nodes to obtain the first feature vector set output by the graph neural network.

[0036] Among them, the graph neural network includes: multiple network layers.

[0037] Step S104: determining a first allocation scheme for enterprise tasks according to the first feature vector set.

[0038] Step S105, when any data in the data set changes, the task relationship graph is updated to obtain the target task relationship graph, and the feature vector corresponding to the target node in the target task relationship graph is iteratively updated using the graph neural network to obtain a second feature vector set output by the graph neural network.

[0039] Step S106: Determine a second allocation plan for enterprise tasks based on the second feature vector set.

[0040] According to some optional embodiments of the present application, step S101 may be implemented by the following method:

[0041] Step S1, determining a data collection protocol according to data types of different data related to enterprise tasks and verification rules corresponding to different data types.

[0042] For example, a data collection protocol C is set to specify the data collection method and format. Specifically, the data collection protocol includes data type definition and data validation rules:

[0043] C={type(D i ), validate(D i )∣D i ∈D}

[0044] Among them, D i Represents the i-th item in the data set D, type(D i ) specifies D iThe data type, validate(D i ) indicates that D i The validation rules to apply.

[0045] Step S2, extracting enterprise demand information in the enterprise management system according to the data collection protocol, wherein the enterprise demand information includes at least one of the following: demand description information, start time, end time, business field and demand priority information; and extracting enterprise scale information related to the enterprise demand information, wherein the enterprise scale information includes at least one of the following: the number of employees of the enterprise, the annual revenue of the enterprise and the market share of the enterprise; and extracting execution object information of the execution object, wherein the execution object is used to process enterprise tasks, and the execution object information includes at least one of the following: educational background, number of historical processing cases and execution field.

[0046] That is to say, through the interface of the enterprise management system, the data collection task is initiated according to the data collection protocol.

[0047] First, collect enterprise demand information R, where the enterprise demand information R can be expressed in the following form:

[0048] R = {r j ∣r j =(desc(r j ), time(r j ), domain(r j ), priority(r j )), j = 1, ..., n}

[0049] Among them, each demand r j Contains the requirement description information desc(r j ), expected resolution time (r j ), business domain(r j ) and demand priority information priority(r j ). The expected resolution time includes: start time and end time.

[0050] Secondly, collect enterprise scale information S, where the enterprise scale information S can be expressed in the following form:

[0051] S={s k ∣s k =(employees(k), revenue(k), market_share(k)), k=1,…,m}

[0052] Among them, each s kIncluding the number of employees employees(k), the annual revenue revenue(k) and the market share market_share(k).

[0053] Next, collect the execution object information (expert experience) Q of the execution object, where the execution object information Q can be expressed in the following form:

[0054] Q = {q l ∣q l =(education(l),cases(l),skills(l)),l=1,…,p}

[0055] The data of each expert ql includes his / her education background education(l), the number of cases he / she has handled historically cases(l), and specific skills (execution areas) skills(l).

[0056] Step S3, determining the agreed time according to the start time and the end time.

[0057] For each enterprise demand r∈R, identify the associated agreed time t r , including the start time and expected end time of the task.

[0058] t r ={(t start , t end )|r∈R}

[0059] Here start and t end They represent the start and estimated end time of the task respectively. It is worth noting that the time label will directly affect the scheduling and priority setting of the task.

[0060] Step S4, formatting and standardizing the enterprise demand information, enterprise scale information, execution object information and agreed time to obtain a data set.

[0061] Specifically, all the collected data R, t r , S, Q are formatted and standardized to ensure the consistency and accuracy of data input, and stored in the central data set D to provide support for further processing and analysis of the graph neural network. At the same time, it is ensured that the central data set D fully reflects the integration and standardization process of enterprise needs, agreed time, enterprise scale information, importance, and execution object information:

[0062]

[0063] Among them, desc(r) represents the requirement description information, time(r) represents the agreed time of the requirement, domain(r) represents the affected business domain, priority(r) represents the requirement priority information, and Respectively represent the start time and expected end time of the task, s employees 、s revenue and market_share They represent the number of employees, annual revenue and market share of the enterprise respectively. education ,q cases and q skills Respectively represent each expert’s educational background, number of historical cases handled, and expert field.

[0064] According to some other optional embodiments of the present application, generating a task relationship diagram according to the data set in step S102 can be implemented by the following method:

[0065] Step S1, extracting task information corresponding to each enterprise task from the data set, and encoding the task information into a first feature vector, wherein the task information includes at least one of the following: enterprise demand, task time, and task priority.

[0066] For example, a task relationship graph is: G = (V, E). V represents a node set, including all experts and tasks, and E represents an edge set, representing the relationship between nodes: V = {v1, v2, ..., v n}, represents the node set containing experts and tasks; E = {e ij |v i ,v j ∈V,i≠j}, which represents the set of edges between nodes.

[0067] For each task node t∈V, the task related information extracted from the dataset D, such as enterprise requirements, task time, and task priority, is encoded as a node feature vector

[0068]

[0069] The encode function is used to generate a feature vector of a task node according to enterprise requirements (desc), task time (time) and task priority (priority).

[0070] Step S2: extracting execution object information corresponding to each execution object from the data set, and encoding the execution object information into a second feature vector.

[0071] For each expert node e∈V, the expert qualification and experience information extracted from the central dataset D, such as educational background, number of historical cases handled, and specific skills (expert field), is encoded as a node feature vector

[0072] The encode function is used to generate the feature vector of the expert node based on the expert's educational background (education), number of historical cases handled (cases) and specific skills (skills).

[0073] Step S3, determining a first edge weight corresponding to the first edge according to the execution object field and the enterprise demand, wherein the first edge is an edge formed by the execution object and the enterprise task, and the first edge weight includes: a correlation index between the execution object and the enterprise task.

[0074] The first edge is the edge (e, t) between the expert and the enterprise task. The weight of the first edge is determined by the compatibility (correlation) between the expert e and the task t. The compatibility is calculated by the matching degree between the expert's skills and the task requirements:

[0075] w et =α·skill_match(skills(e),req_skills(t))+β·time_fit(availability(e),time(t));

[0076] The skill_match function is used to calculate the matching degree between the expert's skills and the skills required by the task, and the time_fit function is used to calculate the matching degree between the expert's available time and the task time.

[0077] Step S4, determining the second edge weight corresponding to the second edge according to the priority difference evaluation function and the correlation index evaluation function, wherein the second edge is an edge formed between enterprise tasks, and the second edge weight includes: the priority difference and the correlation index between the enterprise tasks.

[0078] The second edge is the edge between tasks (t i , t j ), the second edge weight By task i and t j The dependency (correlation index) and priority difference are calculated as follows:

[0079]

[0080] Among them, the priority_diff function is used to calculate the priority difference between two tasks, the dependency function is used to calculate the dependency between two tasks, and γ and δ are coefficients for adjusting the impact of priority difference and dependency.

[0081] Step S5, generating a task relationship graph according to the first eigenvector, the second eigenvector, the first edge weight, and the second edge weight.

[0082] Using the encoded node feature vector and and the edge weight w et and Improve the structure of the relationship graph G to ensure that the information in the graph accurately reflects the dynamic relationship between experts and tasks and between tasks:

[0083]

[0084] It can be understood that the above function utilizes all node feature vectors and edge weight information to obtain the optimized task relationship graph.

[0085] In some optional embodiments of the present application, when any data in the data set changes, updating the task relationship can be achieved by the following method:

[0086] According to the preset time interval, determine whether the data set has changed. If any data in the data set changes, update the task relationship diagram through the following method:

[0087] For example, every 5 minutes, determine whether the data set has changed. Alternatively, monitor the data set to determine whether the data set has changed. In other words, monitor expert status information and enterprise demand information, and receive data on expert availability changes and new or changed enterprise requirements in real time.

[0088] In the case of a change in task information, the changed task information is encoded as a third feature vector, and the first target feature vector corresponding to the task information is determined based on the first feature vector and the third feature vector corresponding to the task information before the change. In the case of a change in execution object information, the changed execution object information is encoded as a fourth feature vector, and the second target feature vector corresponding to the execution object information is determined based on the second feature vector and the fourth feature vector corresponding to the execution object information before the change. In the case of a change in the execution object field and / or enterprise needs, the first edge weight corresponding to the first edge is updated.

[0089] Specifically, for the affected expert node e, update its node feature vector To reflect the state change:

[0090]

[0091] in, The new state information is converted through a coding function to represent the change of the expert state; for new or updated enterprise requirements t, update its node feature vector To reflect the changes in requirements:

[0092]

[0093] in, Indicates requirement change information converted by encoding function.

[0094] Further, the task relationship graph is updated according to the first target feature vector, the second target feature vector, the updated first edge weight, and the second edge weight.

[0095] Specifically, the improved graph neural network N is used to reprocess the updated relationship graph G, and an adaptive update strategy is adopted to adjust the node features and edge weights to reflect the impact of data changes:

[0096]

[0097] Among them, η is the learning rate and g is the update function.

[0098] As some optional embodiments of the present application, using a graph neural network, iteratively updating the feature vectors corresponding to the target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network can be implemented by the following method:

[0099] Step S1, configure the initialization parameters of the graph neural network, wherein the initialization parameters include: adaptive learning rate and network hierarchy parameters; Step S2, the configured graph neural network iteratively updates the feature vector corresponding to the node according to the feature vector corresponding to the target node in the task relationship graph at time t, the adaptive learning rate at time t, the set of neighbor nodes of the target node, the edge weight of the edge where the target node is located, and the feature vector corresponding to the target node at time t+1, wherein t is a positive number and the target node is any node in the task relationship graph; Step S3, after each iteration, adjust the adaptive learning rate according to the preset learning rate adjustment factor and the network training error rate of each iteration; Step S4, repeat steps S2 and S3 until the maximum number of iterations is reached or the adaptive learning rate reaches the preset convergence condition, and the first feature vector set output by the graph neural network is obtained.

[0100] Specifically, in step S2, the feature vector of each node is iteratively updated by the following formula:

[0101]

[0102] in, represents the feature vector of node i at iteration t, η (t) is the adaptive learning rate at iteration t, N(i) is the set of neighbor nodes of node i, and w ij is the weight of edge (i, j), f and g are the functions of node update and neighbor information aggregation, respectively.

[0103] In step S3, after each iteration, the learning rate is dynamically adjusted to optimize the learning process according to the training error of the network and the predetermined adjustment strategy:

[0104] η (t+1) =η (t) ×(1-δ·error_rate(t))

[0105] Where δ is the learning rate adjustment factor and error_rate(t) is the network training error rate at iteration t.

[0106] It should be noted that the graph neural network includes: multiple network layers. Step S1, configuring the initialization parameters of the graph neural network, can be achieved by the following method:

[0107] Step S11, determining the adaptive learning rates of different network levels according to the baseline learning rate and each network level in the graph neural network.

[0108] This embodiment adopts a hierarchical adaptive learning rate strategy, in which each layer of the network automatically adjusts the learning rate η according to the abstract level of the data it processes. l , high-level abstraction layers use lower learning rates to stabilize the learning process:

[0109] η l =η0·e -λ·l

[0110] Among them, η0 is the baseline learning rate, λ is the decay constant, and l represents the network layer.

[0111] Step S12, determining the initial network weights of the graph neural network according to a preset weight distribution pattern and parameters of the preset weight distribution pattern, wherein the preset weight distribution pattern includes at least: normal distribution or uniform distribution, and the parameters of the preset weight distribution pattern include: mean and standard deviation.

[0112] This embodiment uses the initial network weight θ initial The intelligent distribution strategy adjusts the initial weights according to a predefined weight distribution pattern (such as normal distribution or uniform distribution) to promote information flow and convergence speed in the early stages of the network:

[0113] θ initial =init_weight_distribution(dist_type,dist_params)

[0114] Among them, dist_type represents the selected weight distribution type, and dist_params represents the parameters of the distribution (such as mean and standard deviation).

[0115] Step S13, determining a target activation function for each network level according to a preset activation function and an adaptive adjustment parameter of each network level, wherein the adaptive adjustment parameter is associated with a statistical characteristic of the data in the data set.

[0116] This embodiment adopts a conditional activation function adjustment mechanism, which allows the response characteristics of the activation function to be adaptively adjusted according to the dynamic range of the incoming features, thereby enhancing the network's sensitivity and ability to distinguish different data features:

[0117]

[0118] Among them, σ is the basic activation function (such as ReLU or Sigmoid), α l and β l It is an adaptive adjustment parameter, which is dynamically adjusted according to the statistical characteristics of the input data.

[0119] In some optional embodiments of the present application, determining the first allocation scheme for enterprise tasks according to the first feature vector set can be implemented by the following method:

[0120] Step S1, performing standardization processing on a first feature vector set according to a preset feature vector conversion function to obtain a target vector set, wherein the target vector set includes: an execution object vector and an enterprise task vector.

[0121] Preset feature vector conversion functions such as:

[0122]

[0123] Among them, the normalize function ensures that each feature vector is standardized to unit length for subsequent processing.

[0124] Step S2, determining a scoring function according to the similarity coefficient between the execution object vector and the enterprise task vector.

[0125] Define the task assignment scoring function score(t,e) to calculate the matching score of each task t for each expert e. The scoring function takes into account the similarity between the task and the expert as well as the inherent properties of the task:

[0126]

[0127] in, Using cosine similarity calculation, importance(t) and size(t) are the importance and size of the task, respectively. t and η t is a coefficient that adjusts for the effects of importance and size.

[0128] Step S3: Determine the execution object corresponding to each enterprise task according to the scoring function to obtain a first initial allocation plan.

[0129] A preliminary task allocation plan is generated for each task according to the score function, which includes selecting an expert with the highest score for each task. When generating the preliminary task allocation plan, an optimization problem is used to solve resource allocation:

[0130]

[0131] Among them, x te is a binary variable indicating whether task t is assigned to expert e.

[0132] Step S4, processing the first initial allocation plan according to preset constraints to obtain a first allocation plan, wherein the preset constraints at least include: there is no conflict in the time when the same execution object processes different enterprise tasks.

[0133] Using preset constraints, such as time conflicts and expert availability, preliminary task allocation plans are adjusted to ensure that all allocations are practical and reasonable.

[0134] Examples of preset constraints are:

[0135]

[0136] availability(e,t)=not(V t′∈conflicts(t) x et′ )

[0137] Among them, availability(e,t) is used to check whether a given expert is available within the task time and ensure that there is no time conflict.

[0138] Furthermore, the final preliminary diagnosis task allocation plan is generated, which takes into account the time, scale and importance labels of all tasks, as well as the adaptability and availability of experts, to ensure the optimization and accuracy of the overall task allocation:

[0139] FinalPlan={(t,e)|x′ te =1};

[0140] It can be appreciated that the above set provides the final task assignment list after all evaluations and adjustments.

[0141] As some other optional embodiments of the present application, after determining the second allocation scheme for the enterprise tasks according to the second feature vector set, the following steps may be further performed:

[0142] Step S1, using a preset scheduling algorithm to convert the second allocation plan into a task schedule, wherein the task schedule includes: a plurality of enterprise tasks, an execution object assigned to each enterprise task, and a start time and an end time of each enterprise task.

[0143] The task allocation decision is converted into a specific task schedule, including detailed information such as the experts assigned to each enterprise task, the start time and end time of each enterprise task. This conversion process uses a scheduling algorithm, which is expressed as:

[0144] schedule(t)=(assignee(t),start_time(t),end_time(t))

[0145] Here, assignee(t) is the expert assigned to task t, start_time(t) and end_time(t) represent the start and end time of the task, respectively.

[0146] Step S2, using the graph neural network, adjust the start time and end time of each enterprise task until the preset conditions are met to obtain the target task schedule, where the preset conditions at least include: the total completion time of all enterprise tasks is the shortest. Through this step, the analysis and prediction capabilities of the improved graph neural network are used to optimize the time schedule of each task to minimize time conflicts and resource waste, ensuring that each task is performed by the most appropriate expert at the most appropriate time.

[0147] Step S3, uploading the target task schedule to the enterprise management system. Through this step, the final task schedule is output to the enterprise management system to guide the implementation of the diagnosis work and improve the quality control and overall efficiency of the offline diagnosis work.

[0148] Step S4, converting the task schedule into a target graph, and displaying the target graph on the graphic display interface, wherein the target graph includes at least one of the following: an enterprise task processing flow chart, an enterprise task processing timetable, and an enterprise task allocation chart. Through this step, a visual task schedule on the terminal user interface is generated, providing an intuitive operation interface for the enterprise management system.

[0149] According to the above steps, a plurality of data related to enterprise tasks are obtained from the enterprise management system to obtain a data set; a task relationship graph is generated based on the data set, wherein the task relationship graph includes a plurality of nodes and edges, the nodes are used to represent the enterprise tasks and the execution objects of the enterprise tasks, and the edges are used to represent the correlation indicators between the execution objects and the enterprise tasks, the priority differences between the enterprise tasks, and the correlation indicators between the enterprise tasks; a graph neural network is used to iteratively update the feature vectors corresponding to the nodes to obtain a first feature vector set output by the graph neural network; a first allocation scheme for the enterprise tasks is determined based on the first feature vector set; when any data in the data set changes, the task relationship graph is updated to obtain a target task relationship graph, and a graph neural network is used to iteratively update the feature vectors corresponding to the target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; a second allocation scheme for the enterprise tasks is determined based on the second feature vector set, thereby achieving the purpose of flexibly responding to real-time changing enterprise needs, thereby realizing the technical effect of improving the management efficiency and accuracy of enterprise tasks.

[0150] Figure 2 is a flow chart of another method for allocating enterprise tasks according to an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0151] Step S201, collect task-related data from the enterprise management system, including enterprise needs, agreed time, enterprise size, importance, and expert qualifications and experience information.

[0152] Step S202, constructing a relationship graph, in which nodes include experts and tasks, and edges represent the compatibility between experts and tasks and the priority and dependency relationship between tasks.

[0153] Step S203, use the improved graph neural network to process the relationship graph of step S202, and the improved graph neural network enhances the learning efficiency and accuracy of node features and edge relationships through adaptive learning rate and optimized propagation algorithm.

[0154] Step S204, automatically generate a preliminary diagnostic task allocation plan using the output of the improved graph neural network and preset labels (time, scale, and importance of the task).

[0155] Step S205, update the status of experts and changes in enterprise needs to the relationship diagram in real time, and use the improved graph neural network to dynamically adjust task allocation to cope with sudden changes.

[0156] Step S206, recording the industrial diagnosis experience and project history of each expert according to the expert's professional information database to optimize the matching accuracy between the expert and the task.

[0157] Step S207, output the final task schedule, and use the intelligent analysis and prediction capabilities of the improved graph neural network to guide the enterprise management system to implement the diagnostic work, thereby improving the quality control and overall efficiency of offline diagnostic work.

[0158] According to some optional embodiments of the present application, step S201 specifically includes the following steps.

[0159] Step S2011, set the data collection protocol C, which is used to specify the data collection method and format, including data type definition and data validation rules:

[0160] C={type(D i ), validate(D i )∣D i ∈D}

[0161] Among them, D i Represents any data item in the data set D, type(D i ) specifies D i The data type, validate(D i ) indicates that D i The validation rules to apply.

[0162] Step S2012, through the interface of the enterprise management system, start the data collection task to automatically extract the enterprise requirements R, which include the description of the requirements, the expected resolution time, the affected business areas and the priority of the requirements:

[0163] R = {r j ∣r j =(desc(r j ), time(r j ), domain(r j ), priority(r j )), j = 1, ..., n};

[0164] Among them, each demand r j Contains description desc(r j ), expected resolution time (r j ), the business domain affected j ) and the priority of the requirement (r j ).

[0165] Step S2013: for each enterprise demand r∈R, identify the associated agreed time t r , including the start time and expected end time of the task. The time label will directly affect the scheduling and priority setting of the task:

[0166] t r ={(t start , t end )|r∈R}

[0167] Among them, t start and t end Respectively represent the start and estimated end time of the task.

[0168] Step S2014: collect enterprise scale information S related to each requirement, such as the total number of employees, annual income and market share of the enterprise, to determine resource allocation and priority adjustment when allocating tasks:

[0169] S={s k ∣s k =(employees(k), revenue(k), market_share(k)), k=1,…,m}

[0170] Among them, each s k Includes the company's number of employees (k), annual revenue (k) and market share (k).

[0171] In step S2015, the system will evaluate the qualifications and experience Q of the available experts. The data of each expert q∈Q includes his / her educational background, historical diagnosis cases and specific skills. This information is used in the subsequent matching process between experts and tasks:

[0172] Q = {q l ∣q l =(education(l),cases(l),skills(l)),l=1,…,p}

[0173] Among them, each expert q l The data includes education background education(l), number of historical diagnosis cases cases(l), and specific skills skills(l).

[0174] Step S2016, all the collected data R, t r , S, Q are formatted and standardized to ensure the consistency and accuracy of data input, and stored in the central data set D to provide support for further processing and analysis of the graph neural network. At the same time, ensure that the central data set D fully reflects the needs of the enterprise, the agreed time, the size of the enterprise, the importance, and the integration and standardization process of the expert qualifications and experience information:

[0175]

[0176] Where desc(r) indicates the description of the requirement, time(r) indicates the scheduled time of the requirement, domain(r) indicates the affected business domain, and priority(r) indicates the priority of the requirement. and Respectively represent the start time and expected end time of the task, s employees 、s revenue and market_share Represent the number of employees, annual income and market share of the enterprise respectively, q education ,q cases and q skills Respectively represent each expert's educational background, historical diagnostic cases, and specific skills.

[0177] According to some other optional embodiments of the present application, step S202 specifically includes the following steps.

[0178] Step S2021, define a relationship graph: G = (V, E).

[0179] Where V represents the node set, including all experts and tasks, and E represents the edge set, representing the relationship between nodes: V = {v1,v2,...,v n}, represents the node set containing experts and tasks; E = {e ij |v i ,v j ∈V,i≠j}, which represents the set of edges between nodes.

[0180] Step S2022: for each task node t∈V, the task-related information extracted from the central dataset D, such as task description, time, priority, is encoded into a node feature vector

[0181]

[0182] Among them, the encode function generates the feature vector of the task node according to the task description (desc), time (time) and priority (priority).

[0183] Step S2023, for each expert node e∈V, the expert qualification and experience information extracted from the central data set D, such as educational background, number of cases, and specific skills, is encoded as a node feature vector

[0184]

[0185] The encode function generates a feature vector of the expert node based on the expert's educational background (education), number of cases (cases), and specific skills (skills).

[0186] Step S2024, construct an edge set E, which includes:

[0187] The edge (e, t) between the expert and the task, the weight of the edge is w et It is determined by the compatibility between expert e and task t, which is calculated by the matching degree between the expert's skills and the task requirements:

[0188] w et =α·skill_match(skills(e),req_skills(t))+β·time_fit(availability(e),time(t))

[0189] Among them, the skill_match function calculates the matching degree between the expert's skills and the skills required by the task, and the time_fit function calculates the matching degree between the expert's available time and the task time.

[0190] The edge between tasks (t i ,t j ), the edge weight According to the task i and t j The dependency and priority differences are calculated as follows:

[0191]

[0192] Among them, the priority_diff function calculates the priority difference between two tasks, the dependency function evaluates the dependency between two tasks, and γ and δ are coefficients for adjusting the impact of priority difference and dependency.

[0193] Step S2025, using the encoded node feature vector and and the edge weight w et and Improve the structure of the relationship graph G to ensure that the information in the graph accurately reflects the dynamic relationship between experts and tasks and between tasks:

[0194]

[0195] The above function uses all node feature vectors and edge weight information to integrate into the final relationship graph structure.

[0196] In some optional embodiments of the present application, step S203 specifically includes the following steps.

[0197] Step S2031, configure the initialization parameters of the improved graph neural network N, including adaptive learning rate and network hierarchy parameters.

[0198] Step S2032, input the relationship graph G = (V, E) constructed in step S2032 into the graph neural network N.

[0199] Step S2033, apply the improved graph neural network propagation algorithm, and iteratively update the feature vector of each node through the following formula.

[0200]

[0201] in, represents the feature vector of node i at iteration t, η (t) is the adaptive learning rate at iteration t, N(i) is the set of neighbor nodes of node i, and w ij is the weight of edge (i, j), f and g are the functions of node update and neighbor information aggregation, respectively.

[0202] Step S2034, after each iteration, dynamically adjust the learning rate to optimize the learning process according to the network training error and the predetermined adjustment strategy:

[0203] η (t+1) =η (t) ×(1-δ·error_rate(t))

[0204] Where δ is the learning rate adjustment factor and error_rate(t) is the network training error rate at iteration t.

[0205] Step S2035, repeat steps S2033 and S2034 until the preset convergence condition or the maximum number of iterations is reached, so as to ensure that the node features and edge relationships are fully learned and optimized in the graph neural network.

[0206] Preferably, this embodiment proposes an improved graph neural network N and configures initialization parameters of the improved graph neural network N. This process is specially designed to adapt to the characteristics of enterprise diagnosis tasks and the complexity of expert data. The initialization parameter settings include:

[0207] By adopting a hierarchical adaptive learning rate strategy, each layer of the network automatically adjusts the learning rate η according to the abstract level of the data it processes. l , high-level abstraction layers use lower learning rates to stabilize the learning process:

[0208] η l =η0·e -λ·l

[0209] Among them, η0 is the baseline learning rate, λ is the decay constant, and l represents the network layer.

[0210] By designing the initial network weights θinitial The intelligent distribution strategy adjusts the initial weights according to a predefined weight distribution pattern (such as normal distribution or uniform distribution) to promote information flow and convergence speed in the early stages of the network:

[0211] θ initial =init_weight_distribution(dist_type,dist_params)

[0212] Among them, dist_type represents the selected weight distribution type, and dist_params represents the parameters of the distribution (such as mean and standard deviation).

[0213] Improved graph neural networks innovatively introduce a conditional activation function adjustment mechanism, which allows the response characteristics of the activation function to be adaptively adjusted according to the dynamic range of the incoming features, enhancing the network's sensitivity and ability to distinguish different data features:

[0214]

[0215] Among them, σ is the basic activation function (such as ReLU or Sigmoid), α l and β l It is an adaptive adjustment parameter, which is dynamically adjusted according to the statistical characteristics of the input data.

[0216] As some optional embodiments of the present application, step S204 specifically includes the following steps.

[0217] Step S2041, receiving the optimized node feature vector set output by the improved graph neural network N in step S36 To demonstrate data processing, a feature vector conversion function is designed to ensure that the vector meets the format requirements for subsequent processing:

[0218]

[0219] Among them, the normalize function ensures that each feature vector is standardized to unit length for subsequent processing.

[0220] Step S2042, define the task assignment scoring function score(t,e), calculate the matching score of each task t to each expert e, and the scoring function considers the similarity between the task and the expert and the inherent attributes of the task:

[0221]

[0222] in, Using cosine similarity calculation, importance(t) and size(t) are the importance and size of the task, respectively. tand η t is a coefficient that adjusts for the effects of importance and size.

[0223] Step S2043, a preliminary task allocation plan is generated for each task according to the score function score, which includes selecting an expert with the highest score for each task. When generating the preliminary task allocation plan, an optimization problem is used to solve resource allocation:

[0224]

[0225]

[0226] Among them, x te is a binary variable indicating whether task t is assigned to expert e.

[0227] Step S2044, using preset constraints, such as time conflicts and expert availability, adjust the preliminary task allocation plan to ensure the practicality and rationality of all allocations:

[0228]

[0229] availability(e,t)=not(V t′∈conflicts(t) x et′ )

[0230] Among them, availability(e,t) checks whether a given expert is available within the task time and ensures that there is no time conflict.

[0231] Step S2045, generating a final preliminary diagnosis task allocation plan, which takes into account the time, scale and importance labels of all tasks, as well as the adaptability and availability of experts, to ensure the optimization and accuracy of the overall task allocation:

[0232] FinalPlan={(t,e)|x′ te =1}

[0233] The above collection provides the final task assignment list after all evaluations and adjustments.

[0234] In some optional embodiments of the present application, step S205 specifically includes the following steps.

[0235] Step S2051, monitor the status of experts and enterprise needs, and receive data about changes in expert availability and new or changed enterprise needs in real time.

[0236] Step S2052, in order to cope with these real-time updates, the relationship graph G=(V, E) is adjusted, including:

[0237] For the affected expert node e, update its node feature vector To reflect the state change:

[0238]

[0239] in, The new state information is converted through a coding function to represent the change of the expert state; for new or updated enterprise requirements t, update its node feature vector To reflect the changes in requirements:

[0240]

[0241] in, Indicates requirement change information converted by encoding function.

[0242] Step S2053: Use the improved graph neural network N to reprocess the updated relationship graph G, adopt an adaptive update strategy, and adjust the node features and edge weights to reflect the impact of real-time data:

[0243]

[0244] Here, η is the learning rate and g is the update function.

[0245] Step S2054, dynamically adjust the task allocation plan based on the updated node feature vectors and edge weights to ensure that the plan can respond to the latest availability of experts and urgent changes in the company's needs.

[0246] Step S2055, outputting the updated task allocation plan, providing a responsive and timely task management solution, optimizing resource allocation and reducing task execution delays.

[0247] As other optional embodiments of the present application, step S207 specifically includes the following steps.

[0248] Step S2071, obtaining the final optimized node feature vector set and task allocation decision from the improved graph neural network N.

[0249] Step S2072, convert the task allocation decision into a specific task schedule, including detailed information such as the specific execution expert, start and end time of each task. This conversion process uses a scheduling algorithm, which is expressed as:

[0250] schedule(t)=(assignee(t),start_time(t),end_time(t))

[0251] Here, assignee(t) is the expert assigned to task t, start_time(t) and end_time(t) represent the start and end time of the task, respectively.

[0252] Step S2073, using the analysis and prediction capabilities of the improved graph neural network, optimize the schedule of each task to minimize time conflicts and resource waste, and ensure that each task is performed by the most appropriate expert at the most appropriate time.

[0253] Step S2074, generating a visual task schedule on the terminal user interface, providing an intuitive operation interface for the enterprise management system, including a task flow chart, a timetable and a resource allocation chart, so that managers can monitor and adjust the schedule.

[0254] Step S2075, output the final task schedule to the enterprise management system to guide the implementation of the diagnosis work and improve the quality control and overall efficiency of offline diagnosis work.

[0255] Figure 3 is a structural diagram of a device for allocating enterprise tasks according to an embodiment of the present application, such as Figure 3 As shown, the device comprises:

[0256] The acquisition module 31 is used to acquire multiple data related to enterprise tasks from the enterprise management system to obtain a data set.

[0257] The generation module 32 is used to generate a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent the enterprise tasks and the execution objects of the enterprise tasks, and the edges are used to represent the correlation indicators between the execution objects and the enterprise tasks, the priority differences between the enterprise tasks and the correlation indicators between the enterprise tasks.

[0258] The iteration module 33 is used to iteratively update the feature vectors corresponding to the nodes using the graph neural network to obtain a first feature vector set output by the graph neural network.

[0259] The first determination module 34 is used to determine a first allocation scheme for enterprise tasks according to the first feature vector set.

[0260] The updating module 35 is used to update the task relationship graph when any data in the data set changes, obtain the target task relationship graph, and use the graph neural network to iteratively update the feature vector corresponding to the target node in the target task relationship graph to obtain the second feature vector set output by the graph neural network.

[0261] The second determination module 36 is used to determine a second allocation scheme for the enterprise tasks according to the second feature vector set.

[0262] Optionally, the acquisition module 31 is also used to perform the following steps: determine the data collection protocol according to the data types of different data related to the enterprise tasks and the verification rules corresponding to the different data types; extract the enterprise demand information in the enterprise management system according to the data collection protocol, wherein the enterprise demand information includes at least one of the following: demand description information, start time, end time, business field and demand priority information; and extract the enterprise scale information related to the enterprise demand information, wherein the enterprise scale information includes at least one of the following: the number of employees of the enterprise, the annual revenue of the enterprise and the market share of the enterprise; and extract the execution object information of the execution object, wherein the execution object is used to process the enterprise tasks, and the execution object information includes at least one of the following: educational background, number of historical processing cases and execution field; determine the agreed time according to the start time and the end time; format and standardize the enterprise demand information, enterprise scale information, execution object information and agreed time to obtain a data set.

[0263] Optionally, the generation module 32 is also used to perform the following steps: extracting task information corresponding to each enterprise task from the data set, and encoding the task information into a first feature vector, wherein the task information includes at least one of the following: enterprise requirements, task time, and task priority; extracting execution object information corresponding to each execution object from the data set, and encoding the execution object information into a second feature vector; determining a first edge weight corresponding to the first edge according to the execution field and enterprise requirements, wherein the first edge is an edge formed by the execution object and the enterprise task, and the first edge weight includes: a correlation index between the execution object and the enterprise task; determining a second edge weight corresponding to the second edge according to a priority difference evaluation function and a correlation index evaluation function, wherein the second edge is an edge formed between enterprise tasks, and the second edge weight includes: a priority difference and a correlation index between enterprise tasks; generating a task relationship diagram according to the first feature vector, the second feature vector, the first edge weight, and the second edge weight.

[0264] Optionally, the update module 35 is also used to perform the following steps: determine whether the data set has changed according to a preset time interval, and when any data in the data set changes, update the task relationship diagram by the following method: when the task information changes, encode the changed task information into a third feature vector, and determine the first target feature vector corresponding to the task information based on the first feature vector and the third feature vector corresponding to the task information before the change; when the execution object information changes, encode the changed execution object information into a fourth feature vector, and determine the second target feature vector corresponding to the execution object information based on the second feature vector and the fourth feature vector corresponding to the execution object information before the change; when the execution field and / or enterprise needs change, update the first edge weight corresponding to the first edge; update the task relationship diagram based on the first target feature vector, the second target feature vector, the updated first edge weight, and the second edge weight.

[0265] Optionally, the update module 35 is also used to perform the following steps: step S1, configuring the initialization parameters of the graph neural network, wherein the initialization parameters include: an adaptive learning rate and network hierarchical structure parameters; step S2, completing the configuration of the graph neural network, based on the feature vector corresponding to the target node in the task relationship graph at time t, the adaptive learning rate at time t, the set of neighbor nodes of the target node, the edge weight of the edge where the target node is located, and the feature vector corresponding to the target node at time t+1, to iteratively update the feature vector corresponding to the node, wherein t is a positive number, and the target node is any node in the task relationship graph; step S3, after each iteration, adjusting the adaptive learning rate according to the preset learning rate adjustment factor and the network training error rate of each iteration; step S4, repeating steps S2 and S3 until the maximum number of iterations is reached or the adaptive learning rate reaches the preset convergence condition, to obtain the first feature vector set output by the graph neural network.

[0266] Optionally, the graph neural network includes: multiple network layers. The update module 35 is also used to perform the following steps: determine the adaptive learning rates of different network layers according to the baseline learning rate and each network layer in the graph neural network; determine the initial network weights of the graph neural network according to a preset weight distribution mode and the parameters of the preset weight distribution mode, wherein the preset weight distribution mode includes at least: normal distribution or uniform distribution, and the parameters of the preset weight distribution mode include: mean and standard deviation; determine the target activation function of each network layer according to the preset activation function and the adaptive adjustment parameters of each network layer, wherein the adaptive adjustment parameters are associated with the statistical characteristics of the data in the data set.

[0267] Optionally, the first determination module 34 is also used to perform the following steps: standardize the first feature vector set according to a preset feature vector conversion function to obtain a target vector set, wherein the target vector set includes: an execution object vector and an enterprise task vector; determine a scoring function according to a similarity coefficient between the execution object vector and the enterprise task vector; determine the execution object corresponding to each enterprise task according to the scoring function to obtain a first initial allocation plan; process the first initial allocation plan according to preset constraints to obtain a first allocation plan, wherein the preset constraints include at least: there is no conflict in the time when the same execution object processes different enterprise tasks.

[0268] Optionally, the enterprise task allocation device also includes a display module, which is used to perform the following steps after determining the second allocation plan for enterprise tasks based on the second feature vector set: using a preset scheduling algorithm to convert the second allocation plan into a task schedule, wherein the task schedule includes: multiple enterprise tasks, the execution object assigned to each enterprise task, and the start time and end time of each enterprise task; using a graph neural network to adjust the start time and end time of each enterprise task until the preset conditions are met to obtain a target task schedule, wherein the preset conditions at least include: the total completion time of all enterprise tasks is the shortest; uploading the target task schedule to the enterprise management system; converting the task schedule into a target graphic, and displaying the target graphic on a graphic display interface, wherein the target graphic includes at least one of the following: an enterprise task processing flow chart, an enterprise task processing timetable, and an enterprise task allocation chart.

[0269] It should be noted that the above Figure 3 The modules in the embodiment may be program modules (e.g., a set of program instructions for implementing a specific function) or hardware modules. For the latter, they may be expressed in the following forms, but are not limited thereto: the expression form of the above modules is a processor, or the functions of the above modules are implemented by a processor.

[0270] It should be noted that Figure 3 The preferred implementation of the illustrated embodiment can be found in Figure 1 The relevant description of the illustrated embodiment will not be repeated here.

[0271] Figure 4 The hardware structure block diagram of a computer terminal for implementing the method for allocating enterprise tasks is shown in FIG. Figure 4As shown, the computer terminal 40 may include one or more (402a, 402b, ..., 402n are used to illustrate) processors 402 (the processor 402 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 404 for storing data, and a transmission module 406 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 4 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 4 More or fewer components as shown, or with Figure 4 Different configurations are shown.

[0272] It should be noted that the one or more processors 402 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 40. As involved in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0273] The memory 404 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the method for allocating enterprise tasks in the embodiment of the present application. The processor 402 executes various functional applications and data processing by running the software programs and modules stored in the memory 404, that is, the method for allocating enterprise tasks described above is realized. The memory 404 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 404 may further include a memory remotely arranged relative to the processor 402, and these remote memories may be connected to the computer terminal 40 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0274] The transmission module 406 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 40. In one example, the transmission module 406 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 406 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0275] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 40 .

[0276] It should be noted that, in some optional embodiments, the above Figure 4 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 4 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0277] It should be noted that Figure 4 The computer terminal shown is used to execute Figure 1 The method of allocating enterprise tasks shown in the figure, therefore the relevant explanations and instructions in the method of executing the above commands are also applicable to the electronic device and will not be repeated here.

[0278] The embodiment of the present application further provides a non-volatile storage medium, the non-volatile storage medium includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above-mentioned method for allocating enterprise tasks.

[0279] A program for a non-volatile storage medium to perform the following functions: obtaining multiple data related to enterprise tasks from an enterprise management system to obtain a data set; generating a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent enterprise tasks and execution objects of enterprise tasks, and the edges are used to represent correlation indicators between execution objects and enterprise tasks, priority differences between enterprise tasks, and correlation indicators between enterprise tasks; using a graph neural network, iteratively updating feature vectors corresponding to nodes to obtain a first feature vector set output by the graph neural network; determining a first allocation plan for enterprise tasks based on the first feature vector set; when any data in the data set changes, updating the task relationship graph to obtain a target task relationship graph, using a graph neural network to iteratively update feature vectors corresponding to target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; determining a second allocation plan for enterprise tasks based on the second feature vector set.

[0280] An embodiment of the present application also provides an electronic device, including: a memory and a processor, the processor being used to run a program stored in the memory, wherein the above enterprise task allocation method is executed when the program is run.

[0281] The processor is used to run a program that performs the following functions: obtain multiple data related to enterprise tasks from an enterprise management system to obtain a data set; generate a task relationship graph based on the data set, wherein the task relationship graph includes multiple nodes and edges, the nodes are used to represent enterprise tasks and execution objects of enterprise tasks, and the edges are used to represent correlation indicators between execution objects and enterprise tasks, priority differences between enterprise tasks, and correlation indicators between enterprise tasks; use a graph neural network to iteratively update feature vectors corresponding to nodes to obtain a first feature vector set output by the graph neural network; determine a first allocation plan for enterprise tasks based on the first feature vector set; when any data in the data set changes, update the task relationship graph to obtain a target task relationship graph, use the graph neural network to iteratively update feature vectors corresponding to target nodes in the target task relationship graph to obtain a second feature vector set output by the graph neural network; determine a second allocation plan for enterprise tasks based on the second feature vector set.

[0282] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0283] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0284] In the above-mentioned embodiments of the present application, the collected information is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary protection measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0285] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0286] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0287] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0288] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0289] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for allocating enterprise tasks, characterized in that: include: Acquire multiple data related to enterprise tasks from the enterprise management system to obtain a data set; Generate a task relationship graph according to the data set, wherein the task relationship graph includes a plurality of nodes and edges, the nodes are used to represent enterprise tasks and execution objects of the enterprise tasks, and the edges are used to represent the correlation index between the execution object and the enterprise tasks, the priority difference between the enterprise tasks, and the correlation index between the enterprise tasks; Using a graph neural network, iteratively updating the feature vectors corresponding to the nodes to obtain a first feature vector set output by the graph neural network; Determine a first allocation scheme for the enterprise tasks according to the first feature vector set; When any data in the data set changes, the task relationship graph is updated to obtain a target task relationship graph, and the feature vector corresponding to the target node in the target task relationship graph is iteratively updated using the graph neural network to obtain a second feature vector set output by the graph neural network; A second allocation scheme for the enterprise tasks is determined based on the second feature vector set.

2. The method according to claim 1, characterized in that Obtain multiple data related to enterprise tasks from the enterprise management system to obtain a data set, including: Determine the data collection protocol based on the data types of different data related to the enterprise tasks and the corresponding validation rules for different data types; extracting enterprise demand information in the enterprise management system according to the data collection protocol, wherein the enterprise demand information includes at least one of the following: demand description information, start time, end time, business field, and demand priority information; and Extracting enterprise scale information related to the enterprise demand information, wherein the enterprise scale information includes at least one of the following: the number of employees of the enterprise, the annual revenue of the enterprise, and the market share of the enterprise; and, Extracting execution object information of the execution object, wherein the execution object is used to process enterprise tasks, and the execution object information includes at least one of the following: educational background, number of historical processing cases, and execution field; Determine the agreed time according to the start time and the end time; The enterprise demand information, the enterprise scale information, the execution object information and the agreed time are formatted and standardized to obtain the data set.

3. The method according to claim 2, characterized in that According to the data set, a task relationship diagram is generated, including: Extracting task information corresponding to each of the enterprise tasks from the data set, and encoding the task information into a first feature vector, wherein the task information includes at least one of the following: enterprise demand, task time, and task priority; Extracting execution object information corresponding to each execution object from the data set, and encoding the execution object information into a second feature vector; Determine a first edge weight corresponding to a first edge according to the execution field and the enterprise demand, wherein the first edge is an edge formed by the execution object and the enterprise task, and the first edge weight includes: the correlation index between the execution object and the enterprise task; Determining a second edge weight corresponding to the second edge according to a priority difference evaluation function and a correlation index evaluation function, wherein the second edge is an edge formed between the enterprise tasks, and the second edge weight includes: a priority difference and a correlation index between the enterprise tasks; The task relationship graph is generated according to the first feature vector, the second feature vector, the first edge weight, and the second edge weight.

4. The method according to claim 3, characterized in that When any data in the data set changes, updating the task relationship diagram includes: Determine whether the data set has changed according to a preset time interval, and if any data in the data set has changed, update the task relationship diagram by the following method: In the case where the task information changes, the changed task information is encoded into a third feature vector, and a first target feature vector corresponding to the task information is determined according to the first feature vector corresponding to the task information before the change and the third feature vector; In the case where the execution object information changes, the changed execution object information is encoded into a fourth feature vector, and a second target feature vector corresponding to the execution object information is determined according to the second feature vector corresponding to the execution object information before the change and the fourth feature vector; When the execution domain and / or the enterprise demand changes, updating the first edge weight corresponding to the first edge; The task relationship graph is updated according to the first target feature vector, the second target feature vector, the updated first edge weight, and the second edge weight.

5. The method according to claim 1, characterized in that Using the graph neural network, iteratively updating the feature vector corresponding to the target node in the target task relationship graph, to obtain a second feature vector set output by the graph neural network, including: Step S1, configuring the initialization parameters of the graph neural network, wherein the initialization parameters include: adaptive learning rate and network hierarchy parameters; Step S2, the configured graph neural network iteratively updates the feature vector corresponding to the target node in the task relationship graph at time t, the adaptive learning rate at time t, the neighbor node set of the target node, the edge weight of the edge where the target node is located, and the feature vector corresponding to the target node at time t+1, wherein t is a positive number and the target node is any node in the task relationship graph; Step S3, after each iteration, adjusting the adaptive learning rate according to a preset learning rate adjustment factor and a network training error rate of each iteration; Step S4, repeat step S2 and step S3 until the maximum number of iterations is reached or the adaptive learning rate reaches a preset convergence condition, and obtain the first feature vector set output by the graph neural network.

6. The method according to claim 5, characterized in that The graph neural network includes: multiple network layers; Configuring the initialization parameters of the graph neural network includes: Determining adaptive learning rates for different network layers according to a baseline learning rate and each network layer in the graph neural network; Determine the initial network weight of the graph neural network according to a preset weight distribution mode and parameters of the preset weight distribution mode, wherein the preset weight distribution mode at least includes: normal distribution or uniform distribution, and the parameters of the preset weight distribution mode include: mean and standard deviation; According to a preset activation function and an adaptive adjustment parameter of each of the network layers, a target activation function of each of the network layers is determined, wherein the adaptive adjustment parameter is associated with a statistical characteristic of the data in the data set.

7. The method according to claim 1, characterized in that Determining a first allocation scheme for the enterprise task according to the first feature vector set includes: According to a preset feature vector conversion function, the first feature vector set is standardized to obtain a target vector set, wherein the target vector set includes: an execution object vector and an enterprise task vector; Determining a scoring function according to a similarity coefficient between the execution object vector and the enterprise task vector; According to the scoring function, determining the execution object corresponding to each enterprise task, and obtaining a first initial allocation plan; The first initial allocation plan is processed according to preset constraints to obtain the first allocation plan, wherein the preset constraints at least include: there is no conflict in the time when the same execution object processes different enterprise tasks.

8. The method according to claim 1, characterized in that After determining a second allocation scheme for the enterprise tasks according to the second feature vector set, the method further includes: Using a preset scheduling algorithm, converting the second allocation plan into a task schedule, wherein the task schedule includes: a plurality of the enterprise tasks, an execution object assigned to each of the enterprise tasks, and a start time and an end time of each of the enterprise tasks; Using the graph neural network, the start time and end time of each of the enterprise tasks are adjusted until the preset conditions are met to obtain a target task schedule, wherein the preset conditions at least include: the total completion time of all the enterprise tasks is the shortest; Uploading the target task schedule to the enterprise management system; The task schedule is converted into a target graph, and the target graph is displayed on a graphic display interface, wherein the target graph includes at least one of the following: an enterprise task processing flow chart, an enterprise task processing timetable, and an enterprise task allocation chart.

9. A device for allocating enterprise tasks, characterized in that: include: An acquisition module is used to acquire multiple data related to enterprise tasks from the enterprise management system to obtain a data set; A generating module, configured to generate a task relationship graph according to the data set, wherein the task relationship graph includes a plurality of nodes and edges, wherein the nodes are used to represent enterprise tasks and execution objects of the enterprise tasks, and the edges are used to represent correlation indicators between the execution objects and the enterprise tasks, priority differences between the enterprise tasks, and correlation indicators between the enterprise tasks; An iteration module, used to iteratively update the feature vectors corresponding to the nodes using a graph neural network to obtain a first feature vector set output by the graph neural network; A first determining module, configured to determine a first allocation scheme for the enterprise task according to the first feature vector set; An updating module, configured to update the task relationship graph when any data in the data set changes, obtain a target task relationship graph, and iteratively update a feature vector corresponding to a target node in the target task relationship graph using the graph neural network to obtain a second feature vector set output by the graph neural network; The second determination module is used to determine a second allocation plan for the enterprise task according to the second feature vector set.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the enterprise task allocation method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program, when running, executes the enterprise task allocation method described in any one of claims 1 to 8.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for allocating enterprise tasks as described in any one of claims 1 to 8 is implemented.

Images