Recruitment strategy optimization method and device, equipment and storage medium

Abstract

The invention discloses a recruitment strategy optimization method and device, equipment and a storage medium, and the method comprises the steps: carrying out the layering of an initial population corresponding to a target recruitment demand according to a target non-dominated sorting function, obtaining a non-dominated sorting result, and further calculating the individual congestion degree in each non-dominated layer in the initial population; selecting a plurality of excellent individuals in the initial population according to the individual crowding degree and the non-dominated sorting result in each non-dominated layer; and performing genetic iteration based on each excellent individual, the multi-target fitness function and the predicted recruitment index under each demand dimension, and when the iteration population satisfies an iteration termination condition, determining a target individual according to the fitness value of each individual in the iteration population so as to determine a target recruitment strategy. Through the mode, multiple factors are comprehensively considered, a data-driven, efficient and accurate long-term recruitment strategy is provided for an enterprise, the human resource management efficiency and effect are remarkably improved, and the labor cost of direct manpower of the enterprise is reduced.

Description

Technical Field

[0001] The present application relates to the field of recruitment technology, and in particular to a recruitment strategy optimization method, device, equipment and storage medium. Background Art

[0002] Human resource recruitment is an indispensable but challenging part of business operations. To ensure the smooth progress of production activities, companies usually need to formulate human resource plans for half a year or even longer. This process requires weighing many factors, such as recruitment goals, talent sources, recruitment channels, cost-effectiveness of each channel, and employee turnover. At the same time, wage costs are also affected by variables such as recruitment scale, time span, and personnel turnover. However, in long-term planning, due to the inability to fully consider all factors, companies often give priority to meeting production needs and put factors such as wage costs in a secondary position. This practice may lead to disordered recruitment plans and ultimately harm the interests of employees and enterprises. Therefore, the challenge facing human resource managers is how to strike a balance between meeting production needs and controlling costs, while taking into account the various complex factors in the recruitment process to develop a reasonable, sustainable and cost-optimal recruitment strategy. Summary of the invention

[0003] The main purpose of this application is to provide a recruitment strategy optimization method, device, equipment and storage medium, aiming to solve the technical problem of how to formulate efficient and accurate long-term recruitment strategies while meeting multiple factors such as production needs and labor costs.

[0004] To achieve the above objectives, the present application proposes a recruitment strategy optimization method, which includes:

[0005] stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain a non-dominated sorting result of the initial population;

[0006] Calculate the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result;

[0007] Individuals are selected from the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result to determine a plurality of excellent individuals;

[0008] Genetic iteration is performed based on each excellent individual, multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population;

[0009] When the iteration population meets the iteration termination condition, a target individual is determined according to the fitness value of each individual in the iteration population, and a target recruitment strategy is determined based on the target individual.

[0010] In one embodiment, the step of stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population includes:

[0011] The fitness value of each individual in the initial population corresponding to the target recruitment demand is calculated based on the predicted recruitment indicators under each demand dimension and the multi-objective fitness function;

[0012] Calculating parameters according to the target non-dominated sorting function, the fitness value of each individual in the initial population, and the constraint violation degree of each individual in the initial population to determine the dominance parameter of each individual in the initial population;

[0013] The initial population is stratified according to the dominance parameters of each individual in the initial population to obtain a non-dominated sorting result of the initial population.

[0014] In one embodiment, the step of performing genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment index under each demand dimension to obtain an iterative population includes:

[0015] Generate offspring individuals of each excellent individual according to the adaptive genetic algorithm, and construct a temporary population based on the offspring individuals of each excellent individual and each excellent individual;

[0016] Calculate the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function;

[0017] Individuals are screened according to the fitness values ​​of the individuals in the temporary population, and an iterative population is constructed based on the screening results.

[0018] In one embodiment, the step of generating offspring individuals of each excellent individual according to the adaptive genetic algorithm and constructing a temporary population based on the offspring individuals of each excellent individual and each excellent individual includes:

[0019] Generate offspring individuals of each excellent individual according to the adaptive genetic algorithm;

[0020] When there are offspring individuals with excellent individuals that do not meet the recruitment constraints, the offspring individuals that do not meet the recruitment constraints are repaired according to the constraint repair mechanism to obtain the repaired offspring individuals;

[0021] A temporary population is constructed based on the repaired offspring individuals and excellent individuals.

[0022] In one embodiment, the step of calculating the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function includes:

[0023] Calculate the sub-objective function value of each individual in the temporary population under multiple evaluation dimensions according to the predicted recruitment index under each demand dimension;

[0024] Normalizing the sub-objective function values ​​of each individual in the temporary population under each evaluation dimension to obtain the evaluation function values ​​of each individual in the temporary population under each evaluation dimension;

[0025] A weighted calculation is performed according to the target penalty term, the target weight corresponding to each sub-objective function, and the evaluation function value of each individual in the temporary population under each evaluation dimension to determine the fitness value of each individual in the temporary population.

[0026] In one embodiment, before the step of performing genetic iteration based on each outstanding individual, the multi-objective fitness function and the predicted recruitment index under each demand dimension to obtain an iterative population, the step further includes:

[0027] Perform data processing on multi-source information of human resources to obtain processed multi-source data;

[0028] Perform multi-dimensional indicator forecasting based on the multi-source data processed, determine the human resource demand indicator, the predicted turnover rate indicator and the salary trend indicator;

[0029] The predicted recruitment indicators under each demand dimension are determined according to the manpower demand indicator, the predicted churn rate indicator and the salary trend indicator.

[0030] In one embodiment, before the step of stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population, the step further includes:

[0031] Encode the problem based on the recruitment decision parameters and determine multiple decision vectors;

[0032] Conduct feasibility judgment on each decision vector according to the feasibility judgment function;

[0033] When each decision vector satisfies the recruitment constraint condition, a plurality of initial individuals are determined according to each decision vector;

[0034] Construct an initial population corresponding to the target recruitment needs based on each initial individual.

[0035] In addition, to achieve the above purpose, the present application also proposes a recruitment strategy optimization device, which includes:

[0036] A stratification module, used for stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain a non-dominated sorting result of the initial population;

[0037] A calculation module, used for calculating the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result;

[0038] A selection module, configured to select individuals from the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result, and determine a plurality of excellent individuals;

[0039] The iteration module is used to perform genetic iteration based on each excellent individual, multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population;

[0040] A processing module is used to determine a target individual according to the fitness value of each individual in the iterative population when the iterative population meets the iteration termination condition, and determine a target recruitment strategy based on the target individual.

[0041] In addition, to achieve the above-mentioned purpose, the present application also proposes a recruitment strategy optimization device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the recruitment strategy optimization method described above.

[0042] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the recruitment strategy optimization method described above are implemented.

[0043] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the recruitment strategy optimization method described above.

[0044] The method of the present application includes: stratifying an initial population corresponding to a target recruitment demand according to a target non-dominated sorting function to obtain a non-dominated sorting result of the initial population; calculating the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result; performing individual selection in the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result to determine a plurality of excellent individuals; performing genetic iteration based on each excellent individual, a multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population; when the iterative population meets an iteration termination condition, determining a target individual according to the fitness value of each individual in the iterative population, and determining a target recruitment strategy based on the target individual. Through the above method, multiple excellent individuals are selected from the initial population through the non-dominated sorting results of the population corresponding to the individual crowding degree and the recruitment demand, and genetic iteration is performed based on multiple excellent individuals and multi-objective fitness functions. When the iterative population meets the termination condition, the target individual is selected based on the individual fitness value, thereby generating a target recruitment strategy, which comprehensively considers multiple factors such as human resource demand planning, labor cost, and personnel replacement, and provides enterprises with data-driven, efficient and accurate long-term recruitment strategies, which significantly improves the efficiency and effectiveness of human resource management and reduces the direct labor cost of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0047] Figure 1 A flowchart of the first embodiment of the recruitment strategy optimization method of this application is provided;

[0048] Figure 2 A schematic diagram of a multi-objective optimization algorithm flow chart of a recruitment strategy optimization method provided in Example 1 of the present application;

[0049] Figure 3 A flow chart of the second embodiment of the recruitment strategy optimization method of this application;

[0050] Figure 4 A brief flowchart of the recruitment strategy optimization method provided in Example 2 of the present application;

[0051] Figure 5 This is a schematic diagram of the module structure of the recruitment strategy optimization device of the embodiment of the present application;

[0052] Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the recruitment strategy optimization method in the embodiment of the present application.

[0053] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0055] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0056] The main solution of the embodiment of the present application is: stratify the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population; calculate the individual crowding in each non-dominated layer in the initial population according to the non-dominated sorting result; select individuals in the initial population according to the individual crowding in each non-dominated layer and the non-dominated sorting result to determine multiple excellent individuals; perform genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment indicators under each demand dimension to obtain an iterative population; when the iterative population meets the iteration termination condition, determine the target individual according to the fitness value of each individual in the iterative population, and determine the target recruitment strategy based on the target individual.

[0057] In long-term planning, companies often prioritize production needs and put factors such as wage costs in a secondary position because they cannot fully consider all factors. This practice may lead to disordered recruitment plans and ultimately harm the interests of employees and companies. Therefore, the challenge facing human resource managers is how to strike a balance between meeting production needs and controlling costs, while taking into account the various complex factors in the recruitment process to develop a reasonable, sustainable and cost-optimized recruitment strategy.

[0058] The present application provides a solution, which selects multiple excellent individuals from the initial population through the non-dominated sorting results of the population corresponding to the individual crowding and recruitment needs, performs genetic iteration based on the multiple excellent individuals and the multi-objective fitness function, and when the iterative population meets the termination condition, selects the target individual based on the individual fitness value, thereby generating a target recruitment strategy, comprehensively considering multiple factors such as human resource demand planning, labor cost, and personnel replacement, providing enterprises with data-driven, efficient and accurate long-term recruitment strategies, significantly improving the efficiency and effectiveness of human resource management, and reducing the direct labor cost of enterprises.

[0059] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a recruitment strategy optimization device that can realize the above functions, etc. The following takes the recruitment strategy optimization device as the execution subject as an example to illustrate this embodiment and the following embodiments.

[0060] Based on this, the present application embodiment provides a recruitment strategy optimization method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the recruitment strategy optimization method of this application.

[0061] In this embodiment, the method includes steps S10 to S50:

[0062] Step S10, stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain a non-dominated sorting result of the initial population.

[0063] It should be noted that the target non-dominated sorting function is a method used to stratify individuals in a population in multi-objective optimization. The main purpose is to sort the individuals in the population according to their dominance relationship in the multi-objective space, thereby forming multiple non-dominated hierarchies. The individuals in each level have similar performance in the multi-objective space, and the lower the level, the better the individual. When an enterprise has a target recruitment demand, the parameters that need to be decided in the target recruitment demand, including but not limited to the number of recruits, employment cycle, and employment positions, are encoded as real vectors as decision vectors of the optimization algorithm. Based on multiple decision vectors, an initial population is generated, and the initial population includes multiple individuals, each of which represents a possible recruitment strategy, and each individual includes but is not limited to parameters such as the number of recruits, employment cycle, and employment positions.

[0064] It can be understood that the target non-dominated sorting function is used to stratify the initial population according to the target recruitment demand, thereby obtaining multiple non-dominated layers and individuals included in each non-dominated layer. In this embodiment, the non-dominated sorting result of the initial population includes but is not limited to multiple non-dominated layers corresponding to the initial population, individuals included in each non-dominated layer, and individual stratification results.

[0065] In a feasible implementation, step S10 may include steps A11 to A13:

[0066] Step A11, calculating the fitness value of each individual in the initial population corresponding to the target recruitment demand according to the predicted recruitment index under each demand dimension and the multi-objective fitness function.

[0067] It should be noted that multi-source information such as recruitment data, personnel flow data, and market salary data in the area where the enterprises with recruitment needs are located are obtained, and recruitment indicators in different demand dimensions in a certain period of time in the future are predicted based on the multi-source information, so as to obtain predicted recruitment indicators in each demand dimension. In this embodiment, the predicted recruitment indicators in each demand dimension include but are not limited to key indicators such as manpower demand, employee turnover rate, and market salary trend.

[0068] It can be understood that the multi-objective fitness function is used to evaluate the performance of each individual on each goal. In this embodiment, the multi-objective fitness function F(x) considers the following sub-goals: human resource demand satisfaction f1(x), recruitment cost control f2(x), staff turnover rate f3(x), recruitment quality f4(x), where x is the decision vector corresponding to the individual. The multi-objective fitness function F(x) is constructed using the Chebyshev weighting method: F(x) = max{w1f1'(x), w2f2'(x), w3f3'(x), w4f4'(x)}+P(x), where f i '(x) is the normalized sub-objective function value, w i is the weight corresponding to each sub-objective function. The weight can be pre-set or adaptively modified. P(x) is the penalty term used to handle constraint violations.

[0069] In the specific implementation, processing constraints to ensure the effectiveness of the optimization results is a key link in the multi-objective optimization problem. In this embodiment, the constraints are divided into two categories: equality constraints and inequality constraints: i (x)=0,i=1,2...,p, inequality constraint: g i (x)≤0,i=1,2...,q, p and q are the number of equality and inequality constraints respectively. Design a feasibility judgment function Feasible(x) to evaluate whether the decision vector x satisfies all constraints: Where ε is the allowable error of the equality constraint. The constraint processing strategies are as follows: a) Penalty function method: Introduce the penalty term P(x) into the multi-objective fitness function F(x) to transform the constrained optimization problem into an unconstrained optimization problem. Where r is the penalty factor, and an adaptive adjustment strategy can be adopted. b) ε constraint method: Select a main sub-objective function f m (x), transform other sub-objective functions into constraints: f i (x)≤ε,i≠m, by adjusting the value of ε,i, different non-inferior solutions can be obtained. c) Feasibility rule: When comparing individuals, the following rules are adopted: feasible solutions are better than infeasible solutions; between two feasible solutions, the Pareto dominance relationship is used for comparison; between two infeasible solutions, the one with the smaller degree of constraint violation is better.

[0070] It should be noted that, in this embodiment, a constraint repair function is designed to map an infeasible solution to a feasible domain: This can be done in the following ways: a) Boundary repair: truncate the value beyond the boundary of the decision variable to the boundary value. b) Heuristic repair: design heuristic rules based on the characteristics of the problem to adjust the infeasible solution to the nearest feasible solution. c) Projection method: project the infeasible solution onto the boundary of the feasible domain.

[0071] It can be understood that in this embodiment, an adaptive mechanism is also introduced to dynamically adjust the constraint processing strategy: a) Penalty factor adaptation: dynamically adjust the penalty factor r according to the proportion of feasible solutions in the population. b) ε level adaptation: dynamically adjust ε, i in the ε constraint method according to the solution distribution. c) Constraint relaxation: appropriately relax the constraints at the beginning of optimization, and gradually tighten them with iterations to balance exploration and utilization.

[0072] In the specific implementation, a constraint satisfaction function C(x) is designed based on the recruitment constraints to quantify the degree to which the decision vector x satisfies the constraints: The value range of C(x) is (0,1], and the larger the value, the higher the degree of constraint satisfaction. In this embodiment, multi-objective optimization is performed through the above mechanism and function designed in this embodiment.

[0073] In the specific implementation, based on the predicted recruitment indicators and multiple fitness functions under each demand dimension, the fitness value of each individual in the initial population can be calculated.

[0074] Step A12, performing parameter calculation according to the target non-dominated sorting function, the fitness value of each individual in the initial population, and the constraint violation degree of each individual in the initial population, to determine the dominance parameter of each individual in the initial population.

[0075] It should be noted that the domination parameters include but are not limited to the domination count and dominated set of individuals. The degree of violation of the constraints of each individual is calculated. When sorting by the target non-dominated sorting function and the fitness value of each individual in the initial population, the degree of constraint violation of the individual needs to be considered, and the domination parameters of each individual are clarified by combining the fitness value of each individual in the initial population and the degree of constraint violation. For example: If both individuals A and B meet all constraints, they are compared according to the traditional non-dominated relationship. If both individuals A and B violate the constraints, but the constraint violation degree of A is less than the constraint violation degree of B, then A dominates B. If individual A meets all constraints and individual B violates the constraints, then A dominates B. If individual A violates the constraints and individual B meets all constraints, then B dominates A.

[0076] Step A13, stratifying the initial population according to the dominance parameters of each individual in the initial population to obtain a non-dominated sorting result of the initial population.

[0077] It should be noted that the individuals in the initial population are stratified according to the dominance parameters of each individual in the initial population. For example: First layer: all individuals with dominance counts of 0 are placed in the first layer. Second layer: the individuals in the dominated set of individuals in the first layer are placed in the second layer, and the dominance counts of these individuals are updated. Subsequent layers: repeat the above steps until all individuals are assigned to a certain layer. Non-dominated sorting result: The final stratification result is the non-dominated sorting result of the initial population. The individuals in each layer perform similarly in the multi-objective space, and the lower the level, the better the individual.

[0078] In a feasible implementation manner, before step S10, steps B11 to B14 may be included:

[0079] Step B11, encoding the question based on the recruitment decision parameters, and determining multiple decision vectors.

[0080] Step B12, performing feasibility judgment on each decision vector according to the feasibility judgment function.

[0081] Step B13, when each decision vector satisfies the recruitment constraint condition, a plurality of initial individuals are determined according to each decision vector.

[0082] Step B14, constructing an initial population corresponding to the target recruitment demand based on each initial individual.

[0083] It should be noted that the recruitment decision parameters, including but not limited to the number of recruits, employment period, and employment positions, are encoded as decision vectors in the optimization algorithm, and each decision vector represents a possible recruitment strategy. Determine whether each decision vector satisfies the recruitment constraint condition. In this embodiment, the recruitment constraint condition includes two types: equality constraint and inequality constraint: i (x)=0,i=1,2...,p, inequality constraint: g i (x)≤0,i=1,2...,q.

[0084] It can be understood that when the decision vector meets the recruitment constraint conditions, it is used as an individual in the initial population; when the decision vector does not meet the recruitment constraint conditions, the decision vector can be eliminated, or the decision vector can be repaired by applying the constraint repair function, and the repaired decision vector can be used as an individual in the initial population.

[0085] Step S20, calculating the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result.

[0086] It should be noted that for the initial population: the individuals contained in each non-dominated layer are determined based on the non-dominated sorting results, and the crowding distance of individuals in the same non-dominated layer is calculated. The crowding degree is used to measure the distribution density of individuals in their level to maintain the diversity of the population. Generally, individuals with greater crowding degrees have fewer individuals around them and are more sparsely distributed.

[0087] Step S30 , selecting individuals from the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result, and determining a plurality of excellent individuals.

[0088] It should be noted that the selection method uses tournament selection to select multiple excellent individuals from the initial population based on the crowding degree and non-dominated sorting results of individuals in each non-dominated layer, and gives priority to individuals with low non-dominated levels and high crowding degrees, thereby obtaining multiple excellent individuals corresponding to the initial population.

[0089] Step S40, performing genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment index under each demand dimension to obtain an iterative population.

[0090] It should be noted that a crossover operation is performed on the selected excellent individuals to generate new offspring individuals. A mutation operation is performed on the selected excellent individuals to increase the diversity of the population. In this embodiment, the selection of the crossover operator and the mutation operator can be guided by a deep reinforcement learning model to dynamically adjust the crossover strategy.

[0091] It can be understood that the excellent individuals in the parent generation and the generated offspring individuals are merged into a temporary population. The fitness value of each individual in the temporary population is calculated according to the multi-objective fitness function and the predicted recruitment indicators under each demand dimension. According to the fitness values ​​of the individuals in the temporary population, the best individuals are selected to form a new generation of population, thereby obtaining iteration.

[0092] In a feasible implementation manner, before step S40, steps C11 to C13 may be included:

[0093] Step C11, performing data processing on multi-source information of human resources to obtain processed multi-source data.

[0094] It should be noted that the human resources multi-source information package refers to the relevant data of the region where the enterprise with recruitment needs is located, including but not limited to recruitment data, personnel turnover data and market salary data. Among them, the recruitment data includes but is not limited to information such as job type, recruitment channel, recruitment cycle, number of applicants, and personnel's place of origin; the personnel turnover data includes but is not limited to information such as reasons for employee resignation, expected contract cycle, length of service, job level, turnover rate, and loss rate; the market salary data includes but is not limited to information such as industry region, job level, work experience, and salary price. The human resources multi-source information is cleaned, standardized, and feature-engineered to obtain processed human resources multi-source information. In this embodiment, the processed multi-source information refers to the processed human resources multi-source information.

[0095] Step C12, performing multi-dimensional indicator prediction based on the multi-source data processing to determine the manpower demand indicator, the predicted turnover rate indicator and the salary trend indicator.

[0096] It should be noted that multi-dimensional indicator prediction is performed based on processing multi-source data to determine the human resource demand indicator, the predicted turnover rate indicator, and the salary trend indicator. In this embodiment, the indicator prediction can be performed in the following ways: a) time series analysis, processing linear time series data, considering factors such as seasons. b) deep learning prediction, capturing long-term dependencies in the data.

[0097] It is understandable that the specific steps when using time series analysis and deep learning prediction are as follows: select the appropriate model for each type of prediction task based on the data characteristics. For example: for manpower demand prediction, the seasonal autoregressive integrated moving average model is used to consider seasonal factors; for employee turnover rate prediction, the random forest model is used, which can handle multiple features and capture nonlinear relationships; for market salary trend prediction, the long short-term memory network model is used, which is suitable for processing long-term time series data.

[0098] In the specific implementation, when training the model: train each selected model. For example: use historical labor demand data to determine the best neural network parameters and seasonal parameters; use historical employee turnover data for training, and optimize hyperparameters such as the number of trees, maximum depth, etc. through grid search; use historical market salary data for training, and adjust hyperparameters such as the number of network layers, number of neurons, and learning rate. Use the time series cross-validation method to divide the data into multiple training sets and validation sets in chronological order to evaluate the performance of the model in different time periods. Use a variety of evaluation indicators to measure model performance. For example, the root mean square error evaluates the deviation between the predicted value and the actual value. The mean absolute error measures the absolute size of the prediction error. The R2 score evaluates the ability of the model to explain data variation. Based on the evaluation results, perform model optimization. For example: adjust the difference order and moving average term to better capture trends and seasonality; perform feature importance analysis to screen the most relevant features; try different network structures, such as adding convolutional layers or attention mechanisms. Ensemble learning methods can be used to combine the prediction results of multiple models to improve prediction accuracy.

[0099] It should be noted that the trained model is used to generate forecast results for a certain period in the future, such as 6 months or 1 year, so as to obtain the manpower demand index, the predicted turnover rate index and the salary trend index. In this embodiment, an incremental learning mechanism is designed to regularly update the model with new data to adapt to the latest changes in the recruitment market.

[0100] Step C13, determining the predicted recruitment index under each demand dimension according to the manpower demand index, the predicted turnover rate index and the salary trend index.

[0101] It should be noted that based on the structured data forms corresponding to the manpower demand index, the predicted turnover rate index and the salary trend index, the predicted recruitment index under each demand dimension is determined as the input of the subsequent multi-objective optimization algorithm. In this embodiment, the multi-objective optimization algorithm uses technologies such as reinforcement learning and genetic algorithms to seek the best balance between multiple goals such as meeting manpower demand, controlling costs and reducing staff turnover.

[0102] Step S50, when the iteration population meets the iteration termination condition, a target individual is determined according to the fitness value of each individual in the iteration population, and a target recruitment strategy is determined based on the target individual.

[0103] It should be noted that the iteration termination conditions include but are not limited to the number of iterations reaching an upper limit, the population convergence reaching a set value, etc. When the iteration population meets any of the above iteration termination conditions, the iteration is stopped, and a certain proportion of individuals are selected from the iteration population as target individuals based on the fitness value of each individual in the iteration population. In this embodiment, a Pareto optimal solution set is formed based on at least one selected target individual, and the recruitment strategy corresponding to the target individual included in the optimal solution set is the target recruitment strategy.

[0104] It is understandable that when the iteration population does not meet the iteration termination condition, combined with Figure 2 The processing flow of the multi-objective optimization algorithm shown in the figure repeats the following steps: non-dominated sorting: sorting the population in layers based on the Pareto dominance relationship; crowding calculation: calculating the crowding of individuals in the same non-dominated layer to maintain population diversity; using the tournament selection method to select excellent individuals through non-dominated sorting results and crowding; applying adaptive genetic operators to generate offspring individuals; merging parent and offspring populations, evaluating and selecting the best individuals to form a new generation population based on the multi-objective fitness function. The multi-objective optimization algorithm of this embodiment can effectively ensure the feasibility and effectiveness of the multi-objective optimization results, can flexibly respond to complex recruitment strategy optimization problems, and achieve balanced optimization of multiple objectives while meeting various constraints.

[0105] The multi-objective optimization algorithm of this embodiment integrates the constraint processing content involved in the above content. The specific process is as follows: a) Initialization: Generate an initial population that satisfies the constraints. b) Mutation and crossover: For the generated infeasible offspring, apply the constraint repair function. c) Selection: Incorporate constraint satisfaction into the tournament selection. d) Non-dominated sorting: Consider the degree of constraint violation in the Pareto sorting. e) Local search: Perform local search near the boundary of the feasible region to improve the quality of the solution.

[0106] In a specific implementation, at least one target recruitment strategy can be obtained through the method of this embodiment, and the target recruitment strategy includes but is not limited to parameters such as the number of recruits, employment cycle, and employment positions. In order to adapt to market changes and corporate needs, the target recruitment strategy can be continuously optimized based on real-time feedback through an online learning algorithm.

[0107] In this embodiment, the dynamic adjustment of recruitment strategy can be carried out in the following ways: by real-time monitoring and analysis of changes in the recruitment market, changes in corporate demand and feedback on recruitment effects, the recruitment strategy and resource allocation are continuously optimized, including real-time market salary data, industry talent flow trends, competitor recruitment dynamics and internal recruitment effect indicators. Using machine learning algorithms, especially time series analysis and prediction models, the system can identify cyclical changes and sudden events in the recruitment market and predict future talent supply and demand trends. At the same time, through deep learning natural language processing technology, the module can analyze job seekers' resumes and interview feedback to evaluate talent quality and matching in real time. Based on these analysis results, the system uses reinforcement learning algorithms to continuously adjust recruitment strategies, including optimizing job descriptions, adjusting salary plans, changing recruitment channel combinations and modifying interview processes. In addition, the module also integrates budget management functions, and optimizes resource allocation while ensuring recruitment quality through linear programming and dynamic programming algorithms to maximize cost-effectiveness. The system also has an early warning mechanism. When key indicators such as resume collection rate, interview pass rate or acceptance rate of acceptance letters are abnormal, the adjustment process will be automatically triggered and an alarm will be issued to the human resources team. Through this continuous monitoring, analysis, prediction and adjustment cycle, the dynamic adjustment module ensures that the recruitment strategy always remains in the optimal state, can quickly respond to market changes and corporate needs, improve recruitment efficiency and talent quality, while reducing recruitment costs and time cycles.

[0108] It should be noted that, based on the present embodiment, the most suitable recruitment channels and strategy combinations can also be recommended for different positions based on collaborative filtering and content matching algorithms. Using the project-based collaborative filtering algorithm, the system calculates the similarity between positions, constructs a position-feature matrix, and uses cosine similarity for calculation. At the same time, the TF-IDF (TermFrequency-Inverse Document Frequency) algorithm is used to perform text analysis on the position description, extract keywords and calculate weights. The system also quantitatively evaluates the effect of each recruitment channel, considering indicators such as resume delivery volume, resume quality, interview conversion rate, acceptance rate of acceptance letters, recruitment cycle and cost. Based on these data, the system calculates a comprehensive score for each position-channel-strategy combination, and the scoring formula covers the weighted sum of various indicators. When generating recommendations, the system first finds the K historical positions that are most similar to the new position, then summarizes the best channel and strategy combinations for these positions, sorts them according to the comprehensive score, and finally selects the top N best combinations as the recommendation results. Through continuous A / B testing and model parameter updates, the system continuously optimizes the recommendation algorithm to ensure that the most suitable recruitment channel and strategy combination is provided for each position, thereby improving recruitment efficiency and quality.

[0109] It can be understood that, in addition to the above-mentioned contents of this embodiment, the method of this embodiment can also provide an interpretable analysis of recruitment decisions to help human resource managers understand and improve recruitment strategies. First, a decision tree model is used to explain the key factors affecting recruitment success, and the impact of different features on recruitment results is displayed by visualizing the decision tree. At the same time, the importance of each feature is calculated using a random forest or gradient boosting tree model, and a feature importance bar chart is generated. In order to explain a single recruitment decision, the module calculates the contribution of each feature to the predicted result. In addition, the relationship between key features and recruitment results is displayed through a partial dependency graph. In addition, the interpretable analysis also includes a recruitment funnel analysis, which visualizes the conversion rate of each stage; uses a time series decomposition method to analyze the time trend of recruitment demand and success rate; and discovers similar job groups through cluster analysis. All of the above analysis results are integrated into an interactive dashboard, allowing users to select different dimensions and time ranges for data exploration. Finally, the interpretable analysis also integrates a natural language generation function to automatically extract key insights and generate easy-to-understand text descriptions. Through these comprehensive explainability technologies, human resource managers can deeply understand the factors that affect recruitment success and gain insight into the effectiveness of recruitment strategies, so as to make more informed decisions, continuously optimize the recruitment process, and improve overall recruitment efficiency and quality.

[0110] The method of this embodiment includes: stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population; calculating the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result; performing individual selection in the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result to determine a plurality of excellent individuals; performing genetic iteration based on each excellent individual, a multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population; when the iterative population meets the iteration termination condition, determining the target individual according to the fitness value of each individual in the iterative population, and determining the target recruitment strategy based on the target individual. Through the above method, multiple excellent individuals are selected from the initial population through the non-dominated sorting results of the population corresponding to the individual crowding degree and the recruitment demand, and genetic iteration is performed based on multiple excellent individuals and multi-objective fitness functions. When the iterative population meets the termination condition, the target individual is selected based on the individual fitness value, thereby generating a target recruitment strategy, which comprehensively considers multiple factors such as human resource demand planning, labor cost, and personnel replacement, and provides enterprises with data-driven, efficient and accurate long-term recruitment strategies, which significantly improves the efficiency and effectiveness of human resource management and reduces the direct labor cost of enterprises.

[0111] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 3 In the recruitment strategy optimization method, step S40 further includes steps S41 to S43:

[0112] Step S41, generating offspring individuals of each excellent individual according to an adaptive genetic algorithm, and constructing a temporary population based on the offspring individuals of each excellent individual and each excellent individual.

[0113] It should be noted that the excellent individuals are crossovered and mutated by the adaptive algorithm to generate multiple offspring individuals for each excellent individual. In this embodiment, the adaptive genetic algorithm can dynamically adjust the crossover operator and the mutation operator according to the state of the current optimization process (such as the diversity of the population, the convergence, etc.) to generate high-quality offspring individuals. A temporary population can be constructed by the offspring individuals of each excellent individual and multiple excellent individuals.

[0114] In a feasible implementation, step S41 may include steps D11 to D13:

[0115] Step D11, generating offspring individuals of each excellent individual according to an adaptive genetic algorithm.

[0116] Step D12: When there are offspring individuals of excellent individuals that do not meet the recruitment constraints, the offspring individuals that do not meet the recruitment constraints are repaired according to the constraint repair mechanism to obtain repaired offspring individuals.

[0117] Step D13, constructing a temporary population based on the repaired offspring individuals and each excellent individual.

[0118] It should be noted that the feasibility judgment function A constraint check is performed on each offspring individual to determine whether it satisfies the recruitment constraint condition; when there are offspring individuals with excellent individuals that do not satisfy the recruitment constraint condition, the offspring individuals that do not satisfy the recruitment constraint condition are repaired through a constraint repair mechanism to obtain a repaired offspring individual. In this embodiment, the constraint repair mechanism refers to a set constraint repair function, which is used to map an infeasible solution to a feasible domain.

[0119] It can be understood that a temporary population is constructed based on the repaired offspring individuals, the offspring individuals that meet the recruitment constraints without repair, and the excellent individuals.

[0120] Step S42, calculating the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function.

[0121] In a feasible implementation, step S42 may include steps E11 to E13:

[0122] Step E11, calculating the sub-objective function value of each individual in the temporary population under multiple evaluation dimensions according to the predicted recruitment index under each demand dimension.

[0123] Step E12, normalizing the sub-objective function values ​​of each individual in the temporary population under each evaluation dimension to obtain the evaluation function values ​​of each individual in the temporary population under each evaluation dimension.

[0124] Step E13, performing weighted calculation according to the target penalty term, the target weight corresponding to each sub-objective function, and the evaluation function value of each individual in the temporary population under each evaluation dimension, to determine the fitness value of each individual in the temporary population.

[0125] It should be noted that, according to the predicted recruitment indicators under each demand dimension, the sub-objective function values ​​of each individual in the temporary population under multiple evaluation dimensions are calculated. In this embodiment, the sub-objective function values ​​under multiple evaluation dimensions refer to human demand satisfaction f1(x), recruitment cost control f2(x), staff turnover rate f3(x), and recruitment quality f4(x).

[0126] It can be understood that for the temporary population: normalize the sub-objective function value of each individual under each evaluation dimension to obtain the evaluation function value f of each individual under each evaluation dimension i '(x). Use the target weight w corresponding to each sub-objective function i , the target penalty term P(x) and the evaluation function value of each individual under each evaluation dimension are weightedly calculated to obtain the fitness value of each individual F(x)=max{w1f1'(x),w2f2'(x),w3f3'(x),w4f4'(x)}+P(x).

[0127] Step S43, performing individual screening according to the fitness value of each individual in the temporary population, and constructing an iterative population based on the screening result.

[0128] It should be noted that elite individuals with the best fitness value are selected from the temporary population according to the set ratio, the selected elite individuals are retained, and the next generation population is constructed through multiple elite individuals. The next generation population is the iterative population.

[0129] This embodiment generates offspring individuals of each excellent individual according to an adaptive genetic algorithm, constructs a temporary population based on the offspring individuals of each excellent individual and each excellent individual; calculates the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function; performs individual screening according to the fitness value of each individual in the temporary population, and constructs an iterative population based on the screening results. In the above manner, a feasible and effective iterative population can be accurately obtained.

[0130] For example, to help understand the implementation process of the recruitment strategy optimization method obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 4 , Figure 4 A brief flowchart of a recruitment strategy optimization method is provided, specifically:

[0131] The method in this embodiment includes the following steps: 1. Data collection and preprocessing: collect and integrate multi-source information such as recruitment data, personnel turnover data, market salary data, etc., and perform cleaning, standardization and feature engineering to obtain processed multi-source information. 2. Data prediction: Integrate the data in the first step to predict key indicators such as future manpower needs, employee turnover rate and market salary trends, and pass the data to the multi-objective optimization algorithm. 3. Data optimization calculation: Based on the data in the second step, machine learning algorithm technology is used to seek the best balance between multiple goals such as meeting manpower needs, controlling costs, and reducing personnel turnover. 4. Data analysis; based on the results of step 3, the data in step 1, and the data in step 2, recommend the most suitable recruitment channels and strategy combinations for different positions, dynamically adjust feedback and optimize recruitment strategies, and provide explainable analysis of recruitment decisions.

[0132] Through the method of this embodiment, the following beneficial effects can be achieved: 1. Reduce labor costs: In the face of various constraints and target requirements, give a suitable recruitment plan to reduce the labor costs of enterprises. 2. Adaptive learning ability: Continuously optimize recruitment strategies based on real-time feedback to adapt to market changes and dynamic adjustments of enterprise needs. 3. Avoid the limitations of manual planning: Fully consider various practical factors to avoid the limitations of manual capabilities. 4. Multi-objective dynamic balance. 5. Strategy diversity guarantee: By introducing congestion calculation and diversity maintenance mechanism, a series of diverse non-inferior solutions can be generated, providing decision makers with multiple optional recruitment plans.

[0133] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the recruitment strategy optimization method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0134] This application also provides a recruitment strategy optimization device, please refer to Figure 5 , the recruitment strategy optimization device comprises:

[0135] The stratification module 10 is used to stratify the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population.

[0136] The calculation module 20 is used to calculate the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result.

[0137] The selection module 30 is used to select individuals in the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result, and determine a plurality of excellent individuals.

[0138] The iteration module 40 is used to perform genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment index under each demand dimension to obtain an iterative population.

[0139] The processing module 50 is used to determine the target individual according to the fitness value of each individual in the iterative population when the iterative population meets the iteration termination condition, and determine the target recruitment strategy based on the target individual.

[0140] Optionally, the stratification module 10 is further used for:

[0141] The fitness value of each individual in the initial population corresponding to the target recruitment demand is calculated according to the predicted recruitment index under each demand dimension and the multi-objective fitness function; parameter calculation is performed according to the target non-dominated sorting function, the fitness value of each individual in the initial population and the constraint violation degree of each individual in the initial population to determine the dominance parameter of each individual in the initial population; the initial population is stratified according to the dominance parameter of each individual in the initial population to obtain the non-dominated sorting result of the initial population.

[0142] Optionally, the iteration module 40 is further configured to:

[0143] Generate offspring individuals of each excellent individual according to an adaptive genetic algorithm, and construct a temporary population based on the offspring individuals of each excellent individual and each excellent individual; calculate the fitness value of each individual in the temporary population according to the predicted recruitment indicators under each demand dimension and the multi-objective fitness function; perform individual screening according to the fitness value of each individual in the temporary population, and construct an iterative population based on the screening results.

[0144] Optionally, the iteration module 40 is further configured to:

[0145] The offspring individuals of each excellent individual are generated according to the adaptive genetic algorithm; when the offspring individuals of the excellent individuals do not meet the recruitment constraints, the offspring individuals that do not meet the recruitment constraints are repaired according to the constraint repair mechanism to obtain the repaired offspring individuals; a temporary population is constructed according to the repaired offspring individuals and each excellent individual.

[0146] Optionally, the iteration module 40 is further configured to:

[0147] The sub-objective function values ​​of each individual in the temporary population under multiple evaluation dimensions are calculated according to the predicted recruitment indicators under each demand dimension; the sub-objective function values ​​of each individual in the temporary population under each evaluation dimension are normalized to obtain the evaluation function values ​​of each individual in the temporary population under each evaluation dimension; and the fitness value of each individual in the temporary population is determined by weighted calculation based on the target penalty item, the target weight corresponding to each sub-objective function, and the evaluation function value of each individual in the temporary population under each evaluation dimension.

[0148] Optionally, the iteration module 40 is further configured to:

[0149] Data processing is performed on multi-source information of human resources to obtain processed multi-source data; multi-dimensional indicator prediction is performed based on the processed multi-source data to determine the human resource demand indicator, the predicted turnover rate indicator and the salary trend indicator; and the predicted recruitment indicator under each demand dimension is determined based on the human resource demand indicator, the predicted turnover rate indicator and the salary trend indicator.

[0150] Optionally, the stratification module 10 is further used for:

[0151] The problem is encoded based on the recruitment decision parameters to determine multiple decision vectors; the feasibility of each decision vector is judged according to the feasibility judgment function; when each decision vector satisfies the recruitment constraint condition, multiple initial individuals are determined according to each decision vector; and an initial population corresponding to the target recruitment demand is constructed according to each initial individual.

[0152] The recruitment strategy optimization device provided by this application adopts the recruitment strategy optimization method in the above embodiment, which can solve the technical problem of how to formulate an efficient and accurate long-term recruitment strategy while meeting multiple factors such as production demand and labor costs. Compared with the prior art, the beneficial effects of the recruitment strategy optimization device provided by this application are the same as the beneficial effects of the recruitment strategy optimization method provided by the above embodiment, and the other technical features in the recruitment strategy optimization device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0153] The present application provides a recruitment strategy optimization device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the recruitment strategy optimization method in the above-mentioned embodiment one.

[0154] Reference below Figure 6, which shows a schematic diagram of the structure of a recruitment strategy optimization device suitable for implementing the embodiment of the present application. The recruitment strategy optimization device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The recruitment strategy optimization device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0155] like Figure 6 As shown, the recruitment strategy optimization device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the recruitment strategy optimization device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the recruitment strategy optimization device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a recruitment strategy optimization device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0156] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0157] The recruitment strategy optimization device provided by this application adopts the recruitment strategy optimization method in the above embodiment, which can solve the technical problem of how to formulate efficient and accurate long-term recruitment strategies while meeting multiple factors such as production needs and labor costs. Compared with the prior art, the beneficial effects of the recruitment strategy optimization device provided by this application are the same as the beneficial effects of the recruitment strategy optimization method provided by the above embodiment, and the other technical features in the recruitment strategy optimization device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0158] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0159] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0160] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the recruitment strategy optimization method in the above-mentioned embodiment.

[0161] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0162] The above-mentioned computer-readable storage medium may be included in the recruitment strategy optimization device; or it may exist independently without being assembled into the recruitment strategy optimization device.

[0163] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the recruitment strategy optimization device, the recruitment strategy optimization device: stratifies the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population; calculates the individual crowding in each non-dominated layer in the initial population according to the non-dominated sorting result; selects individuals in the initial population according to the individual crowding in each non-dominated layer and the non-dominated sorting result to determine multiple excellent individuals; performs genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment indicators under each demand dimension to obtain an iterative population; when the iterative population meets the iteration termination condition, determines the target individual according to the fitness value of each individual in the iterative population, and determines the target recruitment strategy based on the target individual.

[0164] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0165] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0166] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0167] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned recruitment strategy optimization method, and can solve the technical problem of how to formulate efficient and accurate long-term recruitment strategies while meeting multiple factors such as production needs and labor costs. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the recruitment strategy optimization method provided in the above-mentioned embodiment, and will not be elaborated here.

[0168] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned recruitment strategy optimization method when executed by a processor.

[0169] The computer program product provided by this application can solve the technical problem of how to formulate an efficient and accurate long-term recruitment strategy while meeting multiple factors such as production demand and labor costs. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the recruitment strategy optimization method provided by the above embodiment, and will not be repeated here.

[0170] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A recruitment strategy optimization method, characterized in that: The recruitment strategy optimization method comprises: stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain a non-dominated sorting result of the initial population; Calculate the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result; Selecting individuals from the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result, and determining a plurality of excellent individuals; Genetic iteration is performed based on each excellent individual, multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population; When the iteration population meets the iteration termination condition, a target individual is determined according to the fitness value of each individual in the iteration population, and a target recruitment strategy is determined based on the target individual.

2. The method according to claim 1, characterized in that The step of stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population includes: The fitness value of each individual in the initial population corresponding to the target recruitment demand is calculated based on the predicted recruitment indicators under each demand dimension and the multi-objective fitness function; Calculating parameters according to the target non-dominated sorting function, the fitness value of each individual in the initial population, and the constraint violation degree of each individual in the initial population to determine the dominance parameter of each individual in the initial population; The initial population is stratified according to the dominance parameters of each individual in the initial population to obtain a non-dominated sorting result of the initial population.

3. The method according to claim 1, characterized in that The step of performing genetic iteration based on each excellent individual, the multi-objective fitness function and the predicted recruitment index under each demand dimension to obtain an iterative population includes: Generate offspring individuals of each excellent individual according to the adaptive genetic algorithm, and construct a temporary population based on the offspring individuals of each excellent individual and each excellent individual; Calculate the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function; Individuals are screened according to the fitness values ​​of the individuals in the temporary population, and an iterative population is constructed based on the screening results.

4. The method according to claim 3, characterized in that The steps of generating offspring individuals of each excellent individual according to the adaptive genetic algorithm and constructing a temporary population based on the offspring individuals of each excellent individual and each excellent individual include: Generate offspring individuals of each excellent individual according to the adaptive genetic algorithm; When there are offspring individuals with excellent individuals that do not meet the recruitment constraints, the offspring individuals that do not meet the recruitment constraints are repaired according to the constraint repair mechanism to obtain the repaired offspring individuals; A temporary population is constructed based on the repaired offspring individuals and excellent individuals.

5. The method according to claim 3, characterized in that The step of calculating the fitness value of each individual in the temporary population according to the predicted recruitment index under each demand dimension and the multi-objective fitness function comprises: Calculate the sub-objective function value of each individual in the temporary population under multiple evaluation dimensions according to the predicted recruitment index under each demand dimension; Normalizing the sub-objective function values ​​of each individual in the temporary population under each evaluation dimension to obtain the evaluation function values ​​of each individual in the temporary population under each evaluation dimension; A weighted calculation is performed according to the target penalty term, the target weight corresponding to each sub-objective function, and the evaluation function value of each individual in the temporary population under each evaluation dimension to determine the fitness value of each individual in the temporary population.

6. The method according to any one of claims 1 to 5, characterized in that Before the step of performing genetic iteration based on each excellent individual, multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population, the step further includes: Perform data processing on multi-source information of human resources to obtain processed multi-source data; Perform multi-dimensional indicator forecasting based on the multi-source data processed, determine the human resource demand indicator, the predicted turnover rate indicator and the salary trend indicator; The predicted recruitment indicators under each demand dimension are determined according to the manpower demand indicator, the predicted churn rate indicator and the salary trend indicator.

7. The method according to any one of claims 1 to 5, characterized in that Before the step of stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain the non-dominated sorting result of the initial population, the method further includes: Encode the problem based on the recruitment decision parameters and determine multiple decision vectors; Conduct feasibility judgment on each decision vector according to the feasibility judgment function; When each decision vector satisfies the recruitment constraint condition, a plurality of initial individuals are determined according to each decision vector; Construct an initial population corresponding to the target recruitment needs based on each initial individual.

8. A recruitment strategy optimization device, characterized in that: The recruitment strategy optimization device comprises: A stratification module, used for stratifying the initial population corresponding to the target recruitment demand according to the target non-dominated sorting function to obtain a non-dominated sorting result of the initial population; A calculation module, used for calculating the individual crowding degree in each non-dominated layer in the initial population according to the non-dominated sorting result; A selection module, configured to select individuals from the initial population according to the individual crowding degree in each non-dominated layer and the non-dominated sorting result, and determine a plurality of excellent individuals; The iteration module is used to perform genetic iteration based on each excellent individual, multi-objective fitness function and predicted recruitment indicators under each demand dimension to obtain an iterative population; A processing module is used to determine a target individual according to the fitness value of each individual in the iterative population when the iterative population meets the iteration termination condition, and determine a target recruitment strategy based on the target individual.

9. A recruitment strategy optimization device, characterized in that: The recruitment strategy optimization device includes: a memory, a processor, and a recruitment strategy optimization program stored in the memory and executable on the processor, wherein the recruitment strategy optimization program is configured to implement the recruitment strategy optimization method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a recruitment strategy optimization program, and when the recruitment strategy optimization program is executed by the processor, the recruitment strategy optimization method according to any one of claims 1 to 7 is implemented.

Images