Employee performance evaluation method and system based on multi-dimensional data

Abstract

The invention discloses an employee performance evaluation method and system based on multi-dimensional data, and belongs to the technical field of enterprise talent intelligent management, and the method comprises the steps of multi-dimensional employee data preparation, employee behavior modeling, employee performance fluctuation prediction, employee performance attribution analysis and employee performance evaluation. According to the method, employee behavior modeling is carried out by adopting a graph construction method combining multi-dimensional cooperation characteristics and behavior characteristics, and overall modeling and dynamic sensing of real positions, interaction strength and multi-dimensional behavior states of individual employees in an organization cooperation network are realized; employee performance fluctuation prediction is carried out by using a graph convolution bidirectional long-short term network model optimized by a joint loss function, and employee individual performance and dynamic evolution of mutual influence in an organization cooperation network are comprehensively modeled; the employee performance attribution analysis method based on anti-fact simulation is adopted to perform attribution analysis, and the influence of the key cooperation relation on employee performance change is quantitatively evaluated by simulating the hypothesis situation.

Description

Technical Field

[0001] This invention belongs to the field of intelligent talent management technology, specifically referring to an employee performance evaluation method and system based on multi-dimensional data. Background Technology

[0002] Employee performance evaluation based on multi-dimensional data utilizes artificial intelligence technology to analyze comprehensive employee data from multiple aspects, enabling a comprehensive and dynamic analysis and evaluation of employee work status and performance. It aims to help companies fully understand the true performance of their employees, identify potential performance fluctuations and risks, and clarify influencing factors, thereby supporting precise performance management and talent development decisions, and ultimately improving overall team effectiveness and organizational competitiveness.

[0003] However, existing employee performance evaluation processes suffer from several technical problems: a singular evaluation perspective that ignores the impact of collaboration among employees, making it difficult to identify potential bottlenecks within the organization; reliance on historical individual performance data that overlooks the influence of complex collaborative relationships among employees and fails to capture the potential patterns of dynamic changes in employee performance over time, resulting in poor accuracy in performance evaluation; and difficulty in accurately identifying the specific reasons for fluctuations in employee performance and in clearly distinguishing between individual factors and the impact of team collaboration. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides a multi-dimensional data-based employee performance evaluation method and system. Addressing the technical problems of current employee performance evaluation processes, such as a single evaluation perspective that ignores the impact of collaboration among employees, leading to difficulties in identifying potential bottlenecks within the organization, this solution creatively employs a graph construction method combining multi-dimensional collaborative features and behavioral characteristics to model employee behavior. This achieves holistic modeling and dynamic perception of an individual employee's true position, interaction intensity, and multi-dimensional behavioral state within the organizational collaboration network. Furthermore, it addresses the technical issues of existing employee performance evaluation methods relying on historical individual performance data, ignoring the impact of complex collaborative relationships between employees on employee performance, and failing to capture the potential patterns of dynamic changes in employee performance over time, resulting in poor accuracy in employee performance evaluation. This solution creatively employs a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations. It comprehensively models the dynamic evolution of individual employee performance and its mutual influence within the organizational collaboration network, thereby revealing the structural dependencies and temporal evolution patterns behind performance fluctuations and improving the accuracy and timeliness of employee performance evaluation. Addressing the technical challenges of accurately identifying the specific causes of employee performance fluctuations and clearly distinguishing between individual factors and team collaboration in existing employee performance evaluation processes, this solution creatively uses a counterfactual simulation-based employee performance attribution analysis method. This enables quantitative assessment of the impact of key collaborative relationships on employee performance changes through simulated hypothetical scenarios, effectively revealing causal relationships and assisting managers in identifying the root drivers of performance fluctuations and optimizing team performance.

[0005] The technical solution adopted by this invention is as follows: The employee performance evaluation method based on multi-dimensional data provided by this invention includes the following steps:

[0006] Step S1: Preparation of multi-dimensional employee data;

[0007] Step S2: Employee behavior modeling;

[0008] Step S3: Predicting employee performance fluctuations;

[0009] Step S4: Employee performance attribution analysis;

[0010] Step S5: Employee performance evaluation.

[0011] Further, in step S1, the preparation of multidimensional employee data specifically involves collecting employee file data, historical performance data, attendance records, and work task participation records from the enterprise management system, collecting email records from the office collaboration system, and collecting chat interaction data by calling the WeChat API to obtain multidimensional employee data. Then, the multidimensional employee data is standardized to generate multidimensional employee standard data.

[0012] The employee file data includes employee ID, job information, and department.

[0013] Further, in step S2, the employee behavior modeling is used to construct a graph structure reflecting the collaborative relationships among employees. Specifically, it employs a graph construction method that combines multi-dimensional collaborative features and behavioral features to perform employee behavior modeling and obtain an employee behavior graph, including the following steps:

[0014] Step S21: Establish employee nodes. Specifically, extract all employees who need to be evaluated from the multidimensional employee standard data, treat each employee as an employee node, and use employee ID, job information, and department as node attributes.

[0015] Step S22: Establish a collaborative relationship edge. Specifically, based on the multi-dimensional employee standard data, determine whether there is a collaborative relationship between each pair of employees. If there is a collaborative relationship, establish a collaborative relationship edge between the corresponding two employee nodes.

[0016] Step S23: Set edge weights, specifically by calculating the number of times the task is jointly participated in, the frequency of communication, and the degree of task dependence;

[0017] Step S24: Construct the employee node feature matrix. Specifically, extract work efficiency features, collaboration ability features, and stability features from multidimensional standard employee data, normalize them, and concatenate them to obtain a multidimensional collaboration feature vector. By using the multidimensional collaboration feature vector as node features, the employee node feature matrix is ​​obtained.

[0018] Step S25: Generate employee behavior map, specifically by executing steps S21 to S24 to construct a heterogeneous graph structure and generate employee behavior map.

[0019] Further, in step S3, the employee performance fluctuation prediction, used to uncover potential patterns in employee performance fluctuations, specifically employs a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations, obtaining employee performance fluctuation prediction data, including the following steps:

[0020] Step S31: Construct a graph convolutional bidirectional long short-term network model. Specifically, the information of employee nodes in the employee behavior graph is aggregated through a standard graph convolutional network to obtain the embedded features of employee nodes. Then, the embedded features of employee nodes are temporally modeled through a standard bidirectional long short-term network to obtain the prediction results.

[0021] Step S32: Construct a joint loss function to build a loss function that takes into account both prediction accuracy and graph structure sensitivity. Specifically, it combines the L2 loss function and the graph structure loss function to construct a joint loss function.

[0022] Step S33: Employee performance time series prediction, specifically, by optimizing and training the graph convolutional bidirectional long short term network model through a joint loss function to obtain an employee performance fluctuation prediction model, and then using the employee performance fluctuation prediction model to predict employee performance fluctuations to obtain employee performance fluctuation prediction data, wherein the employee performance fluctuation prediction data includes the predicted score of employee performance and the probability of fluctuation risk.

[0023] Further, in step S4, the employee performance attribution analysis specifically employs a counterfactual simulation-based employee performance attribution analysis method to perform attribution analysis and obtain attribution analysis indicators, including the following steps:

[0024] Step S41: Target employee identification, specifically, based on the employee performance fluctuation prediction data, select employee nodes with a fluctuation risk probability greater than 0.65 as candidate employees; based on the historical performance data in the multidimensional employee standard data, calculate the performance fluctuation range of each employee node; when the performance fluctuation range of a candidate employee node exceeds the abnormal fluctuation threshold, the candidate employee node is designated as the target employee node.

[0025] Step S42: Counterfactual simulation, specifically, each time only one neighboring employee node of the target employee node is set to the ideal state, while keeping other employee nodes unchanged, to construct a counterfactual employee behavior graph, which is then input into the employee performance fluctuation prediction model for prediction, to obtain counterfactual prediction data, and then the difference between the employee performance fluctuation prediction data and the counterfactual prediction data is calculated to obtain the counterfactual bias; the attribution contribution is calculated by combining the edge weights of the counterfactual employee behavior graph and the counterfactual bias; neighboring employee nodes whose counterfactual bias is greater than the bias threshold and whose attribution contribution is greater than the contribution threshold are marked as key influencing employee nodes;

[0026] Step S43: Attribution results are generated. Specifically, by executing steps S41 and S42, attribution analysis indicators for each target employee node are obtained. The attribution analysis indicators include key influencing employee nodes, counterfactual bias of each neighbor node, and attribution contribution.

[0027] Furthermore, in step S5, the employee performance evaluation specifically involves generating an employee performance evaluation report by summarizing the predicted data of employee performance fluctuations and the attribution analysis indicators for each target employee node.

[0028] The employee performance evaluation system based on multi-dimensional data provided by this invention includes: a multi-dimensional employee data preparation module, an employee behavior modeling module, an employee performance fluctuation prediction module, an employee performance attribution analysis module, and an employee performance evaluation module.

[0029] The multidimensional employee data preparation module is used for multidimensional employee data preparation. Through multidimensional employee data preparation, multidimensional employee standard data is obtained, and the multidimensional employee standard data is sent to the employee behavior modeling module and the employee performance attribution analysis module.

[0030] The employee behavior modeling module is used for employee behavior modeling. Through employee behavior modeling, an employee behavior map is obtained, and the employee behavior map is sent to the employee performance fluctuation prediction module and the employee performance attribution analysis module.

[0031] The employee performance fluctuation prediction module is used to predict employee performance fluctuations. Through the prediction of employee performance fluctuations, it obtains employee performance fluctuation prediction data and sends the employee performance fluctuation prediction data to the employee performance attribution analysis module and the employee performance evaluation module.

[0032] The employee performance attribution analysis module is used for employee performance attribution analysis. Through employee performance attribution analysis, attribution analysis indicators are obtained, and the attribution analysis indicators are sent to the employee performance evaluation module.

[0033] The employee performance evaluation module is used to evaluate employee performance and generate employee performance evaluation reports.

[0034] The beneficial effects achieved by the present invention using the above solution are as follows:

[0035] (1) In response to the technical problem that the existing employee performance evaluation process has a single evaluation perspective and ignores the impact of collaboration among employees, making it difficult to find potential bottlenecks in the enterprise organization, this solution creatively adopts a graph construction method that combines multi-dimensional collaboration features and behavioral features to model employee behavior, and realizes the overall modeling and dynamic perception of the real position, interaction intensity and multi-dimensional behavioral status of individual employees in the organizational collaboration network.

[0036] (2) In view of the technical problems in the existing employee performance evaluation process, the traditional evaluation method relies on individual historical performance data, ignores the impact of complex collaborative relationships between employees on employee performance, and is difficult to capture the potential patterns of dynamic changes in employee performance over time, resulting in poor accuracy of employee performance evaluation. This solution creatively adopts a graph convolutional bidirectional long short-term network model with joint loss function optimization to predict employee performance fluctuations, comprehensively models the dynamic evolution of individual employee performance and its mutual influence in the organizational collaborative network, thereby revealing the structural dependence and time evolution patterns behind performance fluctuations, and improving the accuracy and timeliness of employee performance evaluation.

[0037] (3) In response to the technical problems that exist in the existing employee performance evaluation process, such as the difficulty in accurately identifying the specific reasons for employee performance fluctuations and the inability to clearly distinguish between individual factors and the impact of team collaboration, this solution creatively adopts an employee performance attribution analysis method based on counterfactual simulation. This method enables quantitative assessment of the impact of key collaborative relationships on changes in employee performance through simulated hypothetical scenarios, effectively revealing causal relationships and thus assisting managers in identifying the fundamental driving factors of performance fluctuations and optimizing team performance. Attached Figure Description

[0038] Figure 1 A flowchart illustrating the employee performance evaluation method based on multi-dimensional data provided by this invention;

[0039] Figure 2 A schematic diagram of the employee performance evaluation system based on multi-dimensional data provided by the present invention;

[0040] Figure 3 This is a flowchart illustrating step S2;

[0041] Figure 4 This is a flowchart illustrating step S4.

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0044] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0045] Example 1, see Figure 1 The present invention provides an employee performance evaluation method based on multi-dimensional data, which includes the following steps:

[0046] Step S1: Preparation of multi-dimensional employee data;

[0047] Step S2: Employee behavior modeling;

[0048] Step S3: Predicting employee performance fluctuations;

[0049] Step S4: Employee performance attribution analysis;

[0050] Step S5: Employee performance evaluation.

[0051] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the preparation of multi-dimensional employee data specifically involves collecting employee file data, historical performance data, attendance records, and work task participation records from the enterprise management system, collecting email records from the office collaboration system, and collecting chat interaction data by calling the WeChat API to obtain multi-dimensional employee data. Then, the multi-dimensional employee data is standardized to generate multi-dimensional employee standard data.

[0052] The employee file data includes employee ID, job information, and department.

[0053] The standardization process includes data type conversion, unified field naming, outlier removal, duplicate value removal, and timestamp alignment. Specifically, data type conversion involves converting text, numerical, and time-based data in the multidimensional employee data into a structured data format. Unified field naming involves standardizing field names from different sources within the multidimensional employee data using predefined field mapping rules. Timestamp alignment involves aligning time-based data in the multidimensional employee data according to a unified timeline.

[0054] Example 3, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S2, the employee behavior modeling is used to construct a graph structure reflecting the collaborative relationships between employees. Specifically, it adopts a graph construction method that combines multi-dimensional collaborative features and behavioral features to perform employee behavior modeling and obtain an employee behavior graph, including the following steps:

[0055] Step S21: Establish employee nodes. Specifically, extract all employees who need to be evaluated from the multidimensional employee standard data, treat each employee as an employee node, and use employee ID, job information, and department as node attributes.

[0056] Step S22: Establish a collaborative relationship edge. Specifically, based on the multi-dimensional employee standard data, determine whether there is a collaborative relationship between each pair of employees. If there is a collaborative relationship, establish a collaborative relationship edge between the corresponding two employee nodes.

[0057] The collaborative relationships include shared task relationships, upstream and downstream task dependencies, and management relationships;

[0058] The aforementioned shared task relationship specifically refers to two employees jointly participating in a work task;

[0059] The upstream and downstream dependency relationship of the task refers to the upstream and downstream dependency relationship between the work tasks participated in by the two employees;

[0060] The management relationship specifically refers to a superior-subordinate management relationship between two employees.

[0061] Step S23: Set edge weights, specifically by calculating the number of times tasks are jointly participated in, the frequency of communication, and the degree of task dependence. The calculation formula is as follows:

[0062] ;

[0063] In the formula, w ij This represents the edge weight between the i-th employee node and the j-th employee node, where i is the first index of the employee node (representing the employee), and j is the second index of the employee node (not equal to the first index). It is the weight of the number of times the task was participated in, F. proj It refers to the number of times we jointly participate in a task. It is the communication frequency weight, F comm It's the frequency of communication. It is the task dependency weight, F dep It is task dependency;

[0064] The formula for calculating the communication frequency is:

[0065] ;

[0066] In the formula, It represents the total number of emails exchanged between the i-th employee and the j-th employee. It represents the number of interactions between the i-th and j-th employees on WeChat Work. It is the number of days that the i-th employee and the j-th employee communicate at least once;

[0067] The formula for calculating the task dependency is:

[0068] ;

[0069] In the formula, It represents the number of tasks in the work tasks participated in by the i-th employee that directly depend on the work tasks participated in by the j-th employee. It represents the total number of work tasks participated in by the i-th employee;

[0070] Step S24: Construct the employee node feature matrix. Specifically, extract work efficiency features, collaboration ability features, and stability features from multidimensional standard employee data, normalize them, and concatenate them to obtain a multidimensional collaboration feature vector. By using the multidimensional collaboration feature vector as node features, the employee node feature matrix is ​​obtained.

[0071] The formula for calculating the work efficiency characteristic is as follows:

[0072] ;

[0073] In the formula, F delay It is a characteristic of work efficiency, specifically task latency rate, U i Let be the number of tasks for the i-th employee, and u be the index of the tasks that the i-th employee participates in. It is an indicator function that takes the value 1 when a certain condition is met, and 0 otherwise. It is the completion time of the u-th task. It is the deadline for the u-th task;

[0074] The formula for calculating the collaborative capability characteristic is as follows:

[0075] ;

[0076] In the formula, F collab It is a collaborative capability characteristic, specifically the number of collaborative tasks, Num u It represents the number of employees involved in the u-th task;

[0077] The formula for calculating the stability characteristic is as follows:

[0078] ;

[0079] In the formula, F absent It is a stability characteristic, specifically the frequency of absences. It is the number of days the i-th employee is absent. It is the total number of days that the i-th employee should attend;

[0080] Step S25: Generate employee behavior graph. Specifically, by executing steps S21 to S24, a heterogeneous graph structure is constructed to generate the employee behavior graph. The calculation formula is as follows:

[0081] ;

[0082] In the formula, G is the employee behavior graph, V is the set of employee nodes, E is the edge set, W is the edge weight matrix, and X is the employee node feature matrix.

[0083] By performing the above operations, this solution addresses the technical problem that existing employee performance evaluation processes suffer from a single evaluation perspective, neglecting the impact of collaboration among employees and making it difficult to identify potential bottlenecks in the organization. It creatively adopts a graph construction method that combines multi-dimensional collaborative features and behavioral features to model employee behavior, achieving overall modeling and dynamic perception of the actual location, interaction intensity, and multi-dimensional behavioral status of individual employees in the organizational collaboration network.

[0084] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S3, the employee performance fluctuation prediction is used to mine potential patterns of employee performance fluctuations. Specifically, it uses a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations and obtain employee performance fluctuation prediction data. This includes the following steps:

[0085] Step S31: Construct a graph convolutional bidirectional long short-term network model. Specifically, the information of employee nodes in the employee behavior graph is aggregated through a standard graph convolutional network to obtain the embedded features of employee nodes. Then, the embedded features of employee nodes are temporally modeled through a standard bidirectional long short-term network to obtain the prediction results.

[0086] Step S32: Construct a joint loss function to balance prediction accuracy and graph structure sensitivity. Specifically, this involves combining the L2 loss function and the graph structure loss function to construct a joint loss function, calculated as follows:

[0087] ;

[0088] ;

[0089] ;

[0090] In the formula, L total It is the joint loss function, where 'a' is the L2 loss weight, and L... one It is the L2 loss function, b is the graph structure loss weight, L two This is the graph structure loss function, where N is the number of employee nodes. It is the predicted performance value of the i-th employee. It is the actual performance value of the i-th employee;

[0091] Step S33: Employee performance time series prediction, specifically, by optimizing and training the graph convolutional bidirectional long short term network model through the joint loss function to obtain the employee performance fluctuation prediction model, and then using the employee performance fluctuation prediction model to predict the employee performance fluctuation to obtain employee performance fluctuation prediction data. The employee performance fluctuation prediction data includes the predicted score of employee performance and the probability of fluctuation risk.

[0092] By performing the above operations, this solution addresses the technical problems in existing employee performance evaluation processes. Traditional evaluation methods rely on individual historical performance data, neglecting the impact of complex collaborative relationships among employees on employee performance, and struggle to capture the potential patterns of dynamic changes in employee performance over time, leading to poor accuracy in employee performance evaluation. This solution creatively employs a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations. It comprehensively models the dynamic evolution of individual employee performance and its mutual influence within the organizational collaborative network, thereby revealing the structural dependencies and temporal evolution patterns behind performance fluctuations and improving the accuracy and timeliness of employee performance evaluation.

[0093] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the employee performance attribution analysis specifically adopts an employee performance attribution analysis method based on counterfactual simulation to perform attribution analysis and obtain attribution analysis indicators, including the following steps:

[0094] Step S41: Target employee identification, specifically, based on the employee performance fluctuation prediction data, select employee nodes with a fluctuation risk probability greater than 0.65 as candidate employees; based on the historical performance data in the multidimensional employee standard data, calculate the performance fluctuation range of each employee node; when the performance fluctuation range of a candidate employee node exceeds the abnormal fluctuation threshold, the candidate employee node is designated as the target employee node.

[0095] The performance fluctuation range is specifically the standard deviation of an employee's performance score within a quarter, which is used to measure the degree of fluctuation in employee performance within a quarter.

[0096] Preferably, the formula for calculating the abnormal fluctuation threshold is:

[0097] ;

[0098] In the formula, It is the threshold for abnormal fluctuations. It is the average fluctuation range of all employee performance. It is the standard deviation, specifically the standard deviation of the performance fluctuation of all employees;

[0099] Step S42: Counterfactual simulation, specifically, each time only one neighboring employee node of the target employee node is set to the ideal state, while keeping other employee nodes unchanged, to construct a counterfactual employee behavior graph, which is then input into the employee performance fluctuation prediction model for prediction, to obtain counterfactual prediction data, and then the difference between the employee performance fluctuation prediction data and the counterfactual prediction data is calculated to obtain the counterfactual bias; the attribution contribution is calculated by combining the edge weights of the counterfactual employee behavior graph and the counterfactual bias; neighboring employee nodes whose counterfactual bias is greater than the bias threshold and whose attribution contribution is greater than the contribution threshold are marked as key influencing employee nodes;

[0100] The formula for calculating the counterfactual bias is:

[0101] ;

[0102] In the formula, This is a counterfactual bias; A represents the weighting of employee performance. B represents employee performance deviation, and B represents volatility risk weighting. It is volatility risk bias;

[0103] The formula for calculating the attribution contribution is as follows:

[0104] ;

[0105] In the formula, g(k) is the attribution contribution of the k-th neighboring employee node, k is the first index of the neighboring employee node, and w k,target W is the edge weight between the k-th neighboring employee node and the target employee node, where N(target) is the set of neighboring employee nodes of the target employee node. m,target It is the edge weight between the m-th neighboring employee node and the target employee node;

[0106] Preferably, the deviation threshold is set to 0.2, and the contribution threshold is set to 0.3;

[0107] Step S43: Attribution results are generated. Specifically, by executing steps S41 and S42, attribution analysis indicators for each target employee node are obtained. The attribution analysis indicators include key influencing employee nodes, counterfactual bias of each neighbor node, and attribution contribution.

[0108] By performing the above operations, this solution addresses the technical challenges of accurately identifying the specific reasons for employee performance fluctuations and clearly distinguishing between individual factors and the impact of teamwork in existing employee performance evaluation processes. It creatively employs a counterfactual simulation-based employee performance attribution analysis method to conduct attribution analysis. This method enables the quantitative assessment of the impact of key collaborative relationships on changes in employee performance through simulated hypothetical scenarios, effectively revealing causal relationships and thus assisting managers in identifying the root drivers of performance fluctuations and optimizing team performance.

[0109] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the employee performance evaluation specifically involves generating an employee performance evaluation report by summarizing the predicted data of employee performance fluctuations and the attribution analysis indicators of each target employee node.

[0110] Example 7, see Figure 2 Based on the above embodiments, the employee performance evaluation system based on multi-dimensional data provided by the present invention includes: a multi-dimensional employee data preparation module, an employee behavior modeling module, an employee performance fluctuation prediction module, an employee performance attribution analysis module, and an employee performance evaluation module.

[0111] The multidimensional employee data preparation module is used for multidimensional employee data preparation. Through multidimensional employee data preparation, multidimensional employee standard data is obtained, and the multidimensional employee standard data is sent to the employee behavior modeling module and the employee performance attribution analysis module.

[0112] The employee behavior modeling module is used for employee behavior modeling. Through employee behavior modeling, an employee behavior map is obtained, and the employee behavior map is sent to the employee performance fluctuation prediction module and the employee performance attribution analysis module.

[0113] The employee performance fluctuation prediction module is used to predict employee performance fluctuations. Through the prediction of employee performance fluctuations, it obtains employee performance fluctuation prediction data and sends the employee performance fluctuation prediction data to the employee performance attribution analysis module and the employee performance evaluation module.

[0114] The employee performance attribution analysis module is used for employee performance attribution analysis. Through employee performance attribution analysis, attribution analysis indicators are obtained, and the attribution analysis indicators are sent to the employee performance evaluation module.

[0115] The employee performance evaluation module is used to evaluate employee performance and generate employee performance evaluation reports.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0117] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0118] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An employee performance evaluation method based on multi-dimensional data, characterized by: The method includes the following steps: Step S1: Preparation of multi-dimensional employee data; Step S2: Employee behavior modeling, specifically, adopts a graph construction method that combines multi-dimensional collaborative features and behavioral features to model employee behavior and obtain an employee behavior graph, including the following steps: Step S21: Establish employee nodes; Step S22: Establish collaborative relationship edges; Step S23: Set edge weights; Step S24: Construct employee node feature matrix; Step S25: Generate employee behavior graph; Step S3: Employee performance fluctuation prediction. Specifically, this involves using a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations and obtain employee performance fluctuation prediction data. This includes the following steps: Step S31: Constructing a graph convolutional bidirectional long short-term network model; Step S32: Constructing a joint loss function; Step S33: Employee performance time series prediction. Step S4: Employee performance attribution analysis, specifically using a counterfactual simulation-based employee performance attribution analysis method to perform attribution analysis and obtain attribution analysis indicators, including the following steps: Step S41: Target employee identification; Step S42: Counterfactual simulation; Step S43: Attribution result generation; Step S5: Employee performance evaluation.

2. The employee performance evaluation method based on multi-dimensional data according to claim 1, characterized in that: In step S2, the employee behavior modeling is used to construct a graph structure reflecting the collaborative relationships among employees. Specifically, it employs a graph construction method that combines multi-dimensional collaborative features and behavioral features to model employee behavior and obtain an employee behavior graph, including the following steps: Step S21: Establish employee nodes. Specifically, extract all employees who need to be evaluated from the multidimensional employee standard data, treat each employee as an employee node, and use employee ID, job information, and department as node attributes. Step S22: Establish a collaborative relationship edge. Specifically, based on the multi-dimensional employee standard data, determine whether there is a collaborative relationship between each pair of employees. If there is a collaborative relationship, establish a collaborative relationship edge between the corresponding two employee nodes. Step S23: Set edge weights, specifically by calculating the number of times the task is jointly participated in, the frequency of communication, and the degree of task dependence; Step S24: Construct the employee node feature matrix. Specifically, extract work efficiency features, collaboration ability features, and stability features from multidimensional standard employee data, normalize them, and concatenate them to obtain a multidimensional collaboration feature vector. By using the multidimensional collaboration feature vector as node features, the employee node feature matrix is ​​obtained. Step S25: Generate employee behavior map, specifically by executing steps S21 to S24 to construct a heterogeneous graph structure and generate employee behavior map.

3. The employee performance evaluation method based on multi-dimensional data according to claim 2, characterized in that: In step S3, the employee performance fluctuation prediction is used to uncover potential patterns in employee performance fluctuations. Specifically, it employs a graph convolutional bidirectional long short-term network model optimized with a joint loss function to predict employee performance fluctuations and obtain employee performance fluctuation prediction data. This includes the following steps: Step S31: Construct a graph convolutional bidirectional long short-term network model. Specifically, the information of employee nodes in the employee behavior graph is aggregated through a standard graph convolutional network to obtain the embedded features of employee nodes. Then, the embedded features of employee nodes are temporally modeled through a standard bidirectional long short-term network to obtain the prediction results. Step S32: Construct a joint loss function to build a loss function that takes into account both prediction accuracy and graph structure sensitivity. Specifically, it combines the L2 loss function and the graph structure loss function to construct a joint loss function. Step S33: Employee performance time series prediction, specifically, by optimizing and training the graph convolutional bidirectional long short term network model through a joint loss function to obtain an employee performance fluctuation prediction model, and then using the employee performance fluctuation prediction model to predict employee performance fluctuations to obtain employee performance fluctuation prediction data, wherein the employee performance fluctuation prediction data includes the predicted score of employee performance and the probability of fluctuation risk.

4. The employee performance evaluation method based on multi-dimensional data according to claim 3, characterized in that: In step S4, the employee performance attribution analysis specifically employs a counterfactual simulation-based employee performance attribution analysis method to perform attribution analysis and obtain attribution analysis indicators, including the following steps: Step S41: Target employee identification, specifically, based on the employee performance fluctuation prediction data, select employee nodes with a fluctuation risk probability greater than 0.65 as candidate employees; based on the historical performance data in the multidimensional employee standard data, calculate the performance fluctuation range of each employee node; when the performance fluctuation range of a candidate employee node exceeds the abnormal fluctuation threshold, the candidate employee node is designated as the target employee node. Step S42: Counterfactual simulation, specifically, each time only one neighboring employee node of the target employee node is set to the ideal state, while keeping other employee nodes unchanged, to construct a counterfactual employee behavior graph, which is then input into the employee performance fluctuation prediction model for prediction, to obtain counterfactual prediction data, and then the difference between the employee performance fluctuation prediction data and the counterfactual prediction data is calculated to obtain the counterfactual bias; the attribution contribution is calculated by combining the edge weights of the counterfactual employee behavior graph and the counterfactual bias; neighboring employee nodes whose counterfactual bias is greater than the bias threshold and whose attribution contribution is greater than the contribution threshold are marked as key influencing employee nodes; Step S43: Attribution results are generated. Specifically, by executing steps S41 and S42, attribution analysis indicators for each target employee node are obtained. The attribution analysis indicators include key influencing employee nodes, counterfactual bias of each neighbor node, and attribution contribution.

5. The employee performance evaluation method based on multi-dimensional data according to claim 4, characterized in that: In step S5, the employee performance evaluation specifically involves generating an employee performance evaluation report by summarizing employee performance fluctuation prediction data and attribution analysis indicators for each target employee node.

6. The employee performance evaluation method based on multi-dimensional data according to claim 5, characterized in that: In step S1, the preparation of multidimensional employee data specifically involves collecting employee file data, historical performance data, attendance records, and work task participation records from the enterprise management system, as well as collecting email records from the office collaboration system and collecting chat interaction data by calling the WeChat API to obtain multidimensional employee data. Then, the multidimensional employee data is standardized to generate multidimensional employee standard data. The employee file data includes employee ID, job information, and department.

7. A system for evaluating employee performance based on multi-dimensional data, used to implement the employee performance evaluation method based on multi-dimensional data as described in any one of claims 1-6, characterized in that: It includes modules for multidimensional employee data preparation, employee behavior modeling, employee performance fluctuation prediction, employee performance attribution analysis, and employee performance evaluation.

8. The employee performance evaluation method system based on multi-dimensional data according to claim 7, characterized in that: The multidimensional employee data preparation module is used for multidimensional employee data preparation. Through multidimensional employee data preparation, multidimensional employee standard data is obtained, and the multidimensional employee standard data is sent to the employee behavior modeling module and the employee performance attribution analysis module. The employee behavior modeling module is used for employee behavior modeling. Through employee behavior modeling, an employee behavior map is obtained, and the employee behavior map is sent to the employee performance fluctuation prediction module and the employee performance attribution analysis module. The employee performance fluctuation prediction module is used to predict employee performance fluctuations. Through the prediction of employee performance fluctuations, it obtains employee performance fluctuation prediction data and sends the employee performance fluctuation prediction data to the employee performance attribution analysis module and the employee performance evaluation module. The employee performance attribution analysis module is used for employee performance attribution analysis. Through employee performance attribution analysis, attribution analysis indicators are obtained, and the attribution analysis indicators are sent to the employee performance evaluation module. The employee performance evaluation module is used to evaluate employee performance and generate employee performance evaluation reports.

Images