Job task allocation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a job task allocation method and device, electronic equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the steps of predicting a job task scheduling strategy and an execution probability of the job task scheduling strategy based on a first object state feature through a preset original task scheduling model, and updating the first object state feature to obtain a second object state feature; and calculating a reward score of the job task scheduling strategy, and performing state evaluation on the first object state feature and the second object state feature through a state evaluation model to obtain a first state score and a second state score. And calculating a target loss according to the execution probability, the reward score, the first state score and the second state score, performing parameter updating on the original task scheduling model according to the target loss, and predicting a target job task scheduling strategy based on a preset target object state feature through the target task scheduling model. According to the embodiment of the invention, the job task allocation efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and device for allocating job tasks, an electronic device, and a storage medium. Background Art

[0002] In related technologies, manual experience is used to assign pending tasks to task execution objects according to the task content requirements and the professional requirements of the execution objects. This process relies heavily on manual experience and judgment, which consumes a lot of time and cost. In addition, this task allocation method is difficult to adapt to the rapid changes in task objectives and the dynamic changes in the status of the execution objects. Although some digital task management software already exists, its functions are mostly limited to the recording, distribution and process follow-up of task information. The allocation of task tasks is still highly dependent on manual experience and post-intervention, which seriously affects the efficiency of task allocation. Therefore, how to improve the efficiency of task allocation has become a technical problem that needs to be solved urgently. Summary of the Invention

[0003] The main purpose of the embodiments of the present application is to provide a method and device for allocating job tasks, an electronic device, and a storage medium, aiming to improve the efficiency of job task allocation.

[0004] To achieve the above objectives, a first aspect of an embodiment of the present application provides a method for allocating job tasks, the method comprising:

[0005] Obtaining a first object state feature;

[0006] Predicting a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature through a preset original task scheduling model;

[0007] Simulate and update the first object state feature according to the job task scheduling strategy to obtain a second object state feature;

[0008] Calculating a reward score for the job task scheduling strategy;

[0009] Performing a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and performing a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score;

[0010] Calculating a target loss according to the execution probability, the reward score, the first state score, and the second state score, and updating parameters of the original task scheduling model according to the target loss to obtain a target task scheduling model;

[0011] The target task scheduling model is used to predict the target operation task scheduling strategy based on the preset target object state characteristics.

[0012] In some embodiments, the job task scheduling strategy includes a job processing task and a task execution object of the job processing task, and calculating the reward score of the job task scheduling strategy includes:

[0013] Obtaining an execution efficiency score of the task execution object in executing the job processing task;

[0014] Obtaining an original object score of the task execution object;

[0015] The task type of the job processing task is obtained, and a reward evaluation is performed according to the task type, the execution efficiency score, and the original object score to obtain the reward score.

[0016] In some embodiments, the task type includes a collaboration type, and performing a reward evaluation based on the task type, the execution efficiency score, and the original object score to obtain the reward score includes:

[0017] If the task type of the job processing task is the collaborative type, the task execution object that executes the job processing task is used as a collaborative execution object;

[0018] Obtaining an object collaboration score of the collaboration execution object;

[0019] The reward score is obtained by performing a weighted calculation on the execution efficiency score, the original object score, and the object collaboration score.

[0020] In some embodiments, performing weighted calculation on the execution efficiency score, the original object score, and the object collaboration score to obtain the reward score includes:

[0021] Get the scoring rule text;

[0022] Performing text analysis on the scoring rule text to obtain scoring indicator weights; the scoring indicator weights include an efficiency weight parameter, an object weight parameter, and a collaboration weight parameter;

[0023] Multiplying the efficiency weight parameter and the execution efficiency score to obtain a first score;

[0024] Multiplying the object weight parameter and the original object score to obtain a second score;

[0025] Multiplying the collaboration weight parameter and the object collaboration score to obtain a third score;

[0026] The first score, the second score, and the third score are summed to obtain the reward score.

[0027] In some embodiments, predicting a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature using a preset original task scheduling model includes:

[0028] Predicting an original probability of a preset task scheduling strategy based on the first object state feature by the original task scheduling model;

[0029] Selecting the preset task scheduling strategy with the largest original probability as the job task scheduling strategy;

[0030] The original probability of the job task scheduling strategy is used as the execution probability.

[0031] In some embodiments, obtaining the first object state feature includes:

[0032] Initialize the mapping list between preset tasks and preset execution objects;

[0033] Performing a list value evaluation on the mapping list using a preset multi-dimensional value alignment model to obtain a matching score between the preset task and the preset execution object;

[0034] Obtaining an object performance score of the preset execution object;

[0035] The first object state feature is determined based on the matching score and the object performance score.

[0036] In some embodiments, the reward score includes an execution efficiency score, an original object score, and an object collaboration score. After calculating the reward score of the job task scheduling strategy, the following steps are performed:

[0037] Obtaining the number of state exceptions triggered when simulating an update of the state feature of the first object;

[0038] determining a first score increment according to the number of abnormal states;

[0039] Get the object performance score of the preset execution object;

[0040] Performing weighted calculation on the execution efficiency score, the original object score, and the object collaboration score to obtain a second score increment;

[0041] The object performance score is updated according to the first score increment and the second score increment.

[0042] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a job task allocation device, the device comprising:

[0043] A state feature acquisition module, configured to acquire a state feature of a first object;

[0044] A job prediction module, configured to predict a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature using a preset original task scheduling model;

[0045] a job simulation module, configured to simulate and update the first object state feature according to the job task scheduling strategy to obtain a second object state feature;

[0046] A score calculation module, used to calculate the reward score of the job task scheduling strategy;

[0047] a state evaluation module, configured to perform a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and to perform a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score;

[0048] a model optimization module, configured to calculate a target loss based on the execution probability, the reward score, the first state score, and the second state score, and update parameters of the original task scheduling model based on the target loss to obtain a target task scheduling model;

[0049] The task scheduling output module is used to predict the target job task scheduling strategy based on the preset target object state characteristics through the target task scheduling model.

[0050] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in the first aspect when executing the computer program.

[0051] To achieve the above-mentioned purpose, the fourth aspect of the embodiments of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the method described in the first aspect.

[0052] The job task allocation method and device, electronic device, and storage medium proposed in this application use the original task scheduling model to perform a strategic prediction of the current object state, and introduce execution probability, reward score, and state score to construct a target loss function, thereby being able to measure the pros and cons of different task scheduling strategies under the expected execution effect, and based on this, optimize the task scheduling model so that the final generated scheduling strategy not only has a high task completion efficiency, but also maintains strong adaptability when the task goal changes or the object state changes dynamically. Through this method, the job task allocation process no longer relies on subjective judgment based on human experience, but instead the strategy model automatically generates the optimal scheduling strategy based on the current state of the object and continuously optimizes it adaptively. This effectively solves the problems of task allocation relying on human experience, untimely response, and poor adaptability in the existing technology, and fundamentally improves the intelligence level and execution efficiency of job task allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flowchart of the method for allocating job tasks provided by an embodiment of the present application;

[0054] Figure 2 yes Figure 1 Flowchart of step S101 in FIG.

[0055] Figure 3 yes Figure 1 Flowchart of step S102 in FIG.

[0056] Figure 4 yes Figure 1 Flowchart of step S403 in FIG.

[0057] Figure 5 yes Figure 4 Flowchart of step S304 in FIG.

[0058] Figure 6 is another flow chart of the method for allocating job tasks provided by an embodiment of the present application;

[0059] Figure 7 This is a schematic diagram of the structure of the job task allocation device provided in an embodiment of the present application;

[0060] Figure 8 This is a schematic diagram of a management platform structure for task allocation provided by an embodiment of the present application;

[0061] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0063] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0065] First, let’s analyze some of the terms used in this application:

[0066] Artificial intelligence (AI) is a new technical discipline that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. A branch of computer science, AI seeks to understand the essence of intelligence and create new intelligent machines that can respond in a manner similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thinking. It also encompasses theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0067] Natural language processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese and English). A branch of artificial intelligence, NLP is an interdisciplinary field between computer science and linguistics, often referred to as computational linguistics. Natural language processing encompasses grammatical analysis, semantic analysis, and discourse comprehension. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It encompasses data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research related to language processing, and linguistics research related to language computing.

[0068] An agent is an agent that can perceive its environment and take actions to achieve specific goals. It can be software, hardware, or a system, and possesses autonomy, adaptability, and interaction. An agent perceives changes in its environment, makes judgments and decisions based on learned knowledge and algorithms, and then executes actions to influence the environment or achieve a predetermined goal.

[0069] In related technologies, manual experience is used to assign pending tasks to task execution objects according to the task content requirements and the professional requirements of the execution objects. This process relies heavily on manual experience and judgment, which consumes a lot of time and cost. In addition, this task allocation method is difficult to adapt to the rapid changes in task objectives and the dynamic changes in the status of the execution objects. Although some digital task management software already exists, its functions are mostly limited to the recording, distribution and process follow-up of task information. The allocation of task tasks is still highly dependent on manual experience and post-intervention, which seriously affects the efficiency of task allocation. Therefore, how to improve the efficiency of task allocation has become a technical problem that needs to be solved urgently.

[0070] Based on this, embodiments of the present application provide a method and apparatus for allocating job tasks, an electronic device, and a storage medium, aiming to improve the efficiency of allocating job tasks.

[0071] The job task allocation method and device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the job task allocation method in the embodiments of the present application is described.

[0072] The job task allocation method provided in the embodiment of the present application relates to the field of artificial intelligence technology. The job task allocation method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or can be configured as a server cluster or a distributed system composed of multiple physical servers, or can be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the job task allocation method, etc., but is not limited to the above forms.

[0073] Figure 1 This is an optional flowchart of the method for allocating job tasks provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S107.

[0074] Step S101: Acquire a first object state feature.

[0075] Step S102 : predicting a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature using a preset original task scheduling model.

[0076] Step S103: simulate and update the first object state feature according to the job task scheduling strategy to obtain the second object state feature.

[0077] Step S104: Calculate the reward score of the job task scheduling strategy.

[0078] Step S105 , performing a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and performing a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score.

[0079] Step S106 , calculating the target loss according to the execution probability, the reward score, the first state score, and the second state score, and updating the parameters of the original task scheduling model according to the target loss to obtain the target task scheduling model.

[0080] Step S107 : predicting the target job task scheduling strategy based on the preset target object state characteristics through the target task scheduling model.

[0081] In the steps S101 to S107 shown in the embodiment of the present application, the original task scheduling model is used to make a strategic prediction of the current object state, and the execution probability, reward score and state score are introduced to construct a target loss function, so that the pros and cons of different task scheduling strategies under the expected execution effect can be measured, and the task scheduling model is optimized based on this, so that the scheduling strategy finally generated not only has a high task completion efficiency, but also maintains strong adaptability when the task goal changes or the object state changes dynamically. Through this method, the allocation process of the job task no longer relies on subjective judgment based on manual experience, but the strategy model automatically generates the optimal scheduling strategy based on the current state of the object and continuously optimizes it adaptively. This effectively solves the problems of task allocation relying on manual experience, untimely response, and poor adaptability in the prior art, and fundamentally improves the intelligence level and execution efficiency of the job task allocation.

[0082] It should be noted in advance that, at the beginning of task assignment, the embodiment of the present application will construct a corresponding agent for each job execution object waiting to be assigned a task, that is, in the present embodiment, each job execution object to be assigned is an agent. However, in some task assignment related scenarios, often only the overall task goal of a group level is mastered, and the specific task content that each execution object needs to undertake is not clear. For example, in scenarios such as production line scheduling, collaborative maintenance, and cross-team collaboration, task goals are usually proposed in the form of overall goals and have not yet been broken down into specific, operational tasks. Therefore, in the task assignment startup phase, it is necessary to first understand the task intent of the overall task goal, and then split the overall task goal into multiple subtasks, and these subtasks can be used to constitute the action set of the agent.

[0083] The task assignment method provided in the embodiment of the present application can be applied to a variety of application scenarios. For example, for a project, it is necessary to distribute the project to different employees in different departments. In this embodiment, each independent agent corresponds to an employee. In other cross-organizational collaboration implementation scenarios, for example, a complex project needs to be completed by multiple companies in the same industry, and each participating company can be mapped to an agent.

[0084] In step S101 of some embodiments, the first object state feature refers to the feature vector of information such as schedulability, capability status, historical execution records, resource occupancy, priority, etc. of the job execution object before the current task is assigned. These state features constitute the input basis for the subsequent original task scheduling model to perform strategy prediction.

[0085] In addition to the aforementioned feature vector, the first object state feature of this embodiment also includes a matching score and an object performance score. Figure 2 In some embodiments, step S101 may include but is not limited to steps S201 to S204:

[0086] Step S201: Initialize a mapping list between preset tasks and preset execution objects.

[0087] Step S202 : performing list value evaluation on the mapping list through a preset multi-dimensional value alignment model to obtain a matching score between the preset task and the preset execution object.

[0088] Step S203: Obtain the object performance score of the preset execution object.

[0089] Step S204: determining the first object state feature according to the matching score and the object performance score.

[0090] In step S201 of some embodiments, the preset task refers to a set of job tasks to be assigned within the current scheduling cycle. In this embodiment, the preset task can be a set of subtasks obtained by decomposing the task based on the overall goal through a large language model. The preset execution object refers to all objects that may accept task scheduling. The mapping list between preset tasks and preset execution objects is used to enumerate the matching combinations between each preset task and each preset execution object. It should be noted that in some embodiments, each preset task is configured with task attribute information, and the preset task can only be assigned to a preset object with the same attributes, and the job nature of the preset object can be used as the basis for judgment. For example, the task of making a PPT is assigned to the preset object of the planning position, and the task of purchasing materials is assigned to the preset object of the purchasing position.

[0091] In step S202 of some embodiments, the multidimensional value alignment model is an evaluation mechanism proposed in this embodiment. Its construction goal is to comprehensively score the degree of fit between tasks and objects based on multiple value dimensions (such as efficiency, fairness, transparency, compliance, and social responsibility). The match score is the numerical value output by the multidimensional value alignment model that measures the degree of fit between a specific execution object and a specific task, indicating whether the object is well aligned with the current task requirements in terms of value objectives.

[0092] Specifically, in this embodiment, the value dimensions to be evaluated include efficiency, fairness, transparency, compliance, and social responsibility. Efficiency includes task completion efficiency and GPU resource utilization. Fairness refers to the balanced distribution of tasks, avoiding bias toward specific pre-defined execution targets. Transparency refers to the explainability of the decision-making process. Compliance refers to compliance with laws and regulations. Social responsibility refers to avoiding discriminatory behavior or negative social impacts.

[0093] After the historical behavior data of the preset execution object (i.e., the behavioral records of the task execution in history) and the specific task are input into the multi-dimensional value alignment model, the model will obtain the score of the preset execution object corresponding to each value dimension. Each value dimension has a corresponding weight parameter. The scores of all value dimensions are weighted and summed to obtain the matching score between the preset execution object and a specific task. For details, please refer to the following analytical formula:

[0094]

[0095] Among them, V value represents the matching score, i represents the sequence of value dimensions, n represents the total number of value dimensions, w i Represents the weight data corresponding to the current value dimension, v i Indicates the score corresponding to the current value dimension.

[0096] In step S203 of some embodiments, the object performance score is a comprehensive score used to quantify the capability level, stability, execution speed, and task completion quality of a particular execution object in historical task execution. The object performance scores of all pre-set execution objects have the same preset value at the beginning of task assignment. The object performance scores are updated after subsequent execution of the task assignment strategy. The preset value may be 100, but is not limited thereto.

[0097] In step S204 of some embodiments, the matching score and the object performance score are fused as extended features with the original state information of the preset execution object to ultimately form the first object state feature. In some embodiments, the matching score and the object performance score are weighted and fused to obtain a comprehensive score, which is used as a dimension of information to form the first object state feature. For example, the matching score and the object performance score can be weighted and calculated according to the following analytical formula:

[0098] S total =θ·V value +(1-θ)credit normalized (2),

[0099] Among them, S total Indicates the comprehensive score. value Indicates the matching score. normalized represents the object performance score. θ represents the preset weight data. In this embodiment, θ may be 0.7, but is not limited thereto.

[0100] In steps S201 to S204 shown in the embodiment of the present application, a mapping list of preset tasks and preset execution objects is constructed before task assignment, and the mapping list is evaluated for list value based on a multi-dimensional value alignment model to obtain a matching score. At the same time, the first object state feature is finally determined in combination with the obtained object performance score, so that the generated first object state feature can more comprehensively and meticulously reflect the scheduling feasibility and task adaptability of the execution object. By introducing the matching score and the object performance score, the first object state feature not only contains basic state information, but also integrates the dynamic adaptation relationship between task requirements and object capabilities and the object's own execution capability performance, thereby providing more accurate and rich input information for the subsequent strategy prediction of the original task scheduling model. Compared with the traditional method of task assignment that relies solely on static state data, this embodiment can improve the pertinence and accuracy of the task assignment strategy through a multi-dimensional, multi-indicator feature construction method before scheduling model training and reasoning, thereby effectively improving the rationality and execution efficiency of the overall job task assignment.

[0101] In step S102 of some embodiments, the original task scheduling model refers to a task scheduling strategy prediction model that has not yet been optimized, whose input is the first object state feature and whose output is the original probability distribution corresponding to each preset task scheduling strategy. The preset task scheduling strategy refers to a set of optional task allocation schemes determined for all preset execution objects, specifically reflecting the correspondence between the preset execution objects and the ultimately assigned tasks.

[0102] See also Figure 3 In some embodiments, step S102 may include but is not limited to steps S301 to S303:

[0103] Step S301 : predicting the original probability of a preset task scheduling strategy based on the first object state feature through the original task scheduling model.

[0104] Step S302 : Selecting the preset task scheduling strategy with the highest original probability as the job task scheduling strategy.

[0105] Step S303: Using the original probability of the job task scheduling strategy as the execution probability.

[0106] In step S301 of some embodiments, the raw probability represents the likelihood of executing each preset task scheduling strategy under the current object state. This probability is typically calculated via the output layer of the strategy network, for example by using a Softmax function to normalize each strategy score into a probability distribution. Each preset task scheduling strategy includes an assignment relationship between each preset practice object and each preset task.

[0107] In step S302 of some embodiments, the job task scheduling strategy refers to the specific strategy actually selected to guide the current job task allocation action, representing the task allocation plan that the original task scheduling model considers to be optimal or most likely to be executed under the current state characteristics.

[0108] In step S303 of some embodiments, the numerical range of the execution probability is generally between 0 and 1. The execution probability can be used in weighted processing or scheduling strategy modification in subsequent reward score calculation and target loss definition.

[0109] In steps S301 to S303 shown in the embodiment of the present application, the original probability of the preset task scheduling strategy is predicted based on the state characteristics of the first object through the original task scheduling model, and a probabilistic evaluation can be performed based on the current object state among multiple feasible preset task scheduling strategies to quantify the executability and preference of each strategy. Furthermore, by selecting the preset task scheduling strategy with the largest original probability as the job task scheduling strategy, the task allocation decision can be selected based on the principle of maximum probability, thereby improving the rationality and stability of the strategy selection. At the same time, the original probability of the job task scheduling strategy is used as the execution probability. Through steps S301 to S303, while maintaining the diversity and dynamic adaptability of task scheduling, it is possible to ensure that each task allocation decision process has a clear selection basis and quantitative support, significantly improving the intelligence level and execution efficiency of the job task allocation process.

[0110] In step S103 of some embodiments, simulation update means predicting the object state changes after the execution of the strategy based on the current task scheduling strategy without actually changing the real environment state, such as an increase in task load, a change in resource usage, etc., and mapping it into a new object state feature, which is recorded as the second object state feature.

[0111] In step S104 of some embodiments, the job task scheduling policy includes a job processing task and a task execution object of the job processing task. The job processing task refers to a preset task to be executed in the currently executed job task scheduling policy. The task execution object refers to a preset execution object that executes the job processing task.

[0112] In this embodiment, the reward score refers to an indicator value used to measure the performance of the current scheduling strategy in target dimensions such as efficiency, resource utilization, multiple value dimensions, and collaborative competitive actions.

[0113] See also Figure 4 In some embodiments, step S104 may include but is not limited to steps S401 to S403:

[0114] Step S401: Obtain the execution efficiency score of the task execution object in executing the job processing task.

[0115] Step S402: Obtain the original object score of the task execution object.

[0116] Step S403 , obtaining the task type of the job processing task, and performing a reward evaluation based on the task type, the execution efficiency score, and the original object score to obtain a reward score.

[0117] In step S401 of some embodiments, the execution efficiency score refers to a score value that characterizes the actual execution performance of the task execution object when completing the specified job processing task. The method of obtaining the execution efficiency score may include: collecting the object's original execution data such as the job completion time, resource occupancy ratio (such as CPU utilization, memory consumption), energy consumption index or task response time in real time during the execution of the job processing task, and performing quantitative calculations based on these data according to preset efficiency evaluation rules (such as weighted average, normalization processing), thereby generating a corresponding score.

[0118] In step S402 of some embodiments, the original object score refers to a comprehensive alignment score predicted by another multi-dimensional value alignment model based on the action of the task execution object when performing the current job processing task. Different from the multi-dimensional value alignment model in step S202, the input of the multi-dimensional value alignment model in this step embodiment is the behavior of the task execution object when performing the current job processing task, but it is also a mechanism for comprehensively evaluating the task execution object based on multiple value dimensions such as efficiency, fairness, transparency, compliance, and social responsibility. The model evaluates the performance of each object under each value dimension and performs weighted fusion to output the final original object score. The specific calculation method of the original object score is the same as that of the analytical formula (1) and will not be repeated again.

[0119] In step S403 of some embodiments, the task type refers to the basic attributes used to characterize the interaction mode of the current job processing task, specifically divided into collaborative tasks and competitive tasks. Collaborative tasks refer to situations where multiple execution objects need to work together to complete the task, such as logistics distribution optimization and intelligent fleet scheduling. Competitive tasks refer to situations where multiple execution objects independently compete for the opportunity to complete the task, and the system determines the task ownership based on the performance of each object, such as advertising bidding and online transactions.

[0120] It should be noted that after the large language model is used to decompose the tasks based on the overall goal and obtain multiple preset tasks, the content of the preset tasks can be understood based on the large language model to determine whether the preset tasks are collaborative tasks or competitive tasks. For example, it can be determined whether the complexity of the preset tasks is a high-complexity task, and whether the preset tasks are long-term tasks or short-term tasks from the perspective of timeliness. In this embodiment, the task type of the preset tasks with high complexity and long duration will be confirmed as collaborative.

[0121] For details on how to calculate the bonus points, please refer to Figure 5 In some embodiments, step S403 may also include but is not limited to steps S501 to S503:

[0122] Step S501: If the task type of the job processing task is a collaborative type, the task execution object that executes the job processing task is used as a collaborative execution object.

[0123] Step S502: Obtain the object collaboration score of the collaboration execution object.

[0124] Step S503 : performing weighted calculation on the execution efficiency score, the original object score, and the object collaboration score to obtain a reward score.

[0125] In step S501 of some embodiments, the collaborative execution object refers to an execution individual that needs to cooperate with other objects to complete the job processing task in collaborative job processing, and its behavior and performance directly affect the overall completion effect and efficiency of the collaborative task.

[0126] In step S502 of some embodiments, the process of obtaining the object collaboration score includes two stages. First, after the collaborative task is completed, the total reward value of the collaborative task is calculated based on the overall performance of the task execution (such as overall completion, total resource utilization, overall execution efficiency, overall quality score, etc.). This total reward value is used to measure the overall collaborative achievement. Specifically, please refer to the following analytical formula:

[0127]

[0128] Among them, R global Indicates efficiency reward, reflecting the quality and efficiency of the overall task completion. Example: If the task is completed within the specified time, where t actual is the actual completion time, t max is the maximum allowed time. individual,i Represents the individual reward of collaborative execution object i, which can be measured by the number of completed task submodules or the amount of key resource input. i Represents the contribution weight, which is dynamically adjusted according to the role and task importance of the collaborative execution object.

[0129] Subsequently, the total reward value is distributed among the collaborative execution objects based on the Shapley Value allocation method to determine the contribution each object should receive to the overall collaborative outcome, thereby determining the object collaboration score for each collaborative execution object.

[0130] In step S503 of some embodiments, the present embodiment also responds to different events to adjust the weight parameters of different dimensions. Specifically, step S503 further includes the following steps:

[0131] First, obtain the scoring rule text. In this embodiment, the scoring rule text refers to text data used to express the basis for setting weights related to the task evaluation system, which is usually derived from industry standard guidelines, industry dynamic information, resource market trends, internal management specifications or other document materials that can affect the adjustment of task allocation strategies. It can also include but is not limited to the following types: First, various risk assessment reports, such as supply chain risk warnings, project execution risk warnings, etc. This type of information can drive the system to increase the weight of stability or reliability related dimensions in task scheduling; second, customer or user demand change documents, such as major project demand updates, customer-proposed service level requirement adjustment notifications, thereby dynamically adjusting the weights of efficiency or quality related indicators; third, social responsibility and sustainable development goal reports, such as those involving environmental protection requirements, social equity indicators or carbon emission limit information, which are used to enhance the weights of social responsibility indicators; fourth, internal operation strategy adjustment notifications, such as decision-making instructions within the organization to improve resource utilization efficiency and reduce operating costs for a specific stage, thereby affecting the efficiency and resource utilization weight setting. The embodiment of this application does not strictly limit the specific type of scoring rule text.

[0132] Subsequently, text analysis is performed on the scoring rule text to obtain scoring indicator weights. These scoring indicator weights include an efficiency weight parameter, an object weight parameter, and a collaboration weight parameter, corresponding to the importance ratios of the execution efficiency score, the original object score, and the object collaboration score, respectively. It is understood that the dimensions of the scoring indicator weights can be increased based on actual needs.

[0133] It should be noted that the efficiency weight parameter, object weight parameter and collaboration weight parameter are initially preset values. If a scoring rule text is received during the execution of task assignment, the scoring rule text is parsed to determine the weight parameters that need to be adjusted and their values ​​are modified. The weight parameters that need to be adjusted can be determined among the efficiency weight parameters, object weight parameters and collaboration weight parameters through natural language processing technology or a preset rule parsing module, and the adjustment range is made according to the scoring rule text to determine the final scoring indicator weight. Then multiply the efficiency weight parameter and the execution efficiency score to obtain the first score; multiply the object weight parameter and the original object score to obtain the second score; multiply the collaboration weight parameter and the object collaboration score to obtain the third score; sum the first score, the second score and the third score to obtain the bonus score. For details, please refer to the following analytical formula:

[0134] R total =λ1R e +λ2R v +λ3R c (4),

[0135] Among them, R totalrepresents the reward score, λ1 represents the efficiency weight parameter, R e represents the execution efficiency score, λ2 represents the object weight parameter, R v represents the original object score, λ3 represents the collaborative weight parameter, R c Represents the object collaboration score.

[0136] The calculation method of the execution efficiency score can refer to the following analytical formula:

[0137]

[0138] Among them, R e represents the execution efficiency score. α represents the time efficiency weight and can be set to 0.6. c Indicates the actual completion time of the task, t max represents the maximum completion time allowed for a task. β represents the resource efficiency weight. η represents the resource utilization rate.

[0139] If the task type of the job processing task is competitive type, then R total =λ1R e +λ2R v .

[0140] In order to achieve dynamic adjustment of scoring rules, this embodiment supports real-time push of updated scoring rule text through the RESTful API interface, so as to achieve updates based on events such as external market changes, regularly check the rule text update status every hour, or immediately trigger rule refresh when a specific event is detected (such as a change in task type, or the emergence of sensitive data processing requirements). To this end, a scoring logic parsing and execution module based on a rule engine (such as the open source tool Drools) can be integrated into the model, which can dynamically reconfigure weight parameters based on preset or real-time adjusted rule sets. For example, in cases where the task urgency is high, the rule engine can automatically increase the efficiency weight, so that the reward score calculation can highlight the importance of the speed of task completion. When it comes to sensitive data processing, the system can automatically increase the compliance-related weights in the object score to meet the requirements.

[0141] In steps S501 to S503 shown in the embodiment of the present application, by explicitly setting the task execution object as a collaborative execution object when the job processing task type is a collaborative type, and introducing the object collaboration score as a new evaluation dimension, the collaborative ability performance of the execution object can be comprehensively considered in the reward evaluation process, so that when generating the reward score, it not only relies on the traditional execution efficiency and value alignment, but also can perform special optimization for the multi-object coordination characteristics unique to collaborative tasks. Compared with the method of reward evaluation based only on single performance indicators, this embodiment introduces the object collaboration score and performs a weighted fusion calculation with the execution efficiency score and the original object score, so that the reward score can more truly and comprehensively reflect the actual contribution level of the execution object in the collaborative environment, effectively improving the scheduling model's judgment ability and execution effect in the collaborative task allocation and optimization process, and thus significantly improving the overall quality and efficiency of the completion of collaborative tasks.

[0142] Steps S401 to S403 shown in the embodiment of the present application, by respectively obtaining the execution efficiency score of the task execution object under the specific job processing task and the original object score of the object itself, and combining the task type of the job processing task for reward evaluation, can accurately measure the actual execution effect of the job task scheduling strategy on the basis of fully considering the differences between task characteristics and object capabilities. Compared with the traditional method that relies solely on execution efficiency or static object scoring, this embodiment introduces task type as an adjustment factor, and dynamically adjusts the scoring criteria according to the characteristics of different tasks, thereby effectively avoiding scoring distortion or evaluation bias problems, ensuring that the reward score can truly reflect the comprehensive matching degree between task execution effect and object capability, and further improving the scientificity and accuracy of scheduling strategy optimization in the job task allocation process.

[0143] See also Figure 6 In some embodiments, after step S104, it is necessary to update the object performance score of the task execution object. The task assignment method provided in the embodiment of the present application also includes but is not limited to steps S601 to S605:

[0144] Step S601: Obtain the number of state anomalies triggered when simulating an update of a first object state feature.

[0145] Step S602: Determine a first score increment according to the number of abnormal statuses.

[0146] Step S603: Obtain the object performance score of the preset execution object.

[0147] Step S604 : performing weighted calculation on the execution efficiency score, the original object score, and the object collaboration score to obtain a second score increment.

[0148] Step S605: updating the object performance score according to the first score increment and the second score increment.

[0149] In step S601 of some embodiments, the number of state anomalies refers to the number of abnormal records that do not meet the expected task completion conditions or resource usage standards during the simulation scheduling prediction process of the first object state characteristics, such as resource consumption exceeding the limit, task completion delay, failure to meet collaboration requirements, etc. Anomaly detection can use machine learning methods, such as isolation forest or one-class support vector machine (One-Class SVM) algorithms, which input behavioral characteristics such as task completion time, resource usage, and collaboration frequency during the object simulation execution process, output a probability score of whether it is an anomaly, and determine whether it is an anomaly based on a set threshold.

[0150] In step S602 of some embodiments, the first score increment refers to the numerical change in the object's reputation score that is reversely adjusted according to the number of anomalies. The more anomalies there are, the smaller the first score increment is, or even a negative value. The system can map the number of anomalies to a score adjustment value based on a preset reputation decay coefficient through linear decay, exponential decay, or a segmented penalty mechanism to punish objects that frequently exhibit abnormal behavior and strengthen the system's suppression of inefficiency and violations. For example, if a minor violation occurs (such as a timeout), the first score increment is -10, and if a serious violation occurs, the first score increment is -50. If V total If the score is >0.8 and there is no violation record, the first score increment is 10. If the task execution subject abandons the task, the first score increment is -20.

[0151] In step S603 of some embodiments, the initial value of the object performance score may be set to 100. The object performance score is not only used to reflect the ability to complete a single task, but also serves as a basis for prioritizing task assignments. Objects with high scores can be given priority in obtaining task assignment rights.

[0152] In step S604 of some embodiments, the weighted calculation process may refer to the following analytical formula:

[0153] Δcredit i =k·(R e +R v +R c ) (6),

[0154] Among them, Δcredit i Indicates the second score increment when the number of simulation updates is i. e Represents the execution efficiency score. v Represents the original object score. R c Represents the object collaboration score. k is a constant coefficient, which can be 10.

[0155] In step S605 of some embodiments, the updating process may refer to the following analytical formula:

[0156] The weighted calculation process can be referred to the following analytical formula:

[0157] credit i =credit i-1 +Δcredit i (7),

[0158] Among them, credit i Indicates the performance score of the object when the number of simulation updates is i. i-1 Δcredit represents the object performance score when the number of simulation updates is i-1. i Indicates the second score increment.

[0159] It is understandable that the updated object performance score can be used as the constituent information of the second object status feature. When the object performance score is lower than a certain threshold (such as 50), the corresponding task execution object is determined to be a violation object. A certain punishment mechanism is required for the violation object, for example: restricting the violation object from participating in high-value tasks. Alternatively, the violation object is prohibited from being assigned new tasks within a certain time window (such as 24 hours). Mark the object status as "restricted" in the task allocation database. Alternatively, suspend all operating permissions of the Agent until manual review is passed. For example, three consecutive violations or a single serious violation (such as a malicious attack). Restore after submitting an appeal and passing the compliance check.

[0160] When the object performance score is higher than a certain threshold (such as 90), the corresponding task execution object is determined to be a high-value object. High-value objects can be exchanged for additional computing resources, bandwidth, or higher execution permissions.

[0161] Steps S601 to S605 shown in the embodiment of the present application introduce a dynamic adjustment mechanism for object performance scores based on anomaly detection results and changes in comprehensive scores, so that the system can reflect the actual performance of each execution object in real time and accurately after the execution of the task, and provide timely rewards and punishments to enhance the fairness and rationality of task allocation.

[0162] In step S105 of some embodiments, the state evaluation model and the original task scheduling model form an Actor-Critic reinforcement learning framework. The state evaluation model scores the first object state feature and the second object state feature respectively to obtain a first state score and a second state score. The first state score represents the expected performance of the object under the first object state feature. The second state score represents the expected performance in the new state that the object may reach after executing the predicted scheduling strategy. By comparing the first state score and the second state score, the degree to which the scheduling strategy improves or worsens the object state can be quantified, which serves as an important reference for subsequent reward evaluation and target loss calculation.

[0163] In step S106 of some embodiments, the target loss may be calculated with reference to the following analytical formula:

[0164] L actor (θ)=-E s,a [logπ θ (a|s)×(R total +γV(s')-V(s))] (8),

[0165] Among them, L actor (θ) represents the target loss, π θ (a|s) represents the execution of the job task scheduling strategy under the first object state feature, and θ represents the model parameters of the original task scheduling model. total Represents the reward score. V(s) represents the first state score, V(s') represents the second state score, and γ represents the discount factor.

[0166] The policy gradient update formula based on the temporal difference (TD) error is used to update the parameter θ of the original task scheduling model until the first state score reaches a preset value (e.g., 0.7 or 0.8) or the TD error converges.

[0167] According to the final optimized parameter θ, the original task scheduling model is updated to the target task scheduling model.

[0168] In step S107 of some embodiments, the target object state feature may be the current state feature of any object to be scheduled, and the target task scheduling model may output a task allocation strategy for the object for subsequent execution.

[0169] See also Figure 7 The present application also provides a device for allocating tasks, which can implement the above-mentioned method for allocating tasks. The device includes:

[0170] The state feature acquisition module 701 is used to acquire the state feature of the first object.

[0171] The job prediction module 702 is configured to predict a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature using a preset original task scheduling model.

[0172] The job simulation module 703 is used to simulate and update the first object state feature according to the job task scheduling strategy to obtain the second object state feature.

[0173] The score calculation module 704 is used to calculate the reward score of the job task scheduling strategy.

[0174] The state evaluation module 705 is configured to perform a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and perform a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score.

[0175] The model optimization module 706 is used to calculate the target loss based on the execution probability, the reward score, the first state score and the second state score, and update the parameters of the original task scheduling model according to the target loss to obtain the target task scheduling model.

[0176] The task scheduling output module 707 is used to predict the target job task scheduling strategy based on the preset target object state characteristics through the target task scheduling model.

[0177] The specific implementation of the job task allocation device is basically the same as the specific embodiment of the above-mentioned job task allocation method, and will not be repeated here.

[0178] In addition to the above Figure 7 In addition to the device structure, the embodiment of the present application also provides a management platform structure, so that the task allocation method provided by the embodiment of the present application can be implemented on the management platform. Figure 8 , Figure 8 This is a schematic diagram of the structure of a management platform for task allocation provided in an embodiment of the present application.

[0179] The management platform provided by this application has an overall architecture that is constructed step by step from the data layer, rule layer, decision layer, execution and feedback layer, security and governance layer, and interaction layer. The modules of each layer work together to support the platform's intelligent task allocation and dynamic optimization functions. At the data layer, the platform collects and stores agent behavior data, task data, and detection data (used to store information related to external events), providing basic support for subsequent strategy generation and decision optimization. At the same time, through the data collection, storage, and processing modules, the integrity and real-time update of the data are guaranteed. At the rule layer, the platform has a value model that is used to define multi-dimensional value standards (such as efficiency, fairness, transparency, etc.) in different task scenarios, and dynamically adjusts the agent's reputation score and task allocation authority through an incentive and punishment mechanism to achieve positive incentives for agent behavior and suppress violations. At the decision layer, the platform uses the task decomposition and scheduling module to decompose complex overall tasks into multiple subtasks based on complexity and timeliness, namely preset tasks, and combines the agent's historical performance to dynamically determine the subtask allocation strategy through the collaboration and competition management module. At the same time, using the capability evolution module, it continuously tracks and improves the agent's comprehensive capability performance. At the execution and feedback layer, the platform is responsible for task execution and introduces a dynamic adjustment mechanism for task execution feedback. This mechanism dynamically adjusts weighting parameters by real-time monitoring of execution events (such as receiving scoring rule text), enhancing the system's rapid response and adaptability to external environmental changes. At the security and governance layer, the platform uses a behavior detection module to monitor agent behavior in real time, combined with a risk prevention and control module to predict potential risks. The external event detection module responds to changes in industry standards to ensure the stability and compliance of platform operations. At the interaction layer, the platform provides users with a dashboard module that visually displays various task execution and agent status indicators. Furthermore, a multi-dimensional visualization module dynamically displays multiple levels of indicators (such as task completion, resource utilization, and reputation changes), enabling users to intuitively understand the system's operational status and task execution. The layers of the overall architecture work together seamlessly, forming a complete closed loop from data support, rule constraints, intelligent decision-making, dynamic feedback, to visual interaction, enabling the platform to achieve intelligent scheduling and governance optimization for multiple agents.

[0180] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned method for allocating job tasks when executing the computer program. The electronic device can be any smart terminal including a tablet computer, an in-vehicle computer, or the like.

[0181] See also Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0182] The processor 901 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0183] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called by the processor 901 to execute the job task allocation method of the embodiments of this application.

[0184] Input / output interface 903, used to implement information input and output;

[0185] Communication interface 904, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0186] Bus 905 , which transmits information between various components of the device (e.g., processor 901 , memory 902 , input / output interface 903 , and communication interface 904 );

[0187] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0188] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned job task allocation method is implemented.

[0189] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor 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.

[0190] The job task allocation method, job task allocation device, electronic device and storage medium provided in the embodiments of the present application use the original task scheduling model to perform a strategic prediction on the current object state, and introduce the execution probability, reward score and state score to construct a target loss function, so as to measure the pros and cons of different task scheduling strategies under the expected execution effect, and optimize the task scheduling model based on this, so that the scheduling strategy finally generated not only has a high task completion efficiency, but also maintains strong adaptability when the task goal changes or the object state changes dynamically. Through this method, the allocation process of job tasks no longer relies on subjective judgment based on manual experience, but the strategy model automatically generates the optimal scheduling strategy based on the current state of the object, and continuously optimizes it adaptively. This effectively solves the problems of task allocation relying on manual experience, untimely response, and poor adaptability in the prior art, and fundamentally improves the intelligence level and execution efficiency of job task allocation.

[0191] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0192] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0193] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0194] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0195] The terms "first", "second", "third", "fourth", etc. (if any) in the specification 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 sequential order. 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 variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes 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.

[0196] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0197] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as 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 devices or units, which can be electrical, mechanical or other forms.

[0198] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0199] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0200] 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 prior art, 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 multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0201] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for allocating job tasks, characterized in that: The method comprises: Obtaining a first object state feature; Predicting a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature through a preset original task scheduling model; Simulate and update the first object state feature according to the job task scheduling strategy to obtain a second object state feature; Calculating a reward score for the job task scheduling strategy; Performing a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and performing a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score; Calculating a target loss according to the execution probability, the reward score, the first state score, and the second state score, and updating parameters of the original task scheduling model according to the target loss to obtain a target task scheduling model; The target task scheduling model is used to predict the target operation task scheduling strategy based on the preset target object state characteristics.

2. The method according to claim 1, characterized in that The job task scheduling strategy includes a job processing task and a task execution object of the job processing task, and calculating the reward score of the job task scheduling strategy includes: Obtaining an execution efficiency score of the task execution object in executing the job processing task; Obtaining an original object score of the task execution object; The task type of the job processing task is obtained, and a reward evaluation is performed according to the task type, the execution efficiency score, and the original object score to obtain the reward score.

3. The method according to claim 2, characterized in that The task type includes a collaboration type, and the reward evaluation is performed according to the task type, the execution efficiency score, and the original object score to obtain the reward score, including: If the task type of the job processing task is the collaborative type, the task execution object that executes the job processing task is used as a collaborative execution object; Obtaining an object collaboration score of the collaboration execution object; The reward score is obtained by performing a weighted calculation on the execution efficiency score, the original object score, and the object collaboration score.

4. The method according to claim 3, characterized in that The weighted calculation of the execution efficiency score, the original object score, and the object collaboration score to obtain the reward score includes: Get the scoring rule text; Performing text analysis on the scoring rule text to obtain scoring indicator weights; the scoring indicator weights include an efficiency weight parameter, an object weight parameter, and a collaboration weight parameter; Multiplying the efficiency weight parameter and the execution efficiency score to obtain a first score; Multiplying the object weight parameter and the original object score to obtain a second score; Multiplying the collaboration weight parameter and the object collaboration score to obtain a third score; The first score, the second score, and the third score are summed to obtain the reward score.

5. The method according to claim 1, wherein The predicting of the job task scheduling strategy and the execution probability of the job task scheduling strategy based on the first object state feature by using a preset original task scheduling model includes: Predicting an original probability of a preset task scheduling strategy based on the first object state feature by the original task scheduling model; Selecting the preset task scheduling strategy with the largest original probability as the job task scheduling strategy; The original probability of the job task scheduling strategy is used as the execution probability.

6. The method according to claim 1, characterized in that The acquiring of the first object state feature includes: Initialize the mapping list between preset tasks and preset execution objects; Performing a list value evaluation on the mapping list using a preset multi-dimensional value alignment model to obtain a matching score between the preset task and the preset execution object; Obtaining an object performance score of the preset execution object; The first object state feature is determined based on the matching score and the object performance score.

7. The method according to claim 1, characterized in that The reward score includes an execution efficiency score, an original object score, and an object collaboration score. After calculating the reward score of the job task scheduling strategy, the following steps are performed: Obtaining the number of state exceptions triggered when simulating an update of the state feature of the first object; determining a first score increment according to the number of abnormal states; Get the object performance score of the preset execution object; Performing weighted calculation on the execution efficiency score, the original object score, and the object collaboration score to obtain a second score increment; The object performance score is updated according to the first score increment and the second score increment.

8. A task assignment device, characterized in that: The device comprises: A state feature acquisition module, configured to acquire a state feature of a first object; A job prediction module, configured to predict a job task scheduling strategy and an execution probability of the job task scheduling strategy based on the first object state feature using a preset original task scheduling model; a job simulation module, configured to simulate and update the first object state feature according to the job task scheduling strategy to obtain a second object state feature; A score calculation module, used to calculate the reward score of the job task scheduling strategy; a state evaluation module, configured to perform a state evaluation on the state feature of the first object using a preset state evaluation model to obtain a first state score, and to perform a state evaluation on the state feature of the second object using the state evaluation model to obtain a second state score; a model optimization module, configured to calculate a target loss based on the execution probability, the reward score, the first state score, and the second state score, and update parameters of the original task scheduling model based on the target loss to obtain a target task scheduling model; The task scheduling output module is used to predict the target job task scheduling strategy based on the preset target object state characteristics through the target task scheduling model.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images