An employee performance data automatic statistical analysis method, system, device and medium
By using employee profile feature vector matching target word embedding model and deep learning scoring, the problem of low efficiency in traditional employee performance analysis has been solved. It has achieved accurate performance scoring and trend prediction for diverse job types and career stages, improving the comprehensiveness and foresight of performance analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 张建刚
- Filing Date
- 2026-02-26
- Publication Date
- 2026-05-29
Smart Images
Figure CN122114733A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of performance statistical analysis, and in particular relates to a method, system, equipment and medium for automatic statistical analysis of employee performance data. Background Technology
[0002] Against the backdrop of deepening digital transformation and upgrading human resource management needs, the core value of employee performance statistical analysis lies in objectively presenting employee performance and predicting performance trends through the mining and analysis of employee performance data, thereby providing data support for human resource decision-making. As enterprises expand, job divisions become more refined, and career development systems improve, employee performance data exhibits multi-source and heterogeneous characteristics. It includes not only quantitative data but also unstructured text data such as performance evaluations, work summaries, and collaborative feedback. Simultaneously, the employee population exhibits diverse job types and career stages. Traditional performance statistical analysis models are no longer adequate for the management needs of the new era, making the evolution of performance statistical analysis towards automation and intelligence an inevitable trend in the industry.
[0003] Traditional employee performance statistical analysis relies primarily on manual statistics, with HR professionals compiling quantitative data and summarizing textual evaluations. This approach is not only inefficient but also struggles with in-depth data mining from massive datasets. It fails to coordinate the analysis of quantitative and textual data, and it is even more difficult to conduct differentiated analysis of performance characteristics for employees in different positions and at different career stages. Furthermore, the performance analysis results are highly subjective and have limited reference value. It cannot achieve accurate calculation of performance scores or scientific prediction of performance trends, thus failing to meet the needs of enterprises for refined performance management. Summary of the Invention
[0004] Therefore, it is necessary to provide a method, system, device, and medium for automatic statistical analysis of employee performance data that can adapt to diverse job types and career stages, addressing the aforementioned technical issues.
[0005] Firstly, this application provides a method for automatic statistical analysis of employee performance data, including: Acquire employee profile information, quantitative employee performance data, and text data of employee performance records; Based on employee profile information, we extract employee profile feature vectors, and based on these employee profile feature vectors, we match the target word embedding model of employee performance record text data from the two-dimensional word embedding model library of job type and career stage. Employee performance record text data is input into the target word embedding model to generate employee performance record text semantic vectors. Employee performance quantitative data and employee performance record text semantic vectors are then input into the employee performance scoring deep learning model to generate employee performance score data. Employee performance score data within the historical analysis period is selected to construct an employee performance score sequence. This sequence is then input into an employee performance score prediction machine learning model to generate employee performance score prediction information. The system aggregates employee performance rating data and employee performance rating prediction information to generate statistical analysis results on employee performance.
[0006] In one embodiment, the employee profile feature vector includes an employee job type feature vector and an employee career stage feature vector. The employee profile feature vector is extracted based on the employee profile information, and a target word embedding model for the employee performance record text data is obtained by matching the employee profile feature vector from a two-dimensional word embedding model library of job type and career stage. This model includes: Based on employee profile information, feature vectors for employee job type and employee career stage are extracted. Employee job type information is obtained by identifying the employee job type feature vector. Employee career stage information is obtained by identifying the employee's career stage feature vector. Based on employee job type information and employee career stage information, the target word embedding model of employee performance record text data is obtained by matching from the two-dimensional word embedding model library of job type and career stage.
[0007] In one embodiment, the job type dimension of the job type career stage dual-dimensional word embedding model library includes management job type, business job type and functional job type, and the career stage dimension of the job type career stage dual-dimensional word embedding model library includes growth stage type, maturity stage type and senior stage type. The job type and career stage dual-dimensional word embedding model library includes word embedding models for management positions during growth stage, management positions during maturity stage, management positions during senior stage, business positions during growth stage, business positions during maturity stage, business positions during senior stage, functional positions during growth stage, functional positions during maturity stage, and functional positions during senior stage.
[0008] In one embodiment, employee job type feature vectors and employee career stage feature vectors are extracted based on employee profile information, including: Load the employee profile knowledge graph and map the employee profile information to the employee profile knowledge graph to obtain an employee profile knowledge graph instance. The attribute features of job type nodes in the job type subgraph and the attribute features of career stage nodes in the career stage subgraph of the employee profile knowledge graph instance are extracted respectively. Calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the job type subgraph within the employee profile knowledge graph instance. Also calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the career stage subgraph within the career stage subgraph. The employee job type feature vector is constructed based on the job type node attribute features, job type subgraph degree centrality, job type subgraph betweenness centrality, and job type subgraph proximity centrality. The employee career stage feature vector is constructed based on the career stage node attribute features, career stage subgraph degree centrality, career stage subgraph betweenness centrality, and career stage subgraph proximity centrality.
[0009] In one embodiment, the machine learning model for predicting employee performance scores is a long short-term memory network model.
[0010] In one embodiment, the deep learning model for employee performance rating includes a semantic space autoencoder, a semantic vector concatenation layer, and a semantic space decoder. The quantitative feature vector of employee performance and the semantic vector of employee performance record text are input into the deep learning model for employee performance scoring to generate employee performance scoring data, including: The employee performance quantification feature vector is input into the semantic space autoencoder to generate the employee performance quantification parameter semantic vector; Input the semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text into the semantic vector concatenation layer, and concatenate the semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text to obtain the comprehensive semantic vector of employee performance. The comprehensive semantic vector of employee performance is input into the semantic space decoder to generate employee performance score data.
[0011] In one embodiment, the expression for the multi-dimensional performance rating prediction loss function of the deep learning model for employee performance rating is as follows:
[0012] In the formula, For multi-dimensional performance score prediction loss function, The number of training samples. The total number of rating dimensions for employee performance rating data. For the first Weighting coefficients for rating dimensions For the first The first training sample of employee performance score data Training labels for the rating dimension For the first The first training sample of employee performance score data Model predictions for the rating dimension.
[0013] Secondly, this application also provides an automatic statistical analysis system for employee performance data, including: The employee performance data acquisition module is used to acquire employee profile information, quantitative employee performance data, and text data of employee performance records. The word embedding model matching module is used to extract employee profile feature vectors based on employee profile information, and to match the target word embedding model of employee performance record text data from the two-dimensional word embedding model library of job type and career stage based on the employee profile feature vectors. The employee performance rating generation module is used to input employee performance record text data into the target word embedding model to generate employee performance record text semantic vectors, and input employee performance quantitative data and employee performance record text semantic vectors into the employee performance rating deep learning model to generate employee performance rating data; The employee performance rating prediction module is used to select employee performance rating data within the historical analysis period, construct an employee performance rating sequence, and input the employee performance rating sequence into the employee performance rating prediction machine learning model to generate employee performance rating prediction information. The performance statistics result generation module is used to summarize employee performance score data and employee performance score prediction information to generate employee performance statistical analysis results.
[0014] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method as described in any of the first aspects of this application.
[0015] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the first aspects of this application.
[0016] The aforementioned method, system, equipment, and medium for automatic statistical analysis of employee performance data significantly improves the accuracy of text semantic extraction by matching target word embedding models using employee profile feature vectors. This enhances the adaptability of word embedding models, uncovers implicit performance information in the text, and improves the comprehensiveness and reliability of performance analysis. By collaboratively scoring performance using text semantic vectors and quantitative data, it deeply integrates structured and unstructured performance data, improving the accuracy and reliability of employee performance scoring. Furthermore, by constructing historical performance scoring sequences and predicting performance trends, it accurately predicts employee performance development trends, enhancing the forward-looking nature of performance analysis. This provides forward-looking data support for talent management strategy formulation, helping enterprises to proactively plan for talent development and performance improvement. Finally, by integrating current performance scores and future prediction information, it enhances the systematic decision-making capabilities of performance statistical analysis, helping enterprises optimize human resource management efficiency and strengthen their core talent competitiveness. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 A flowchart illustrating an automatic statistical analysis method for employee performance data provided in one embodiment of this application. Figure 1 ; Figure 2 A flowchart illustrating an automatic statistical analysis method for employee performance data provided in one embodiment of this application. Figure 2 ; Figure 3 This is a flowchart illustrating an automatic statistical analysis system for employee performance data, provided as an embodiment of this application. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0020] In one exemplary embodiment of this application, such as Figure 1As shown, an automatic statistical analysis method for employee performance data is provided. This embodiment illustrates the application of this method to a performance analysis terminal. It is understood that this method can also be applied to a performance analysis server, and further to a performance analysis system including both a performance analysis terminal and a performance analysis server, and is implemented through the interaction between the two. In this embodiment, the method includes the following steps: Step S101: Obtain employee profile information, employee performance quantitative data, and employee performance record text data.
[0021] Optionally, the performance analysis terminal can connect to enterprise management data sources such as human resource management systems, business operations management systems, business communication systems, collaborative office systems, and performance appraisal systems through standardized data access interfaces authorized within the enterprise. After authorization and verification through the standardized data access interfaces, the performance analysis terminal can extract employee profile information, quantitative performance data, and text data of employee performance records for the target employees.
[0022] Optionally, the employee profile information for target performance analysis employees may include, but is not limited to, the employee's basic identity information, job affiliation information, years of service information, career advancement trajectory information, professional skills and qualifications information, and job description information. After authorization and verification through a standardized data access interface, the performance analysis terminal can obtain the employee profile information for target performance analysis employees from the enterprise's human resource management system via the standardized data access interface.
[0023] Optionally, employee performance quantification data may include, but is not limited to, one or more of the following: business indicator completion data, work efficiency data, task achievement rate data, attendance rate data, customer satisfaction data, and quality pass rate data. The performance analysis terminal can obtain employee performance quantification data from the performance appraisal system through a standardized data access interface after authorization and verification.
[0024] Optionally, the employee performance record text data for target performance analysis employees may include, but is not limited to, one or more of the following: work summary text data, work report text data, project review report text data, internal evaluation text data, customer feedback evaluation text data, and business communication record text data. After authorization and verification through the standardized data access interface, the performance analysis terminal can extract the employee performance record text data for target performance analysis employees from the collaborative office system, performance appraisal system, and / or business communication system via the standardized data access interface.
[0025] Step S102: Extract the employee profile feature vector based on the employee profile information, and match the target word embedding model of the employee performance record text data from the two-dimensional word embedding model library of job type and career stage based on the employee profile feature vector.
[0026] Specifically, the performance analysis terminal can first load an employee profile knowledge graph, and then extract employee profile feature vectors from the employee profile information based on the employee profile knowledge graph. Based on these employee profile feature vectors, the performance analysis terminal can match the target word embedding model of the employee performance record text data from a two-dimensional word embedding model library of job type and career stage.
[0027] For example, the performance analysis terminal can load an employee profile knowledge graph and map the employee profile information of the target performance analysis employee to the employee profile knowledge graph to obtain an employee profile knowledge graph instance for the target performance analysis employee. The performance analysis terminal can extract the attribute features of the job type nodes in the job type subgraph and the attribute features of the career stage nodes in the career stage subgraph of the employee profile knowledge graph instance. The performance analysis terminal can calculate the degree centrality, betweenness centrality, and proximity centrality of the employee node corresponding to the target performance analysis employee in the job type subgraph within the employee profile knowledge graph instance, and also calculate the degree centrality, betweenness centrality, and proximity centrality of the career stage subgraph for the employee node corresponding to the target performance analysis employee in the career stage subgraph. The performance analysis terminal can construct an employee job type feature vector based on the job type node attribute features, job type subgraph degree centrality, job type subgraph betweenness centrality, and job type subgraph proximity centrality, and construct an employee career stage feature vector based on the career stage node attribute features, career stage subgraph degree centrality, career stage subgraph betweenness centrality, and career stage subgraph proximity centrality.
[0028] Furthermore, the performance analysis terminal can identify employee job type information based on employee job type feature vectors and employee career stage information based on employee career stage feature vectors. The performance analysis terminal can also identify the job type dimension of the target word embedding model corresponding to the target performance analysis employee based on the employee job type information. Finally, the performance analysis terminal can identify the career stage dimension of the target word embedding model corresponding to the target performance analysis employee based on the employee career stage information.
[0029] Optionally, the performance analysis terminal can also identify the job type dimension of the target word embedding model corresponding to the target performance analysis employee based on the employee's job type feature vector, and the performance analysis terminal can also identify the career stage dimension of the target word embedding model corresponding to the target performance analysis employee based on the employee's career stage feature vector, without limitation here.
[0030] Furthermore, the performance analysis terminal can match the target word embedding model of the employee's performance record text data from the job type and career stage dual-dimensional word embedding model library based on the job type dimension and career stage dimension of the target word embedding model corresponding to the target performance analysis employee.
[0031] For example, the job type dimension of the word embedding model in the job type and career stage dual-dimensional word embedding model library can include management job type, business job type, and functional job type, and the career stage dimension of the word embedding model in the job type and career stage dual-dimensional word embedding model library can include growth stage type, maturity stage type, and senior stage type. The job type and career stage dual-dimensional word embedding model library may include, but is not limited to, word embedding models for management job growth stage, management job maturity stage, and management job senior stage; word embedding models for business job growth stage, business job maturity stage, and business job senior stage; word embedding models for functional job growth stage, functional job maturity stage, and functional job senior stage.
[0032] Step S103: Input the employee performance record text data into the target word embedding model to generate the employee performance record text semantic vector, and input the employee performance quantitative data and the employee performance record text semantic vector into the employee performance scoring deep learning model to generate employee performance score data.
[0033] Specifically, the performance analysis terminal can input the text data of the employee's performance records into the matched target word embedding model to generate a semantic vector of the employee's performance record text. The performance analysis terminal can also input the quantitative data of the employee's performance and the semantic vector of the employee's performance record text into a deep learning model for employee performance scoring to generate employee performance score data.
[0034] Optionally, the deep learning model for employee performance scoring can be deployed on a performance analysis terminal, a performance analysis server, or a performance analysis system; no limitation is made here.
[0035] Step S104: Select employee performance score data within the historical analysis period of employee performance, construct an employee performance score sequence, and input the employee performance score sequence into the employee performance score prediction machine learning model to generate employee performance score prediction information.
[0036] Specifically, the performance analysis terminal can select employee performance rating data generated by a deep learning model for employee performance rating within the historical performance analysis period for the target employee, and construct an employee performance rating sequence for the target employee. The performance analysis terminal can then input this employee performance rating sequence into a machine learning model for predicting employee performance ratings, generating predicted employee performance rating data for the target employee within the predicted performance analysis period. Finally, the performance analysis terminal can construct predicted employee performance rating information based on the predicted employee performance rating data generated by the machine learning model for predicting employee performance ratings within the predicted performance analysis period for the target employee.
[0037] Optionally, the employee performance rating prediction machine learning model can be deployed on a performance analysis terminal, a performance analysis server, or a performance analysis system, without any limitation.
[0038] Step S105: Summarize employee performance score data and employee performance score prediction information to generate employee performance statistical analysis results.
[0039] Specifically, the performance analysis terminal can aggregate employee performance score data and employee performance score prediction information of employees under target performance analysis, and generate statistical analysis results of employee performance of employees under target performance analysis.
[0040] In the aforementioned method for automatic statistical analysis of employee performance data, target word embedding model matching using employee profile feature vectors significantly improves the accuracy of text semantic extraction, thereby enhancing the adaptability of word embedding models, uncovering implicit performance information in the text, and improving the comprehensiveness and reliability of performance analysis. By collaboratively scoring performance using text semantic vectors and quantitative data, structured and unstructured performance data can be deeply integrated, improving the accuracy and reliability of employee performance scoring. Constructing historical performance scoring sequences and predicting performance trends allows for accurate prediction of employee performance development trends, enhancing the forward-looking nature of performance analysis. This provides forward-looking data support for talent management strategy formulation, helping enterprises to proactively plan for talent development and performance improvement. Integrating current performance scores and future prediction information enhances the systematic decision-making capabilities of performance statistical analysis, helping enterprises optimize human resource management efficiency and strengthen their core talent competitiveness.
[0041] In an optional embodiment of this application, the employee profile feature vector may include an employee job type feature vector and an employee career stage feature vector. Please refer to [reference needed]. Figure 1 and Figure 2Step S102 involves extracting employee profile feature vectors based on employee profile information, and then matching the target word embedding model of the employee performance record text data from a two-dimensional word embedding model library based on the employee profile feature vectors. This may include: Specifically, the performance analysis terminal can extract employee job type feature vectors and employee career stage feature vectors based on employee profile information.
[0042] Step S206: Employee job type information is obtained by identifying the employee job type feature vector.
[0043] For example, the performance analysis terminal can load a job type standard feature vector library. The terminal can calculate the job type standard feature similarity index between the employee job type feature vector of the target performance analysis employee and the corresponding job type standard feature vectors in the job type standard feature vector library. The terminal can then set the employee job type information for the target performance analysis employee based on the job type information corresponding to the job type standard feature vector with the highest similarity index and / or the job type information corresponding to the job type standard feature vector with a similarity index greater than the job type standard feature similarity threshold.
[0044] For example, the performance analysis terminal can also input the employee job type feature vector of the target performance analysis employee into the employee job type recognition machine learning model to generate the employee job type information of the target performance analysis employee, which is not limited here.
[0045] Optionally, the employee job type identification machine learning model may be constructed based on one or more of the following models: logistic regression, support vector machine, random forest, gradient boosting tree, backpropagation neural network, decision tree, and Bayesian classification.
[0046] Step S207: Obtain employee career stage information based on employee career stage feature vector identification.
[0047] For example, the performance analysis terminal can load a professional stage standard feature vector library. The terminal can calculate the professional stage standard feature similarity index between the employee's professional stage feature vector and the corresponding professional stage standard feature vectors in the professional stage standard feature vector library. The terminal can then set the employee's professional stage information based on the professional stage information corresponding to the professional stage standard feature vector with the highest similarity index and / or the professional stage information corresponding to the professional stage standard feature vector with a similarity index greater than the professional stage standard feature similarity threshold.
[0048] For example, the performance analysis terminal can also input the employee career stage feature vector of the target performance analysis employee into the employee career stage recognition machine learning model to generate the employee career stage information of the target performance analysis employee, without limitation.
[0049] Optionally, the employee career stage identification machine learning model can be constructed based on one or more of the following models: logistic regression model, support vector machine model, random forest model, gradient boosting tree model, backpropagation neural network model, decision tree model, and Bayesian classification model.
[0050] To illustrate, the employee career stage identification machine learning model can be isomorphic or heteromorphic to the employee job type identification machine learning model, and no limitation is made here.
[0051] Step S208: Based on employee job type information and employee career stage information, the target word embedding model of employee performance record text data is obtained by matching from the job type and career stage two-dimensional word embedding model library.
[0052] In an optional embodiment of this application, the job type dimension of the job type and career stage dual-dimensional word embedding model library may include management job type, business job type and functional job type, and the career stage dimension of the job type and career stage dual-dimensional word embedding model library may include growth stage type, maturity stage type and senior stage type.
[0053] The job type and career stage dual-dimensional word embedding model library can include word embedding models for management positions during growth stage, management positions during maturity stage, management positions during senior stage, business positions during growth stage, business positions during maturity stage, business positions during senior stage, functional positions during growth stage, functional positions during maturity stage, and functional positions during senior stage.
[0054] In an optional embodiment of this application, extracting employee job type feature vectors and employee career stage feature vectors based on employee profile information may include: Step S202: Load the employee profile knowledge graph and map the employee profile information to the employee profile knowledge graph to obtain an employee profile knowledge graph instance.
[0055] Step S203: Extract the attribute features of job type nodes in the job type subgraph of the employee profile knowledge graph instance and the attribute features of career stage nodes in the career stage subgraph of the employee profile knowledge graph instance.
[0056] Step S204: Calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the employee profile knowledge graph instance within the job type subgraph, and calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the career stage subgraph.
[0057] Step S205: Based on the job type node attribute features, job type subgraph degree centrality, job type subgraph betweenness centrality, and job type subgraph proximity centrality, an employee job type feature vector is constructed. Based on the career stage node attribute features, career stage subgraph degree centrality, career stage subgraph betweenness centrality, and career stage subgraph proximity centrality, an employee career stage feature vector is constructed.
[0058] In an optional embodiment of this application, the machine learning model for predicting employee performance scores can be a long short-term memory network model.
[0059] In an optional embodiment of this application, the deep learning model for employee performance rating may include a semantic space autoencoder, a semantic vector concatenation layer, and a semantic space decoder.
[0060] The employee performance quantitative feature vector and the employee performance record text semantic vector are input into the employee performance scoring deep learning model to generate employee performance scoring data, which may include: Specifically, the performance analysis terminal can input the quantitative feature vector of employee performance into the semantic space autoencoder to generate the semantic vector of quantitative parameters of employee performance.
[0061] Specifically, the performance analysis terminal can input the semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text into the semantic vector splicing layer, and splice the semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text to obtain the comprehensive semantic vector of employee performance.
[0062] Specifically, the performance analysis terminal can input the comprehensive semantic vector of employee performance into the semantic space decoder to generate employee performance score data.
[0063] In an optional embodiment of this application, the expression for the multi-dimensional performance rating prediction loss function of the deep learning model for employee performance rating can be:
[0064] In the formula, A multi-dimensional performance score prediction loss function for a deep learning model of employee performance scoring. The number of training samples. The total number of rating dimensions for employee performance rating data. For the first Weighting coefficients for rating dimensions For the first The first training sample of employee performance score data Training labels for the rating dimension For the first The first training sample of employee performance score data Model predictions for the rating dimension.
[0065] In one exemplary embodiment of this application, such as Figure 2 As shown, an automatic statistical analysis method for employee performance data is provided, which may include: Step S201: Obtain employee profile information, employee performance quantitative data, and employee performance record text data.
[0066] Step S202: Load the employee profile knowledge graph and map the employee profile information to the employee profile knowledge graph to obtain an employee profile knowledge graph instance.
[0067] Step S203: Extract the attribute features of job type nodes in the job type subgraph of the employee profile knowledge graph instance and the attribute features of career stage nodes in the career stage subgraph of the employee profile knowledge graph instance.
[0068] Step S204: Calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the employee profile knowledge graph instance within the job type subgraph, and calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the career stage subgraph.
[0069] Step S205: Based on the job type node attribute features, job type subgraph degree centrality, job type subgraph betweenness centrality, and job type subgraph proximity centrality, an employee job type feature vector is constructed. Based on the career stage node attribute features, career stage subgraph degree centrality, career stage subgraph betweenness centrality, and career stage subgraph proximity centrality, an employee career stage feature vector is constructed.
[0070] Step S206: Employee job type information is obtained by identifying the employee job type feature vector.
[0071] Step S207: Obtain employee career stage information based on employee career stage feature vector identification.
[0072] Step S208: Based on employee job type information and employee career stage information, the target word embedding model of employee performance record text data is obtained by matching from the job type and career stage two-dimensional word embedding model library.
[0073] Step S209: Input the employee performance record text data into the target word embedding model to generate the employee performance record text semantic vector, and input the employee performance quantitative data and the employee performance record text semantic vector into the employee performance scoring deep learning model to generate employee performance score data.
[0074] Step S210: Select employee performance score data within the historical analysis period of employee performance, construct an employee performance score sequence, and input the employee performance score sequence into the employee performance score prediction machine learning model to generate employee performance score prediction information.
[0075] Step S211: Summarize employee performance score data and employee performance score prediction information to generate employee performance statistical analysis results.
[0076] The aforementioned method for automatic statistical analysis of employee performance data acquires multi-source employee data, combines it with employee profile knowledge graphs to complete entity attribute mapping, extracts node attribute features from job type sub-graphs and career stage sub-graphs, calculates various centrality indicators of employee nodes in the two types of sub-graphs, constructs feature vectors for employee job types and career stages, then matches and adapts target word embedding models to generate semantic vectors for performance record text, and combines performance quantification data with deep learning models to generate accurate performance scoring data and construct historical scoring sequences. Finally, it generates performance prediction information through machine learning models, obtaining performance statistical analysis results. This method deeply integrates and efficiently utilizes multi-source data, adapts to diverse job types and career stages, improves the accuracy and relevance of performance scoring and performance trend prediction, provides comprehensive, objective, and forward-looking data support for enterprise human resource management decisions, and helps optimize talent allocation and management efficiency.
[0077] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0078] Based on the same inventive concept, this application also provides an automatic statistical analysis system for employee performance data to implement the aforementioned automatic statistical analysis method for employee performance data. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more embodiments of the automatic statistical analysis system for employee performance data provided below can be found in the limitations of the automatic statistical analysis method for employee performance data described above, and will not be repeated here.
[0079] In one exemplary embodiment, such as Figure 3 As shown, an automatic statistical analysis system 300 for employee performance data is provided, which may include: The employee performance data acquisition module 301 can be used to acquire employee profile information, employee performance quantitative data, and employee performance record text data.
[0080] The word embedding model matching module 302 can be used to extract employee profile feature vectors based on employee profile information, and to match the target word embedding model of employee performance record text data from the two-dimensional word embedding model library of job type and career stage based on the employee profile feature vectors.
[0081] The employee performance rating generation module 303 can be used to input employee performance record text data into the target word embedding model to generate employee performance record text semantic vectors, and input employee performance quantitative data and employee performance record text semantic vectors into the employee performance rating deep learning model to generate employee performance rating data.
[0082] The employee performance rating prediction module 304 can be used to select employee performance rating data within the historical analysis period of employee performance, construct an employee performance rating sequence, and input the employee performance rating sequence into the employee performance rating prediction machine learning model to generate employee performance rating prediction information.
[0083] The performance statistics result generation module 305 can be used to summarize employee performance score data and employee performance score prediction information to generate employee performance statistical analysis results.
[0084] In an optional embodiment of this application, the word embedding model matching module 302 can also be used for: Based on employee profile information, feature vectors for employee job type and employee career stage are extracted.
[0085] Employee job type information is obtained by identifying the employee job type feature vector.
[0086] Employee career stage information is obtained by identifying the employee's career stage feature vector.
[0087] Based on employee job type information and employee career stage information, the target word embedding model of employee performance record text data is obtained by matching from the two-dimensional word embedding model library of job type and career stage.
[0088] In an optional embodiment of this application, the word embedding model matching module 302 can also be used for: Load the employee profile knowledge graph and map the employee profile information to the employee profile knowledge graph to obtain an employee profile knowledge graph instance.
[0089] The attribute features of job type nodes in the job type subgraph and the attribute features of career stage nodes in the career stage subgraph of the employee profile knowledge graph instance are extracted respectively.
[0090] Calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the job type subgraph within the employee profile knowledge graph instance. Also calculate the degree centrality, betweenness centrality, and proximity centrality of employee nodes in the career stage subgraph within the career stage subgraph.
[0091] The employee job type feature vector is constructed based on the job type node attribute features, job type subgraph degree centrality, job type subgraph betweenness centrality, and job type subgraph proximity centrality. The employee career stage feature vector is constructed based on the career stage node attribute features, career stage subgraph degree centrality, career stage subgraph betweenness centrality, and career stage subgraph proximity centrality.
[0092] In an optional embodiment of this application, the employee performance rating generation module 303 can also be used for: The employee performance quantification feature vector is input into the semantic space autoencoder to generate the employee performance quantification parameter semantic vector.
[0093] The semantic vectors of the employee performance quantification parameters and the employee performance record text are input into the semantic vector concatenation layer. The semantic vectors of the employee performance quantification parameters and the employee performance record text are concatenated to obtain the comprehensive semantic vector of employee performance.
[0094] The comprehensive semantic vector of employee performance is input into the semantic space decoder to generate employee performance score data.
[0095] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above-described method for automatic statistical analysis of employee performance data.
[0096] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.
[0097] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0098] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.
Claims
1. A method for automatic statistical analysis of employee performance data, characterized in that, The method includes: Acquire employee profile information, quantitative employee performance data, and text data of employee performance records; Based on the employee profile information, an employee profile feature vector is extracted, and based on the employee profile feature vector, a target word embedding model for the employee performance record text data is obtained by matching it from the two-dimensional word embedding model library of job type and career stage. The employee performance record text data is input into the target word embedding model to generate the employee performance record text semantic vector, and the employee performance quantitative data and the employee performance record text semantic vector are input into the employee performance scoring deep learning model to generate employee performance scoring data; Select employee performance score data within the historical analysis period of employee performance, construct an employee performance score sequence, and input the employee performance score sequence into an employee performance score prediction machine learning model to generate employee performance score prediction information; The employee performance score data and the employee performance score prediction information are combined to generate employee performance statistical analysis results.
2. The method according to claim 1, characterized in that, The employee profile feature vector includes an employee job type feature vector and an employee career stage feature vector. The process of extracting the employee profile feature vector based on the employee profile information, and then matching the target word embedding model of the employee performance record text data from a two-dimensional word embedding model library based on the employee profile feature vector, includes: Based on the employee profile information, the feature vectors of the employee job type and the employee career stage are extracted; Employee job type information is obtained by identifying the employee job type feature vector. Employee career stage information is obtained by identifying the employee's career stage feature vector. Based on the employee job type information and the employee career stage information, the target word embedding model of the employee performance record text data is obtained by matching from the two-dimensional word embedding model library of job type and career stage.
3. The method according to claim 2, characterized in that, The job type dimension of the job type career stage dual-dimensional word embedding model library includes management job type, business job type and functional job type, and the career stage dimension of the job type career stage dual-dimensional word embedding model library includes growth stage type, maturity stage type and senior stage type. The job type career stage dual-dimensional word embedding model library includes word embedding models for management positions during growth stage, maturity stage, and senior stage; word embedding models for business positions during growth stage, maturity stage, and senior stage; word embedding models for functional positions during growth stage, maturity stage, and senior stage; and word embedding models for functional positions during growth stage, maturity stage, and senior stage.
4. The method according to claim 2, characterized in that, The step of extracting the employee job type feature vector and the employee career stage feature vector based on the employee profile information includes: Load the employee profile knowledge graph and map the employee profile information to the employee profile knowledge graph to obtain an employee profile knowledge graph instance. The job type node attribute features of the job type subgraph and the career stage node attribute features of the career stage subgraph in the employee profile knowledge graph instance are extracted respectively. Calculate the degree centrality, betweenness centrality, and proximity centrality of the employee nodes in the employee profile knowledge graph instance within the job type subgraph, and calculate the degree centrality, betweenness centrality, and proximity centrality of the employee nodes within the career stage subgraph. The employee job type feature vector is constructed based on the job type node attribute features, the job type subgraph degree centrality, the job type subgraph betweenness centrality, and the job type subgraph proximity centrality. The employee career stage feature vector is constructed based on the career stage node attribute features, the career stage subgraph degree centrality, the career stage subgraph betweenness centrality, and the career stage subgraph proximity centrality.
5. The method according to claim 1, characterized in that, The machine learning model for predicting employee performance scores is a long short-term memory network model.
6. The method according to any one of claims 1 to 5, characterized in that, The deep learning model for employee performance scoring includes a semantic space autoencoder, a semantic vector concatenation layer, and a semantic space decoder. The step of inputting the employee performance quantification feature vector and the employee performance record text semantic vector into the employee performance scoring deep learning model to generate employee performance scoring data includes: The employee performance quantification feature vector is input into the semantic space autoencoder to generate the employee performance quantification parameter semantic vector; The semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text are input into the semantic vector concatenation layer, and the semantic vector of the employee performance quantification parameter and the semantic vector of the employee performance record text are concatenated to obtain the comprehensive semantic vector of employee performance. The employee performance comprehensive semantic vector is input into the semantic space decoder to generate the employee performance score data.
7. The method according to claim 6, characterized in that, The expression for the multi-dimensional performance rating prediction loss function of the deep learning model for employee performance rating is as follows: In the formula, The loss function for predicting the multi-dimensional performance score is... The number of training samples. The total number of rating dimensions for the employee performance rating data. For the first Weighting coefficients for rating dimensions For the first The first training sample of the employee performance score data Training labels for the rating dimension For the first The first training sample of the employee performance score data Model predictions for the rating dimension.
8. An automatic statistical analysis system for employee performance data, characterized in that, The system includes: The employee performance data acquisition module is used to acquire employee profile information, quantitative employee performance data, and text data of employee performance records. The word embedding model matching module is used to extract employee profile feature vectors based on the employee profile information, and to match the target word embedding model of the employee performance record text data from the two-dimensional word embedding model library of job type and career stage based on the employee profile feature vectors. The employee performance rating generation module is used to input the employee performance record text data into the target word embedding model to generate the employee performance record text semantic vector, and input the employee performance quantitative data and the employee performance record text semantic vector into the employee performance rating deep learning model to generate employee performance rating data; The employee performance rating prediction module is used to select the employee performance rating data within the historical analysis period of employee performance, construct the employee performance rating sequence, and input the employee performance rating sequence into the employee performance rating prediction machine learning model to generate employee performance rating prediction information. The performance statistics result generation module is used to summarize the employee performance score data and the employee performance score prediction information to generate employee performance statistical analysis result information.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.