Human resource scheduling management method based on big data platform

Abstract

The invention relates to the field of data processing, in particular to a human resource scheduling management method based on a big data platform. The method comprises the steps of determining a to-be-configured target scene in response to a resource application instruction issued by a request object in a big data platform; predicting an activity change condition in the target scene based on the historical operation data to obtain an activity change feature of the target scene; verifying the business matching degree between the business change feature and the request object; and after the verification is passed, generating a scene visualization scheduling model of the target scene in a future preset time period according to the business change characteristics, and pushing the scene visualization scheduling model to the request object so as to facilitate the interaction operation between the request object and the scene visualization scheduling model in the big data platform. According to the method, intelligent management of human resource demands is realized to cope with complex and changeable actual activity conditions or business states, and human resource management efficiency is improved in an auxiliary manner.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a human resource scheduling and management method based on a big data platform. Background Art

[0002] Human Resource Management (HRM) is the process of effectively developing, rationally allocating, fully utilizing and scientifically managing the human resources in an organization through planning, organizing, coordinating and controlling.

[0003] Traditional linear models have limitations when dealing with the multivariate influences of scheduling requirements. Scheduling requirements are influenced by multiple variables, including business fluctuations, employee status, and the external environment. Traditional linear models (such as regression analysis) struggle to capture these complex relationships. Therefore, a new technical solution is urgently needed to address at least one of these issues. Summary of the Invention

[0004] In response to the technical problems existing in the prior art, the present application provides a human resource scheduling management method and device based on a big data platform to solve at least one technical problem existing in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a human resources scheduling management method based on a big data platform, the method comprising:

[0006] In response to a resource application instruction issued by a requesting party in a big data platform, a target scenario to be configured is determined; the target scenario is obtained based on the target task division contained in the instruction; the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario;

[0007] Based on the historical operation data matched by the request object, predict the activity changes in the target scenario and obtain the activity change characteristics of the target scenario in the future preset time period; the historical operation data matched by the request object includes at least: the working hours, number of personnel, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks of various activity scenarios in the category to which the request object belongs;

[0008] Verify the business matching degree between the business change characteristics and the request object;

[0009] After verification, a scene visualization scheduling model for the target scenario in a preset future time period is generated based on the business change characteristics, and the scene visualization scheduling model is pushed to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to realize intelligent management of human resource needs.

[0010] In a second aspect, an embodiment of the present application provides a human resources scheduling management device based on a big data platform, the device comprising at least the following units:

[0011] A determination unit is configured to determine a target scenario to be configured in response to a resource application instruction issued by a requesting object in the big data platform; the target scenario is obtained based on the target task division included in the instruction; the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario;

[0012] The prediction unit is configured to predict activity changes in a target scenario based on historical operation data matched by the request object, and obtain activity change characteristics of the target scenario in a future preset time period; the historical operation data matched by the request object includes at least: personnel working hours, personnel quantity, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks in various activity scenarios in the category to which the request object belongs;

[0013] A verification unit configured to verify a business matching degree between the business change feature and the request object;

[0014] The generation unit is configured to generate a scene visualization scheduling model for the target scene in a future preset time period according to the business change characteristics after verification, and push the scene visualization scheduling model to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to realize intelligent management of human resource needs.

[0015] The beneficial effect of this application is to provide a human resource scheduling management method and device based on a big data platform. This application can achieve intelligent management of human resource needs through scenario classification of target tasks and activity trend prediction based on scenario classification, which can better cope with complex and changing actual activity situations or business status, and help improve human resource management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flowchart of a human resources scheduling management method based on a big data platform according to an embodiment of the present application;

[0017] Figure 2 This is a use case diagram of a human resources scheduling management method based on a big data platform according to an embodiment of the present application;

[0018] Figure 3 This is a time sequence diagram of a human resources scheduling management method based on a big data platform according to an embodiment of the present application;

[0019] Figure 4 It is an entity relationship diagram of a human resource scheduling management method based on a big data platform in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0021] In order to solve at least one technical problem in the related art, the embodiment of the present application provides a human resource scheduling management method and device based on a big data platform. In this technical solution, through comprehensive analysis of multi-dimensional data and more advanced prediction methods, it is possible to more accurately capture the complex correlation between multiple variables, thereby improving the accuracy of human resource scheduling and reducing the waste or shortage of manpower caused by unreasonable scheduling. Business matching verification and visual interactive operations enable the scheduling plan to better adapt to actual business needs and various changing factors, and improve the system's adaptability to business fluctuations, employee status changes and external environmental influences. The visual scheduling model and interactive operation interface enable users to understand and operate the scheduling model more intuitively, speed up the decision-making process, improve decision-making efficiency, and enable enterprises to respond to various changes more quickly and make reasonable human resource scheduling decisions. Based on the big data platform, a variety of technical means are comprehensively used to realize the intelligent management of human resource needs, which can better cope with complex and changeable actual activity situations or business status, and assist in improving human resource management efficiency.

[0022] It can be understood that, first of all, the technical solution of this application can collect rich historical operation data matched by the request object based on the big data platform, covering multiple dimensions such as personnel working hours, number of personnel, personnel skills, personnel work status, scheduling plan selection preferences, safety risks, etc. Unlike the traditional linear model that only considers a few variables, it can comprehensively capture various factors that affect scheduling needs. For example, it not only focuses on changes in order volume caused by business fluctuations (traditional linear models may only focus on this), but also considers the impact of factors such as employee status (such as sudden illness, skill improvement) and external environment (such as policy changes, natural disasters) on scheduling. Through comprehensive analysis of multi-dimensional data, the complex relationship between various factors can be more accurately understood, thereby providing a richer information basis for scheduling decisions.

[0023] Secondly, the technical solution of this application predicts the changes in activities in the target scenario based on multi-dimensional historical operation data. Although the specific algorithm is not explicitly mentioned, it can be inferred that it may have adopted more advanced machine learning or deep learning algorithms (such as neural networks, decision tree integration, etc.), which can handle nonlinear relationships. Compared with traditional linear models that can only describe simple linear relationships, the new method can capture the complex nonlinear interactions between multiple variables such as business fluctuations, employee status and external environment. For example, during the e-commerce promotion period, not only the relationship between order volume and required manpower can be taken into account, but also the impact of employee skill improvement on work efficiency, as well as policy changes on working hours restrictions and other factors, to comprehensively predict more accurate human resource needs.

[0024] After obtaining the activity change characteristics of the target scenario, the technical solution of this application adds a step to verify the business matching between the business change characteristics and the request object. This step helps to further screen and confirm the rationality of the prediction results and avoid deviations caused by relying solely on the prediction model. By comprehensively considering the actual business needs and the characteristics of the request object, verifying and adjusting the prediction results can better cope with the uncertainty under the influence of multiple variables. For example, even if a certain human resource allocation plan is predicted, but it is found through business matching verification that it does not meet the actual business process or special needs of the request object, it can be corrected in time to improve the accuracy and practicality of the scheduling plan.

[0025] The technical solution of the present application generates a scene visualization scheduling model for the target scene and pushes it to the requesting object, so that the requesting object can perform interactive operations. This visualization and interactive method helps users intuitively understand the relationship between complex scheduling models and multiple variables, and also facilitates users to adjust the model according to actual conditions. Users can intervene and optimize the scheduling model based on their own understanding of business fluctuations, employee status and the external environment, making up for the lack of flexibility and adaptability of traditional linear models. For example, when there are sudden changes in the external environment (such as natural disasters), users can quickly adjust the scheduling model through interactive operations and reconfigure human resources to respond to emergencies.

[0026] Figure 1 A flowchart of a human resource scheduling management method based on a big data platform is provided in an embodiment of the present application, such as Figure 1 As shown, the method includes the following steps:

[0027] 101, in response to a resource application instruction issued by a requesting party in the big data platform, determining a target scenario to be configured;

[0028] 102. Predicting activity changes in the target scenario based on the historical operation data matched by the request object, and obtaining activity change characteristics of the target scenario in a future preset time period;

[0029] 103, verifying the business matching degree between the business change feature and the request object;

[0030] 104. After verification, a scene visualization scheduling model of the target scene in a future preset time period is generated according to the business change characteristics, and the scene visualization scheduling model is pushed to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to realize intelligent management of human resource needs.

[0031] In the embodiment of the present application, the target scenario is obtained based on the target task contained in the instruction. This is because the specific content and nature of the target task determine the type, quantity and scheduling method of the required human resources. For example, if the target task is to produce a certain product, then a series of links from raw material procurement, production and processing to finished product inspection have their own specific processes and requirements, and these characteristics will affect the definition of the production and manufacturing scenario. By analyzing the target task, the core objectives, key steps, required resources and other elements of the task can be clarified, so that it can be classified into the appropriate target scenario.

[0032] Further optionally, the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario.

[0033] The manufacturing scenario primarily involves the manufacturing and production process of products. In this scenario, there are usually fixed production processes and process requirements, requiring a large amount of production equipment and personnel with corresponding production skills. For example, in automobile manufacturing companies, from stamping, welding, painting to final assembly, each link requires workers with different skills, such as stampers, welders, painters, and assemblers. At the same time, the manufacturing scenario is also affected by factors such as production planning, equipment maintenance, and raw material supply. For example, if there is a delay in the supply of raw materials, it may be necessary to adjust the production plan and personnel scheduling to ensure smooth production.

[0034] Service scenarios focus on providing various services to customers. Service scenarios are characterized by high levels of demand uncertainty, necessitating timely adjustments to service strategies and staffing arrangements based on specific customer needs and circumstances. For example, during peak tourist season, hotel occupancy rates surge, leading to a corresponding increase in demand for front desk staff, housekeeping staff, and catering staff. During off-season, staffing may need to be reduced or work hours adjusted. Furthermore, service scenarios must prioritize customer satisfaction, and soft skills such as communication and adaptability are crucial for service personnel.

[0035] A project implementation scenario typically involves a series of activities carried out for a specific project. Projects have clear objectives, timelines, and resource requirements. Software development projects require collaboration across diverse roles, including requirements analysts, software engineers, and testers. Personnel scheduling within a project implementation scenario must consider factors such as the project's progress, critical path, and personnel skill matching. For example, if a particular skill is in high demand at a particular project stage, personnel with that skill must be promptly deployed to ensure timely project completion.

[0036] In an embodiment of the present application, the historical operation data matched by the request object includes at least: the working hours, number of personnel, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks of personnel in various activity scenarios in the category to which the request object belongs.

[0037] Personnel work time data records the hours worked by personnel across various activity scenarios within the requested category. This data can reflect personnel workloads in different scenarios. In manufacturing scenarios, if workers on a particular production line work excessively long hours, fatigue and reduced productivity may result, necessitating adjustments to personnel scheduling or work arrangements. Furthermore, personnel work time data can be used to analyze personnel efficiency and output, providing a reference for subsequent human resource planning.

[0038] Headcount data refers to the number of staff involved in different activity scenarios. By analyzing historical headcount data, we can understand the patterns of staffing demand across different scenarios and time periods. In service scenarios, historical data can be used to predict how many additional staff members will be needed to meet customer demand during specific promotions or peak business periods. Headcount data can also be used to assess the rationality of staffing and avoid overstaffing or understaffing.

[0039] Personnel skills data covers the various skills possessed by personnel within the category to which the request belongs. Skill requirements vary significantly across different target scenarios. In manufacturing, workers may be required to possess specific operational skills and process knowledge; in project implementation, project team members may need to possess a variety of skills, including project management and technical R&D. Understanding personnel skills data helps optimize personnel placement, ensuring that personnel are competent for their assigned tasks and improving work quality and efficiency.

[0040] Personnel work status data includes information on attendance, work enthusiasm, and performance. In human resource scheduling, personnel work status directly impacts work completion. If an employee has recently shown low work enthusiasm, it may be necessary to understand the cause and take appropriate measures, such as adjusting work tasks or providing incentives. Personnel work status data can also be used to assess overall team performance, providing a basis for team management and personnel training.

[0041] Scheduling preference data records the preferences of requesters for different scheduling options during past scheduling processes. These preferences may be influenced by various factors, such as cost, efficiency, and employee satisfaction. By analyzing this data, we can understand which scheduling options requesters prefer in different situations. This provides a reference for future scheduling decisions and improves the feasibility and satisfaction of scheduling options.

[0042] Security risk data identifies potential safety hazards and risk factors across various activity scenarios. In manufacturing, safety risks may include equipment failure and improper operation. In service scenarios, risks include customer complaints and service disputes. Understanding security risk data helps you take appropriate preventative measures to ensure personnel safety and the smooth operation of your business. Security risk factors also need to be considered when scheduling human resources, ensuring that personnel are appropriately allocated to mitigate the likelihood of risk.

[0043] Figure 2 This is a use case diagram of the human resource scheduling management method based on the big data platform. Figure 2The entire process, from the requester initiating a resource request to the final generation of a visual scheduling model and interactive operations, is described in this article. For example, a requester might submit a resource request to the human resources scheduling management system, for example, a department within an enterprise that needs to add specialized technical personnel for a new project. After receiving the request, the system classifies the target scenario, clarifying whether it is for a specific project, a specific business period (such as an e-commerce promotion), or overall business expansion. The system then retrieves historical operational data related to the requester from the big data platform, covering various aspects such as personnel work hours, number of personnel, skills, work status, scheduling plan preferences, and security risks. For example, this data includes personnel allocation and work efficiency data from past similar projects. Based on this historical operational data, the system predicts activity change characteristics for the target scenario, including trends in task demand characteristics at different classification granularities, such as increases or decreases in task volume and changes in required personnel skills at different project stages. The system then verifies the degree of match between the predicted activity change characteristics and the requester's business, based on the requester's attribute information, to determine whether the personnel demand changes align with the company's current business strategy and actual operations. After verification, a scenario-based visual scheduling model is generated that intuitively displays the scale, type, and scheduling of staffing needs, matching the human resources scheduling plan. For example, a chart displays staffing requirements and job position distribution over different time periods. Finally, the requester can interact with the visual scheduling model on the big data platform, adjusting or confirming it based on actual circumstances. If staffing allocations are found to be unreasonable during certain time periods, adjustments can be made. During this process, the requester initiates resource requests and participates in interactive confirmation of the final model. The big data platform provides historical data for analysis and prediction, and stores the visual model for easy operation. System administrators manage and configure the rule base, providing support for the system to rationally schedule human resources according to established rules.

[0044] As an optional embodiment, in step 101, in response to the resource application instruction issued by the requesting object in the big data platform, determining the target scenario to be configured can be implemented as follows:

[0045] 201, obtaining a target task to be executed from the instruction;

[0046] 202, identifying the task execution requirements of the target task, and setting the classification granularity in the scene classification rule based on the identification result;

[0047] 203, preliminarily classifying the target scenarios corresponding to the activity execution processes in the target tasks according to the set scenario classification rules;

[0048] 204 : Based on the historical resource application records and business execution status of the request object, the target scenario obtained by the preliminary division is optimized to obtain the target scenario to be configured.

[0049] In an embodiment of the present application, the scene classification rules are provided with: a matching relationship between the activity execution process and the scene, the scene classification granularity, and the correlation between the scene classification granularity and the feature dimension. Further optionally, the matching relationship between the activity execution process and the scene clarifies to which scene different activity execution processes should correspond. In a manufacturing enterprise, a series of activity execution processes, from raw material procurement, parts processing, product assembly to quality inspection, respectively correspond to procurement scenes, parts production scenes, product assembly scenes and quality inspection scenes. Through this matching relationship, the scope and main activity content of each scene can be clearly defined, so that when dividing the scenes for the target tasks, there is a clear basis to follow. For example, when an enterprise receives a new order and needs to produce a specific product, it can quickly determine the various scenes involved based on this matching relationship, and then reasonably arrange resources and dispatch personnel.

[0050] The granularity of scenario classification determines the level of detail used in the scenario division for the target task. A coarser classification granularity groups multiple similar activities or tasks into a larger scenario, while a finer classification granularity further subdivides tasks into more specific sub-scenarios. In a software development project, if a coarser classification granularity is used, the project might be divided into only a few broad scenarios, such as requirements analysis, development, and testing. However, if a finer classification granularity is used, the development scenarios can be further subdivided into front-end development scenarios, back-end development scenarios, database development scenarios, and so on. Appropriate classification granularity facilitates more precise resource management and scheduling to meet the needs of different tasks.

[0051] Characteristic dimensions include aspects such as task size, required skills, time span, and resource requirements. The granularity of scenario classification is closely related to these characteristic dimensions. In some tasks with high time requirements, the granularity of scenario classification may be adjusted based on the characteristic dimension of time span. For example, for a short-term urgent project, a finer classification granularity may be adopted to divide the project into different scenarios according to time stages in order to better monitor and manage project progress. For some large-scale tasks involving multiple skills, the characteristic dimension of required skills may be used to divide the tasks into scenarios with different skill requirements in order to rationally allocate personnel with corresponding skills.

[0052] In an embodiment of the present application, the scenario classification rules are dynamically updated based on the development trends of active businesses in the big data platform, changes in the market environment, and dynamically updated business models.

[0053] Big data platforms can collect and analyze data from various activities within an enterprise, thereby revealing business development trends. As a company's business grows and diversifies, new business activities and processes may continue to emerge. For e-commerce companies, new business models such as livestreaming with sales may emerge as their business evolves. At this point, scenario classification rules need to be updated based on these business development trends. New livestreaming with sales scenarios may be added, with clearer definitions of the activity execution process within these scenarios (such as host preparation, product display, and order processing), as well as the corresponding classification granularity and feature dimension relationships (such as host skill requirements and livestream scheduling).

[0054] Changes in the market environment, such as adjustments in competitors' strategies and changes in consumer demand, will also have an impact on a company's business, requiring updates to scenario classification rules. In the face of fierce market competition, companies may need to pay more attention to cost control and efficiency improvement. This may lead to increased attention to characteristic dimensions such as resource requirements and time spans in scenario classification rules, and corresponding adjustments to the classification granularity. For example, some scenarios that could have been merged can be further subdivided to more accurately control costs and improve production efficiency. At the same time, based on changes in market demand, some new scenarios related to marketing, customer service, etc. may be added, and corresponding matching relationships and rules may be established.

[0055] With the development of technology and innovations in corporate management concepts, business models may be constantly updated. For example, companies have adopted new business models such as remote work and agile development. In remote work models, scenario classification rules may need to take into account factors such as employees' geographic location and network environment, and redefine scenarios. In agile development models, which emphasize rapid iteration and team collaboration, scenario classification rules may place greater emphasis on task flexibility and adaptability, adjusting the matching relationship between activity execution processes and scenarios, as well as the relationship between classification granularity and feature dimensions, to meet the requirements of this new business model.

[0056] Based on the above assumptions, in step 201, when the requesting party issues a resource request instruction on the big data platform, the system first extracts the target task to be executed from the instruction. This is the foundation of the entire target scenario determination process, as subsequent operations are centered around the target task. For example, in an enterprise project management scenario, the requesting party may be a project team, and the resource request instruction it issues may include a target task such as "develop a new mobile application." The system needs to accurately capture this task description for further analysis and processing.

[0057] In step 202, the target task is analyzed in depth to identify the various elements necessary for task execution. This includes, but is not limited to, required personnel skills (e.g., developing a mobile app may require programmers with specific programming language skills, or designers with interface design capabilities), resources (e.g., server resources, development tools), timelines (e.g., project delivery deadlines), and collaboration methods (e.g., communication and division of labor among team members). A comprehensive understanding of these requirements allows for a more accurate determination of the characteristics and requirements of the target task.

[0058] According to the identified task execution requirements, set the classification granularity in the scenario classification rules. The classification granularity determines the level of detail in the scenario division of the target task. If the task execution requirements are more complex and involve multiple different links and elements, you may need to set a finer classification granularity to more accurately divide the target scenario. For example, for a large software development project, it may be necessary to subdivide it into multiple sub-scenarios such as requirements analysis, design, coding, and testing. For some relatively simple tasks, such as document organization, a coarser classification granularity may be used to classify them into a unified scenario.

[0059] In step 203, the scenario classification rules include information such as the matching relationship between the activity execution process and the scenario, the scenario classification granularity, and the correlation between the scenario classification granularity and the feature dimension. Based on these rules, the system matches each link of the activity execution process in the target task with the corresponding scenario. For example, if the target task is to produce a certain product, based on the matching relationship between the activity execution process and the scenario, the raw material procurement link is matched to the procurement scenario, and the production and processing link is matched to the production and manufacturing scenario. At the same time, the specific scope and content of each scenario are determined in combination with the set classification granularity. If the classification granularity is finer, the production and manufacturing scenario may be further subdivided, such as into component manufacturing scenarios and finished product assembly scenarios.

[0060] In step 204, the requester's historical resource request records are referenced to understand their resource requirements and scheduling for similar tasks or scenarios. By analyzing this historical data, potential patterns and trends can be identified, such as which resources are frequently required for specific tasks and which scenario partitioning methods are more reasonable and effective. This information can serve as an important reference for optimizing the target scenario.

[0061] It's understandable that in addition to historical resource request records, the business execution of the requesting party is also considered. This includes the actual completion of the task, any problems encountered, and any challenges. For example, if in a previous project, improper scenario division led to resource scheduling disruptions and impacted project progress, the optimization of the target scenario should prevent similar issues from recurring. By comprehensively considering historical records and business execution, the initially generated target scenario is adjusted and optimized to better align with actual business needs and resource scheduling requirements. Scenario classification rules are not static but are dynamically updated based on business development trends, market changes, and evolving business models within the big data platform. As business evolves and the market environment changes, task execution requirements and scenario characteristics may also change. For example, technological advancements may alter the process and skills required for developing mobile applications, requiring corresponding scenario classification rules to adapt to these changes. Furthermore, the emergence of new business models may introduce new scenario types and classification methods, requiring timely updates to ensure that the target scenario divisions remain accurate and reasonable.

[0062] The above steps can more accurately and flexibly determine the target scenario to be configured, providing a more reliable foundation and basis for subsequent human resource scheduling and other tasks.

[0063] As an optional embodiment, in step 202, identifying the task execution requirements of the target task and setting the classification granularity in the scene classification rule based on the identification result can be implemented as follows:

[0064] Extract multidimensional task requirement features in the target task; evaluate the feature complexity index of the target task based on the multidimensional task requirement features; input the multidimensional task requirement features and the feature complexity index into a decision-making agent, and select the optimal dynamic granularity level of the target task; and set the scene classification granularity corresponding to the task requirement features of different dimensions in the target task based on the dynamic threshold interval in the optimal dynamic granularity level.

[0065] Specifically, the aforementioned multiple task requirement characteristics are extracted from the target task, comprehensively collecting information on task-related skills, collaboration, strategy, and functions to provide a data foundation for subsequent analysis. First, based on the extracted multi-dimensional task requirement characteristics, the target task's feature complexity index is assessed. This index measures the task's complexity, potentially using a specific algorithm or model that comprehensively considers factors such as the number and interrelationships of various characteristics. For example, a high frequency of cross-departmental collaboration and diverse skill requirements will result in a higher feature complexity index. The multi-dimensional task requirement characteristics and feature complexity index are then input into a decision-making agent. Based on pre-defined rules, algorithms, or machine learning models, the decision-making agent analyzes the task characteristics and complexity and selects the optimal dynamic granularity level from multiple possible granularity levels. This level is the classification granularity level that best reflects the characteristics of the task. Finally, based on the dynamic threshold range within the optimal dynamic granularity level, the scenario classification granularity is set for the different dimensions of the task requirement characteristics in the target task. For example, if the complexity index of the skill requirement label falls within a certain threshold range, the skill requirements can be subdivided into different granularities such as basic skills, professional skills, and advanced skills, which can be used to divide target scenarios with different skill requirements, thereby achieving the division of target tasks based on different task requirement characteristics and obtaining corresponding target scenario classification results.

[0066] In this embodiment of the present application, the multi-dimensional task requirement characteristics include at least one of the following: skill requirement tags, frequency of cross-departmental collaboration, strategic relevance between the requesting party and the business department, and functional match between the requesting party and the business department. The multi-dimensional task requirement characteristics of the target task are the basic data source for the entire process, covering several key aspects:

[0067] Skill requirement tags are used to analyze the specific content and objectives of the target task and identify the various skills required to complete it. For example, for an AI project, skill requirement tags might include machine learning algorithms (such as neural networks and decision trees), programming languages (Python, Java, etc.), and data processing and analysis skills (data cleaning and feature engineering). By breaking down and analyzing the task, these skills can be accurately extracted in the form of tags.

[0068] The frequency of cross-departmental collaboration investigates how often other departments collaborate during the execution of the target task. This information can be obtained by reviewing project plans, historical records, or communicating with relevant personnel. For example, a large product development project may require frequent collaboration with marketing, production, quality control, and other departments. Recording the number and frequency of collaborations with each department can serve as a concrete indicator of the frequency of cross-departmental collaboration.

[0069] The strategic correlation between the requesting party and the business unit to which it belongs is used to clarify the degree of alignment between the objectives and tasks assigned by the requesting party (e.g., a project group or team) and the overall strategic goals of the business unit. This requires a clear understanding of the business unit's strategic planning and a comparative analysis of the objectives and tasks. For example, if the business unit's strategic focus is expanding into new markets and the objective task is developing a product targeting these markets, then the strategic correlation between them is high. Conversely, if the objective task is unrelated to the strategic focus, the strategic correlation is low. This correlation can be quantified using a scoring or grading method.

[0070] The functional fit between the requester and their business department is used to determine whether the requester's capabilities and resources match the business department's functional requirements. For example, if a business department is primarily responsible for manufacturing, while the requester possesses R&D capabilities, the functional fit between them is low. However, if the requester's skills and resources align with manufacturing requirements, the functional fit is high. Similarly, specific assessment methods can be used to determine the specific degree of fit.

[0071] Specifically, the multidimensional task requirement characteristics of the target task are the basic data source for the entire process, covering several key aspects:

[0072] Skill requirement labels carefully analyze the specific content and objectives of the target task to determine the various skills required to complete it. For example, for an AI project, skill requirement labels might include machine learning algorithms (such as neural networks and decision trees), programming languages (Python, Java, etc.), and data processing and analysis skills (data cleaning and feature engineering). By breaking down and analyzing the task, these skills can be accurately extracted in the form of labels.

[0073] Frequency of cross-departmental collaboration: Investigate the frequency of collaboration with other departments during the execution of the target task. This information can be obtained by reviewing project plans, historical records, or communicating with relevant personnel. For example, a large product development project may require frequent collaboration with marketing, production, quality control, and other departments. Record the number and frequency of collaborations with each department as a concrete indicator of the frequency of cross-departmental collaboration.

[0074] The strategic relevance between the requesting entity and the business unit to which it belongs. This involves clarifying the degree of alignment between the objectives and tasks assigned by the requesting entity (e.g., a project group or team) and the overall strategic objectives of the business unit. This requires a clear understanding of the business unit's strategic planning and a comparative analysis of the objectives and tasks. For example, if the business unit's strategic focus is expanding into new markets and the objective task is developing a product targeting these markets, then the strategic relevance between them is high. Conversely, if the objective task is unrelated to the strategic focus, the strategic relevance is low. This relevance can be quantified using a scoring or grading approach.

[0075] The functional fit between the requester and their business department determines whether the requester's capabilities and resources match the business department's functional requirements. For example, if a business department is primarily responsible for manufacturing, while the requester possesses R&D capabilities, the functional fit between them is low. However, if the requester's skills and resources align with manufacturing requirements, the functional fit is high. Similarly, specific assessment methods can be used to determine the specific degree of fit.

[0076] After extracting multidimensional task requirement features, these features need to be combined to assess the target task's feature complexity index. This index reflects the task's complexity and is calculated using a specific algorithm or model. First, a weight is assigned to each dimension of the task requirement feature, depending on its importance within the task. For example, for a technical R&D task, the skill requirement tag might have a relatively high weight; whereas for a project involving coordination across multiple departments, the frequency of cross-departmental collaboration might have a higher weight. Then, based on the specific values of each feature (such as the number of skills, the frequency of collaboration, and the strategic relevance and functional fit scores), combined with the assigned weights, a composite value is calculated. This value is the target task's feature complexity index. For example, a weighted summation method can be used to calculate the index: Feature Complexity Index = Skill Requirement Tag Weight × Skill Requirement Tag Score + Cross-Departmental Collaboration Frequency Weight × Cross-Departmental Collaboration Frequency Score + Strategic Relevance Weight between the Requester and the Business Department × Strategic Relevance Score + Functional Fit Weight between the Requester and the Business Department × Functional Fit Score.

[0077] Furthermore, a decision-making agent is a system based on rules, algorithms, or machine learning. It selects the optimal dynamic granularity level that best suits the target task from multiple preset granularity levels based on the input multi-dimensional task requirements and feature complexity index. The decision-making agent may have a series of pre-defined rules and policies that correspond to different granularity levels based on different task characteristics and complexity index ranges. For example, when the feature complexity index is low and cross-departmental collaboration is low, the decision-making agent may select a coarser granularity level, dividing the task into larger scenario categories. However, when the feature complexity index is high and involves multiple complex skills and frequent cross-departmental collaboration, the decision-making agent may select a finer granularity level, breaking the task into multiple specific sub-scenarios. The decision-making agent can also be trained through machine learning, using historical data to learn the relationship between task characteristics, complexity index, and optimal granularity level, thereby more accurately selecting the appropriate granularity level for the current target task.

[0078] After determining the optimal dynamic granularity level, it includes dynamic threshold intervals for task requirement characteristics across different dimensions. Based on these threshold intervals, a corresponding scenario classification granularity is set for each dimension's task requirement characteristics. For example, for the skill requirement label dimension, if the threshold intervals in the optimal dynamic granularity level specify that when the number of skill requirements is within a certain range, the skill requirements are divided into three granularity levels: basic skills, intermediate skills, and advanced skills. When the number of skill requirements exceeds a certain threshold, these granularity levels are further subdivided into more sub-granularities, such as specialized skills in a specific field. For the dimension of cross-departmental collaboration frequency, the threshold intervals might be used to classify tasks with low collaboration frequency into independent scenarios, tasks with medium collaboration frequency into general collaboration scenarios, and tasks with high collaboration frequency into close collaboration scenarios. Within each scenario, further detailed classifications can be made based on other characteristics. In this way, based on the dynamic threshold intervals in the optimal dynamic granularity level, appropriate scenario classification granularity is set for each dimension's task requirement characteristics, thereby achieving a reasonable classification of target tasks across different dimensions and obtaining accurate target scenario classification results.

[0079] In the embodiment of the present application, the scene classification granularity corresponding to different task requirement characteristics is used to divide the target task according to different task requirement characteristics to obtain corresponding target scene classification results.

[0080] The principle of the scenario classification granularity corresponding to different task requirement characteristics is based on the diversity and complexity of task requirements. Through differentiated granularity settings, the target tasks can be accurately portrayed, thereby obtaining more reasonable target scenario classification results. Further optionally, different task requirement characteristics reflect different aspects of the target tasks. For example, the skill requirement label reflects the requirements of the task for the professional capabilities of personnel, the frequency of cross-departmental collaboration reflects the collaborative complexity of the task in the organizational structure, and the strategic relevance and functional matching are related to the positioning and role of the task in the business system. In order to comprehensively and meticulously classify the target tasks, it is necessary to set corresponding classification granularity for different characteristics to adapt to their unique attributes and changing patterns.

[0081] Optionally, the decision-making agent selects the optimal dynamic granularity level based on the target task's feature complexity index. The dynamic threshold range within this level provides a basis for setting the classification granularity. For skill requirement labels, if the task requires multiple, highly specialized skills, a finer classification granularity can be set, such as segmenting scenarios by specific skill type. For the frequency of cross-departmental collaboration, if the frequency is low, a coarser granularity can be set, dividing scenarios by the presence or absence of interdepartmental collaboration or the approximate frequency range, so that the classification granularity matches the complexity of the features.

[0082] By dividing the task requirements of each dimension at different scenario classification granularities in this way, the target task can be deconstructed from multiple perspectives. After each task requirement characteristic is classified according to its corresponding granularity and then combined, a target scenario classification result can be formed that comprehensively reflects the characteristics of the target task, achieving accurate classification of the target task.

[0083] Therefore, setting a dedicated classification granularity for different task requirement characteristics can avoid the errors caused by a "one-size-fits-all" classification approach. It can more accurately capture the details and characteristics of the target task, making the classification results more in line with the actual situation and providing a more reliable foundation for subsequent scenario analysis, resource allocation, etc. Accurate target scenario classification results help the big data platform to more clearly understand the task requirements. For example, fine-grained classification based on skill requirement labels can accurately match personnel with corresponding skills; classification based on the frequency of cross-departmental collaboration can plan cross-departmental collaboration processes and resources in advance, thereby achieving optimal allocation of human resources and other resources, and improving resource utilization efficiency. Because it takes into account multiple task requirement characteristics and the classification granularity can be dynamically adjusted according to the complexity of the characteristics, this classification method can better adapt to tasks of different types and complexities, as well as changes in task requirements caused by changes in the business environment. It enables the big data platform to effectively classify and manage scenarios when facing diversified businesses, improving the flexibility and efficiency of overall business operations.

[0084] Further optionally, the decision-making agent is trained by the Q-Learning algorithm with the scheduling efficiency improvement rate, labor cost saving rate, and task completion quality as reward functions.

[0085] The decision-making agent is trained using the Q-Learning algorithm, with the scheduling efficiency improvement rate, labor cost savings rate, and task completion quality as reward functions. Its principle is to use reward feedback to guide the decision-making agent to learn the optimal strategy. The effect is that the decision-making agent can make better decisions to improve the overall business benefits.

[0086] Q-Learning is a model-free reinforcement learning algorithm. Its core is to maintain a Q-value table, which stores the expected cumulative rewards from taking different actions in different states. The decision-making agent continuously interacts with the environment, selects actions based on the current state, obtains rewards, and updates the Q-value, gradually learning the optimal strategy to maximize long-term cumulative rewards. The three indicators of scheduling efficiency improvement, labor cost savings, and task completion quality are closely related to key business operational objectives. The scheduling efficiency improvement rate reflects the speed of task resource scheduling and directly affects the speed of business process advancement; the labor cost savings rate is related to the company's cost investment; and the task completion quality determines the quality of business output. Using these as reward functions can encourage the decision-making agent to learn in a way that improves overall business efficiency.

[0087] During training, the decision-making agent will be in different states when faced with a target task, such as different combinations of task requirement characteristics and different resource allocation scenarios. Based on its current state, it selects an action from the action space, such as a specific scenario classification granularity or resource allocation scheme. Then, based on the actual results of the task execution, it calculates the improvement in scheduling efficiency, labor cost savings, and task completion quality to determine a reward value. This reward value is used to update the Q-value table, allowing the decision-making agent to select a more optimal action based on the updated Q-value the next time it encounters a similar state, continuously optimizing its decision-making strategy.

[0088] In this way, the trained decision-making agent can more accurately select the optimal dynamic granularity level and set a reasonable scenario classification granularity based on the specific circumstances of the task. Furthermore, guided by the reward function, it can make resource allocation decisions that are more conducive to improving scheduling efficiency, saving labor costs, and ensuring task quality, thereby enhancing the big data platform's processing capabilities for target tasks. As the decision-making agent continuously learns and optimizes, the improved scheduling efficiency can accelerate business processes and reduce task wait times. The reduced labor costs can lower the company's operating costs. Improved task completion quality helps enhance business competitiveness, improve the company's reputation and profitability, and ultimately enhance the overall business benefits. The Q-Learning algorithm gives the decision-making agent strong adaptability. Because the reward function provides rewards based on actual task execution results, the decision-making agent can adapt to different target tasks and changes in the business environment. Regardless of how task requirements change, the decision-making agent can continuously learn and adjust its strategy to meet business requirements for scheduling efficiency, cost, and quality.

[0089] As an optional embodiment, in the above steps, evaluating the feature complexity index of the target task based on the multi-dimensional task requirement characteristics can be implemented as follows:

[0090] Calculate the single-dimensional information entropy corresponding to the task requirement characteristics in each dimension; the information entropy is used to represent the discreteness of the feature distribution of the corresponding dimension; calculate the mutual information corresponding to the task requirement characteristics between any two dimensions; the user information is used to represent the dependency relationship between the corresponding two dimensions; perform weighted fusion on the single-dimensional information entropy and the mutual information to obtain the feature complexity index of the task requirement characteristics of each dimension in the target task, wherein the weighted coefficient is obtained by reverse optimization using historical scheduling evaluation data.

[0091] The historical scheduling evaluation data includes at least: historical human resource scheduling efficiency and historical human resource operating costs.

[0092] Specifically, information entropy is a key concept in information theory, used to measure the uncertainty or dispersion of information. In this scenario, calculating the single-dimensional information entropy corresponding to the task requirement characteristics in each dimension can measure the dispersion of the corresponding dimension's feature distribution. For example, for the skill requirement label dimension, if the task's skill requirements are diverse and evenly distributed, its information entropy will be high, indicating high dispersion of the features in this dimension and complex skill requirements. Conversely, if the skill requirements are concentrated in a few categories, the information entropy will be low, indicating low dispersion. Mutual information is used to measure the dependency between two random variables. In this context, calculating the mutual information corresponding to the task requirement characteristics between any two dimensions can clearly indicate the degree of correlation between the two dimensions. For example, a high mutual information value between the frequency of cross-departmental collaboration and the functional match between the requesting party and the business department in which it resides indicates a strong dependency between the two, likely indicating a requesting party with a high functional match and a relatively stable or high frequency of cross-departmental collaboration. A low mutual information value indicates a weak correlation and minimal mutual influence.

[0093] Next, a weighted fusion of single-dimensional information entropy and mutual information is performed, comprehensively considering the discreteness of each dimension and the dependencies between dimensions, resulting in an index that comprehensively reflects the characteristic complexity of the target task. The weighting coefficients are derived through reverse optimization using historical scheduling evaluation data. Historical scheduling evaluation data, such as historical human resource scheduling efficiency and historical human resource operating costs, can reflect the actual situation during past task execution. By analyzing the relationship between this data and the characteristic complexity index under different weighting coefficients, a weighting coefficient that most accurately reflects the characteristic complexity of the task and is consistent with actual business scheduling conditions is reversely optimized, making the characteristic complexity index more scientific and practical.

[0094] Therefore, the complexity of the task can be quantified from multiple perspectives, providing a more accurate basis for the setting of subsequent scene classification granularity and the decision-making of decision-making agents, which helps to improve task scheduling and management efficiency.

[0095] As an optional embodiment, it is assumed that the dynamic threshold intervals in the optimal dynamic granularity level include at least three levels of dynamic threshold intervals. Furthermore, the three levels of dynamic threshold intervals include: a first-level dynamic threshold region where the feature complexity index is less than the first dynamic threshold, a second-level dynamic threshold region where the feature complexity index is greater than the first dynamic threshold and less than the second dynamic threshold, and a third-level dynamic threshold region where the feature complexity index is greater than the second dynamic threshold; the first dynamic threshold is greater than the second dynamic threshold.

[0096] Based on the above assumptions, in the above steps, setting the scene classification granularity corresponding to the task requirement features of different dimensions in the target task based on the dynamic threshold interval in the optimal dynamic granularity level can be implemented as follows:

[0097] Based on the feature complexity index of the task requirement characteristics of each dimension in the target task, determine the dynamic threshold interval corresponding to the task requirement characteristics of each dimension; set the scenario classification granularity corresponding to the task requirement characteristics within the first-level dynamic threshold area to the first-level coarse granularity; set the scenario classification granularity corresponding to the task requirement characteristics within the second-level dynamic threshold area to the second-level medium granularity; set the scenario classification granularity corresponding to the task requirement characteristics within the third-level dynamic threshold area to the third-level fine granularity; wherein, the higher the level of the scenario classification granularity, the finer the classification granularity, and the higher the complexity of the human resources situation required to be scheduled.

[0098] Specifically, the dynamic threshold interval in the optimal dynamic granularity level is divided into three levels. Using the first and second dynamic thresholds as boundaries, the range of the feature complexity index is divided into three regions. The first dynamic threshold region is where the feature complexity index is less than the first dynamic threshold, indicating low task feature complexity. The second dynamic threshold region is where the feature complexity index is greater than the first dynamic threshold and less than the second dynamic threshold, indicating medium task feature complexity. The third dynamic threshold region is where the feature complexity index is greater than the second dynamic threshold, indicating high task feature complexity.

[0099] Based on the previously calculated feature complexity index of each dimension's task requirement features within the target task, compare each dimension's feature complexity index with the three-level dynamic threshold range to determine its corresponding range. For example, if the feature complexity index of a dimension's task requirement features is 5, the first dynamic threshold is 8, and the second dynamic threshold is 4, then the feature complexity index falls within the second-level dynamic threshold range.

[0100] According to the dynamic threshold interval to which the task requirement characteristics belong, the corresponding scenario classification granularity is set. For task requirement characteristics within the first-level dynamic threshold area, the scenario classification granularity is set to the first-level coarse granularity. This is because the complexity of task characteristics in this area is low, and no fine classification is required. Coarse-grained classification can meet the task scheduling requirements. For task requirement characteristics within the second-level dynamic threshold area, the scenario classification granularity is set to the second-level medium granularity. Medium feature complexity requires a moderate classification granularity to balance scheduling efficiency and accuracy. For task requirement characteristics within the third-level dynamic threshold area, the scenario classification granularity is set to the third-level fine granularity. Due to its high feature complexity, more detailed classification is required to accurately grasp the task requirements and reasonably schedule human resources.

[0101] It is worth noting that the higher the level of scenario classification granularity and the finer the classification granularity, the more complex the human resources required for scheduling. This is because fine-grained classification means that tasks are divided more specifically, which may involve more special requirements, professional skills, or complex collaborative relationships, so human resources need to be matched more accurately, and the difficulty and complexity of scheduling also increase accordingly. For example, a level 3 fine-grained task may require personnel with specific professional skills and rich experience, and may also involve the collaboration of multiple departments and multiple positions, while a level 1 coarse-grained task may only require personnel with general skills, and scheduling is relatively simple.

[0102] Therefore, tasks can be classified according to their actual complexity, which helps to optimize the task scheduling process, rationally allocate human resources, and improve task execution efficiency.

[0103] Further optionally, after setting the scene classification granularity corresponding to the task requirement characteristics within the three-level dynamic threshold area to the three-level fine granularity in the above steps, historical scene classification data corresponding to the three-level fine-grained task requirement characteristics can be extracted from the historical task library; the BIRCH hierarchical clustering algorithm is used to extract high-frequency sub-labels corresponding to the task requirement characteristics at the three-level fine granularity from the historical scene classification data; and fine-grained scene classification labels corresponding to the task requirement characteristics at the three-level fine granularity are constructed based on the high-frequency sub-labels.

[0104] Specifically, more accurate fine-grained scene classification labels are constructed by extracting high-frequency sub-labels from historical data. The historical task library stores a large amount of data related to past tasks. After setting the scene classification granularity corresponding to the task requirement characteristics within the three-level dynamic threshold area to the three-level fine-grained level, historical scene classification data corresponding to the three-level fine-grained task requirement characteristics are extracted from the historical task library. These data contain the classification of similar high-complexity tasks in the past and are the basis for subsequent analysis. For example, if the three-level fine-grained task involves a high-end scientific research project task, the task classification data related to the previous high-end scientific research project is extracted from the historical task library, including various attributes of the task, classification methods, etc.

[0105] The BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies) hierarchical clustering algorithm is an efficient clustering algorithm suitable for processing large-scale data. It implements data clustering by constructing a cluster feature tree (CF-Tree). The extracted historical scene classification data is input into the BIRCH hierarchical clustering algorithm, and the algorithm performs cluster analysis based on the characteristics and similarities of the data. During the clustering process, sub-tags with higher frequencies can be identified. These high-frequency sub-tags reflect the common attributes or classification dimensions of the three-level fine-grained task requirement characteristics. For example, for the historical scene classification data of high-end scientific research project tasks, after clustering with the BIRCH algorithm, high-frequency sub-tags such as "experimental equipment requirements", "key technical indicators", and "interdisciplinary cooperation" may be found. These sub-tags are frequently appearing and representative features in this type of task classification.

[0106] Based on the extracted high-frequency sub-tags, fine-grained scenario classification tags corresponding to the task requirement characteristics at the three-level fine-grained level are constructed. The high-frequency sub-tags are integrated and sorted to form a complete classification tag system that can accurately describe the three-level fine-grained task characteristics. For example, based on high-frequency sub-tags such as "experimental equipment requirements", "key technical indicators", and "interdisciplinary cooperation", they are further refined into specific fine-grained scenario classification tags such as "high-precision electron microscope equipment", "nanoscale material synthesis technical indicators", and "biological-chemical interdisciplinary cooperation". This makes the three-level fine-grained task scenario classification more accurate, helps to more accurately understand task requirements, provides more detailed references for human resource scheduling and task allocation, and improves the processing efficiency and quality of high-complexity tasks.

[0107] As an optional embodiment, in 102, the task requirement characteristics at different classification granularities in the target scene are extracted; based on the task requirement characteristics at different classification granularities and the historical operation data matched by the request object, the activity change corresponding to the task requirement characteristics at different classification granularities in the target scene is predicted to obtain the activity change characteristics of the target scene in the future preset time period.

[0108] Among them, the activity change characteristics include at least: the change trend information corresponding to the task requirement characteristics at different classification granularities; the historical operation data matched by the request object includes at least: the working hours, number of personnel, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks of personnel in various activity scenarios in the category to which the request object belongs.

[0109] The principle behind the above steps is to analyze task demand characteristics and historical operational data to predict future activity changes, thereby facilitating better resource scheduling and task management. Specifically, the target scenario contains task demand characteristics at varying levels of granularity. These characteristics describe the requirements at different levels of the task, including task type, difficulty, and required resources. For example, in an engineering project scenario, coarse-grained task demand characteristics might represent the division of project phases (e.g., design phase, construction phase), while fine-grained task demand characteristics might represent the specific tasks within each phase (e.g., drawing and mechanical calculations within the design phase). By extracting these characteristics, the task composition and demand profile of the target scenario can be clearly defined. The historical operational data matched by the request object contains rich information, covering aspects such as personnel work hours, number of personnel, and personnel skills. This data reflects the operational performance of the category to which the request object belongs in similar past activity scenarios. For example, for a certain type of manufacturing enterprise (to which the request object belongs), its historical operational data reveals the working hours, number of personnel involved, required professional skills, and commonly used scheduling schemes for various positions under different production task scenarios, providing a reference for subsequent predictions. By combining task demand characteristics at different classification granularities with historical operational data, a predictive model is constructed using relevant technologies such as data mining and machine learning. This model learns the correlation between task demand characteristics and activity changes in historical data. For example, when a certain task demand characteristic emerges, how will the number of personnel and working hours change? The model then predicts activity changes corresponding to task demand characteristics at different classification granularities in the target scenario. Based on the current task demand characteristics and historical patterns, activity change characteristics within a preset future time period are inferred, such as the changing trend of task demand characteristics, whether demand is increasing or decreasing, and how the number of personnel will be adjusted.

[0110] By predicting activity change characteristics, we can predict in advance the changing trends of task demand characteristics at different classification granularities, enabling more accurate human resource planning. If demand for a specific fine-grained task is predicted to increase, personnel with the corresponding skills can be assigned in advance, avoiding staff shortages or overstaffing and improving resource utilization efficiency. By leveraging historical operational data, such as the scheduling preferences of the requesting parties, and predicting activity changes, we can optimize the scheduling plan for the current target scenario. Based on the predicted changes in task demand, we can select a scheduling strategy that best reflects actual conditions, improving scheduling rationality and effectiveness while reducing scheduling costs and time. Combining security risk information from historical operational data with activity change predictions allows relevant personnel to take proactive safety precautions. If changes in a task activity are predicted to introduce new safety risks, we can develop contingency plans, deploy safety equipment, and provide personnel training in advance, reducing the likelihood of safety incidents and ensuring the safe and stable conduct of all activities within the target scenario.

[0111] For example, in causal inference-enhanced forecasting, the first step is to clearly define the causal variables. The treatment variable is the scheduling preference, such as whether to use part-time customer service staff, a key decision factor. The outcome variable is the characteristic of activity change, such as the improvement in customer service staff productivity measured by the number of inquiries handled per unit time. Next, causal effect estimation is performed, using a doubly robust estimation method to calculate the causal effect of different scheduling options on activity changes.

[0112] For example, it can accurately calculate the specific change in customer service staff productivity after employing part-time customer service staff under specific conditions. It can also output counterfactual predictions, such as "If part-time customer service staff were not employed, staff productivity would have decreased by 10%." Finally, in the prediction fusion stage, the results from causal inference are combined with the predictions from the Transformer-LSTM model. The weighting of the two is dynamically adjusted based on historical prediction errors to produce a final, accurate and reliable prediction of activity change characteristics.

[0113] As an optional embodiment, in 103, for the change trend information corresponding to the task requirement characteristics at different classification granularities, the rationality of each change trend information is verified based on the attribute information of the request object to obtain the business matching degree between each change trend information and the request object; and it is determined whether the business matching degree between each change trend information and the request object reaches the preset qualification threshold.

[0114] In 104, after verification, the personnel demand information of the target scenario in the future time period is determined based on the change trend information that reaches the preset qualification threshold; the personnel demand information at least includes: the personnel demand scale and the required personnel type corresponding to the change trend information that reaches the preset qualification threshold; a human resource scheduling plan is constructed according to the personnel demand; the human resource scheduling plan includes the human resource demand curve under the change trend information of each dimension and the corresponding human resource configuration instructions; a scenario visualization scheduling model that matches the human resource scheduling plan in the future preset time period is generated and published to the corresponding big data platform.

[0115] For example, in step 103, the rationality of the trend information corresponding to the task requirement characteristics at different classification granularities must first be verified based on the attribute information of the requesting object. The attribute information of the requesting object may cover many aspects, such as the requesting object's industry, business scale, organizational structure, past business execution preferences, and current business strategic direction.

[0116] For example, if an e-commerce company's business exhibits significant seasonal fluctuations (a characteristic of the business), and the predicted trend in task demand characteristics shows an unusually large increase in customer service staff workload during a period outside of promotional season, the rationality of this trend needs to be assessed based on the company's business attributes. If this trend information is inconsistent with the company's historical business patterns and current business plans, the business fit between this trend information and the requested object is low.

[0117] Through a series of similar analyses and judgments, each trend is evaluated for its business compatibility with the requested object. This business compatibility is then compared with a preset qualification threshold. Only trend information with a business compatibility that meets the threshold will proceed to the next stage of processing.

[0118] Once the change trend information passes the business matching verification (i.e. reaches the preset qualification threshold), in step 104, the personnel demand information of the target scenario in the future period is determined based on these valid change trend information. The scale of personnel demand is determined based on the prediction of the amount of various tasks in the change trend information. For example, if a certain change trend indicates that the order processing volume will increase significantly in a certain period in the future, combined with the average manpower required to process a single order in the past, the corresponding personnel demand scale can be calculated, that is, how many additional staff are needed to meet the growth in business volume. The type of personnel required is determined based on factors such as the skill requirements and work experience corresponding to the characteristics of different task requirements. For example, if the business change trend involves a large amount of data analysis work, then the type of personnel required may be a data analyst with data analysis skills and experience; if it involves the expansion of new business areas, professionals familiar with the knowledge in this field may be required.

[0119] Based on the identified personnel demand information, a human resource scheduling plan is then constructed. This plan is the core component of the entire process, detailing how to rationally allocate human resources to meet business needs. Based on the trend information for each dimension, corresponding human resource demand curves are plotted. These curves visually illustrate the changing demand for human resources in terms of quantity and type at different points in the future over a predetermined period of time. For example, a demand curve for order processing personnel, plotted with time on the horizontal axis and headcount on the vertical axis, might show a sharp increase in demand in the first few hours of a promotion, followed by a gradual stabilization, and a smaller peak near the end of the promotion. The plan specifies specific human resource allocation instructions, namely, how to allocate personnel based on demand at different points in time. This includes detailed instructions on where to allocate personnel (e.g., secondment from other internal departments or external temporary staff), specific positions to which they are assigned, and how work hours should be allocated. For example, during peak order processing periods, personnel familiar with the business process could be temporarily transferred from the after-sales department to the order processing position, extending their working hours by two hours to ensure timely order processing.

[0120] Finally, based on the constructed human resource scheduling plan, a scenario-specific visualization scheduling model is generated. This model presents complex human resource scheduling information in an intuitive and visual manner, making it easier for the requester and relevant staff to clearly understand and use it. The model may adopt various visualization formats, such as Gantt charts to display the time schedule and staffing allocation of different tasks, and heat maps to display the demand for staff in different regions or business segments. After the model is generated, it is published to the corresponding big data platform. The requester can view and analyze this visualization scheduling model in real time on the big data platform, making necessary adjustments and optimizations based on actual conditions, thereby achieving intelligent management of human resource needs.

[0121] Through the above steps, the complete process from task demand forecasting to human resource scheduling plan formulation and visual presentation is realized.

[0122] exist Figure 3 In the timing diagram shown, the specific process of the human resources scheduling management system involved in the embodiment of the present application is as follows: Figure 3The following sequence diagram illustrates the interaction between the requesting party, the scheduling management system, the big data platform, the prediction module, the verification module, and the visualization module. From the overall process perspective, the requesting party submits a resource request containing a target task to the scheduling management system. The scheduling management system analyzes the target task and requests historical operational data (including personnel, business, preferences, and risks) from the big data platform. The big data platform returns the historical data. The scheduling management system inputs the data into the prediction module, which processes and returns the prediction results (including personnel needs, skills gaps, and risk points). The scheduling management system sends the prediction results to the verification module, which verifies compliance, efficiency, and adaptability, and returns a pass or fail result. If the verification passes, the scheduling management system notifies the visualization module to generate a visual scheduling model (such as a Gantt chart, heat map, or scenario comparison). After the visualization module completes model generation, the scheduling management system pushes the model to the requesting party. The requesting party performs interactive operations (adjusting the plan or confirming execution), and the scheduling management system stores the final plan and the interaction data. If the verification fails, the scheduling management system prompts the requesting party to make manual adjustments, and the requesting party enters the adjustment parameters. The scheduling management system enters a loop and re-predicts and verifies, that is, it first re-predicts to obtain new prediction results, and then verifies the new results again until the verification is passed. The subsequent process is consistent with the verification pass branch.

[0123] Figure 4It is an entity relationship diagram related to human resource scheduling management. This entity relationship diagram shows the entities and their attributes, including the request object, target scenario, historical operation data, activity change characteristics, and scenario visualization scheduling model, as well as the relationship between them. The request object contains contact information (contact_info), requester ID (primary key requester_id), requester type (such as department, project team, business line requester_type), category (category), and scheduling preference (scheduling_preference). The request object is the subject that initiates human resource needs. The target scenario has a scene ID (primary key scene_id), a target task description (task_description), a scene type (such as manufacturing, service, project implementation scene_type), and a preset time period (time_window). The target scenario is used to define the specific context of human resource needs. Historical operation data includes the data ID (primary key data_id), the associated requester ID (foreign key requester_id), as well as information such as personnel working hours (personnel_hours), headcount, personnel skills (skills), personnel status (status), safety risks (safety_risk), and historical scheduling plans (historical_schedule). This information is a crucial foundation for analysis and forecasting. Activity change features include the prediction ID (primary key prediction_id), the associated scenario ID (foreign key scene_id), as well as the demand forecast (demand_forecast), skills gap (skill_gap), risk points (risk_points), and forecast time (prediction_time). These features are predictions of the future performance of the target scenario based on historical data. The scenario visualization scheduling model includes the model ID (primary key model_id), the associated scenario ID (foreign key scene_id), the model type (such as Gantt chart, heat map, or scenario comparison model_type), model data (model_data), and update time (update_time). It is used to visually display human resource scheduling plans. The request object and the target scenario are connected through a "Publish / Correspond" relationship, indicating that the request object publishes resource requests corresponding to the target scenario. The "Ownership / Contains" relationship indicates that historical operation data belongs to the request object and contains historical information related to the request object. Activity change features are generated based on the target scenario through "Generate / Based on" predictions. Activity change features are then used to generate a scenario visualization scheduling model after verification.These relationships reflect the business logic from initiating demands from the request object, to performing scenario analysis and prediction based on historical data, and finally generating a visual scheduling model.

[0124] In the technical solution of this application, intelligent management of human resource needs can be achieved through scenario classification of target tasks and activity trend prediction based on scenario classification, which can better cope with complex and changeable actual activity situations or business status, and help improve human resource management efficiency.

[0125] In another embodiment of the present application, a human resource scheduling management device based on a big data platform is provided. The device includes the following units: a determination unit configured to determine a target scenario to be configured in response to a resource application instruction issued by a requesting object in the big data platform; the target scenario is obtained based on the target task division contained in the instruction; the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario; a prediction unit configured to predict activity changes in the target scenario based on historical operation data matched by the requesting object, and obtain activity change characteristics of the target scenario in a future preset time period; the historical operation data matched by the requesting object includes at least: personnel working hours, personnel quantity, personnel skills, personnel working status, scheduling scheme selection preference, and security risks in various activity scenarios in the category to which the requesting object belongs; a verification unit configured to verify the business matching degree between the business change characteristics and the requesting object; and a generation unit configured to, after verification, generate a scene visualization scheduling model for the target scenario in a future preset time period based on the business change characteristics, and push the scene visualization scheduling model to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to achieve intelligent management of human resource needs.

[0126] The role and function of each unit of the device can be referred to the various steps in the above method embodiment, which will not be expanded here.

[0127] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0128] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0129] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if such changes and modifications of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such changes and modifications.

Claims

1. A human resource scheduling management method based on a big data platform, characterized in that: The method comprises: In response to a resource application instruction issued by a requesting party in a big data platform, a target scenario to be configured is determined; the target scenario is obtained based on the target task division contained in the instruction; the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario; Based on the historical operation data matched by the request object, predict the activity changes in the target scenario and obtain the activity change characteristics of the target scenario in the future preset time period; the historical operation data matched by the request object includes at least: the working hours, number of personnel, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks of various activity scenarios in the category to which the request object belongs; Verify the business matching degree between the business change characteristics and the request object; After verification, a scene visualization scheduling model for the target scenario in a preset future time period is generated based on the business change characteristics, and the scene visualization scheduling model is pushed to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to realize intelligent management of human resource needs.

2. The human resource scheduling management method based on the big data platform according to claim 1 is characterized in that: The step of determining the target scenario to be configured in response to the resource application instruction issued by the requesting object in the big data platform includes: Obtaining a target task to be executed from the instruction; Identifying the task execution requirements of the target task, and setting the classification granularity in the scene classification rule based on the identification result; Preliminarily divide the target scenarios corresponding to the activity execution process in the target task according to the set scenario classification rules; the scenario classification rules are set with: the matching relationship between the activity execution process and the scenario, the scenario classification granularity, and the correlation relationship between the scenario classification granularity and the feature dimension; Based on the historical resource application records and business execution status of the request object, the target scenario obtained by preliminary division is optimized to obtain the target scenario to be configured; Among them, the scenario classification rules are dynamically updated based on the development trends of active business in the big data platform, changes in the market environment, and dynamically updated business models.

3. The human resource scheduling management method based on the big data platform according to claim 2 is characterized in that: The identifying the task execution requirements of the target task and setting the classification granularity in the scene classification rule based on the identification result includes: Extracting multidimensional task requirement features from the target task, wherein the multidimensional task requirement features include at least one of the following: skill requirement tags, frequency of cross-departmental collaboration, strategic relevance between the requesting party and the business department, and functional match between the requesting party and the business department; Evaluating the characteristic complexity index of the target task based on multi-dimensional task requirement characteristics; Inputting the multidimensional task requirement characteristics and the characteristic complexity index into a decision-making agent, selecting the optimal dynamic granularity level of the target task; using the scheduling efficiency improvement rate, labor cost saving rate, and task completion quality as reward functions, training the decision-making agent through the Q-Learning algorithm; Based on the dynamic threshold interval in the optimal dynamic granularity level, setting the scene classification granularity corresponding to the task requirement characteristics of different dimensions in the target task; The scene classification granularity corresponding to different task requirement characteristics is used to divide the target task according to different task requirement characteristics to obtain corresponding target scene classification results.

4. The human resource scheduling management method based on a big data platform according to claim 3 is characterized in that: The evaluating the feature complexity index of the target task based on the multi-dimensional task requirement characteristics includes: Calculate the single-dimensional information entropy corresponding to the task requirement characteristics under each dimension; the information entropy is used to represent the dispersion of the feature distribution of the corresponding dimension; Calculate the mutual information corresponding to the task requirement characteristics between any two dimensions; the user information is used to represent the dependency relationship between the corresponding two dimensions; The single-dimensional information entropy and the mutual information are weightedly fused to obtain the characteristic complexity index of the task requirement characteristics of each dimension in the target task; the weighted coefficient is obtained by reverse optimization using historical scheduling evaluation data; the historical scheduling evaluation data at least includes: historical human resource scheduling efficiency and historical human resource operating cost.

5. The human resource scheduling management method based on a big data platform according to claim 4 is characterized in that: The dynamic threshold intervals in the optimal dynamic granularity level include at least three levels of dynamic threshold intervals; The three-level dynamic threshold intervals are: a first-level dynamic threshold region where the feature complexity index is less than the first dynamic threshold, a second-level dynamic threshold region where the feature complexity index is greater than the first dynamic threshold and less than the second dynamic threshold, and a third-level dynamic threshold region where the feature complexity index is greater than the second dynamic threshold; the first dynamic threshold is greater than the second dynamic threshold; The step of setting the scene classification granularity corresponding to the task requirement features of different dimensions in the target task based on the dynamic threshold interval in the optimal dynamic granularity level includes: Determining the dynamic threshold interval corresponding to each dimensional task requirement feature based on the feature complexity index of each dimensional task requirement feature in the target task; The scenario classification granularity corresponding to the task requirement characteristics within the first-level dynamic threshold area is set to the first-level coarse granularity; the scenario classification granularity corresponding to the task requirement characteristics within the second-level dynamic threshold area is set to the second-level medium granularity; the scenario classification granularity corresponding to the task requirement characteristics within the third-level dynamic threshold area is set to the third-level fine granularity; among them, the higher the level of the scenario classification granularity, the finer the classification granularity, and the more complex the human resources situation that needs to be scheduled.

6. The human resource scheduling management method based on a big data platform according to claim 5 is characterized in that: After setting the scene classification granularity corresponding to the task requirement characteristics within the three-level dynamic threshold area to the three-level fine granularity, the method further includes: Extract historical scenario classification data corresponding to three-level fine-grained task requirement features from the historical task library; The BIRCH hierarchical clustering algorithm is used to extract high-frequency sub-labels corresponding to task requirement features at three levels of fine granularity from historical scene classification data. Based on the high-frequency sub-labels, fine-grained scene classification labels corresponding to the task requirement features at the third level of fine-grainedness are constructed.

7. The human resource scheduling management method based on a big data platform according to claim 1 is characterized in that: The process of predicting activity changes in the target scene based on the historical operation data matched by the request object and obtaining activity change characteristics of the target scene in a future preset time period includes: Extract task requirement features at different classification granularities in the target scenario; Based on the task requirement characteristics at different classification granularities and the historical operation data matched by the request object, the activity change corresponding to the task requirement characteristics at different classification granularities in the target scenario is predicted to obtain the activity change characteristics of the target scenario in the future preset period; Among them, the activity change characteristics include at least: the change trend information corresponding to the task requirement characteristics at different classification granularities; the historical operation data matched by the request object includes at least: the working hours, number of personnel, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks of personnel in various activity scenarios in the category to which the request object belongs.

8. The human resource scheduling management method based on a big data platform according to claim 7 is characterized in that: The verifying the business matching degree between the business change feature and the request object includes: For the change trend information corresponding to the task requirement characteristics at different classification granularities, the rationality of each change trend information is verified based on the attribute information of the request object, and the business matching degree between each change trend information and the request object is obtained; Determine whether the business matching degree between each change trend information and the request object reaches the preset qualification threshold; After the verification is passed, a scene visualization scheduling model for the target scene in a future preset time period is generated according to the business change characteristics, and the scene visualization scheduling model is pushed to the requesting object, including: Determine the personnel demand information for the target scenario in the future period based on the trend information reaching the preset qualification threshold; the personnel demand information includes at least: the scale of personnel demand and the type of personnel required corresponding to the trend information reaching the preset qualification threshold; Constructing a human resource scheduling plan based on the personnel demand; the human resource scheduling plan includes a human resource demand curve under the change trend information of each dimension and corresponding human resource allocation instructions; Generate a scenario visualization scheduling model that matches the human resource scheduling plan in a preset time period in the future and publish it to the corresponding big data platform.

9. A human resource scheduling management device based on a big data platform, characterized in that: The device comprises the following units, wherein: A determination unit is configured to determine a target scenario to be configured in response to a resource application instruction issued by a requesting object in the big data platform; the target scenario is obtained based on the target task division included in the instruction; the target scenario includes at least one of the following: a production and manufacturing scenario, a service scenario, and a project implementation scenario; The prediction unit is configured to predict activity changes in a target scenario based on historical operation data matched by the request object, and obtain activity change characteristics of the target scenario in a future preset time period; the historical operation data matched by the request object includes at least: personnel working hours, personnel quantity, personnel skills, personnel working status, scheduling plan selection preferences, and safety risks in various activity scenarios in the category to which the request object belongs; A verification unit configured to verify a business matching degree between the business change feature and the request object; The generation unit is configured to generate a scene visualization scheduling model for the target scene in a future preset time period according to the business change characteristics after verification, and push the scene visualization scheduling model to the requesting object so that the requesting object can interact with the scene visualization scheduling model in the big data platform to realize intelligent management of human resource needs.

Images