A system for automatically matching functional assessment with employee positions

Abstract

The present invention relates to the field of human resource management technology, specifically to a function assessment and employee position automatic matching system, the system includes a skill map construction module, a skill demand analysis module, an employee skill assessment module, a matching degree analysis module, a career development planning module, a feedback collection and analysis module, a matching strategy optimization module, and a continuous learning and model update module. The present invention, through the application of graph neural networks and graph convolutional networks, enables the system to effectively analyze the complex relationship between skills and realize the construction of a dynamic skill map. The decision tree algorithm and association rule learning provide in-depth insights into employee skill development and feedback optimization. The reinforcement learning algorithm allows the system to continuously optimize itself through a reward mechanism in a simulated environment. The combination of online learning and incremental learning algorithms ensures that the system can continuously receive and adapt to changes in market data and feedback information, so that the entire system has a high degree of adaptability and continuous learning capabilities.

Description

Technical Field

[0001] The present invention relates to the technical field of human resource management, and in particular to a system for automatically matching function evaluation with employee positions. Background Art

[0002] The field of human resource management technology focuses on efficiently managing human resources within an organization through various methods and technical means. The field of human resource management technology covers a series of activities from recruitment, selection, training, evaluation to employee development, aiming to maximize employee productivity and satisfaction while promoting the achievement of organizational goals. In recent years, with the development of information technology, the field of human resource management technology has increasingly adopted automated and intelligent tools to improve management efficiency and decision-making quality.

[0003] Among them, the function assessment and employee position automatic matching system is a tool that uses information technology to automatically match employees with positions. Its main purpose is to automatically recommend the most suitable employees and positions by evaluating employees' abilities, experience and career inclinations, and analyzing the needs and characteristics of positions. It aims to achieve efficient human resource allocation, improve employee job satisfaction and production efficiency, and reduce the cost and time consumption of human resource management.

[0004] Traditional systems rely on static matching algorithms and fixed skill libraries, lacking the ability to deeply analyze and dynamically update the relationships between skills. This leads to low efficiency and accuracy in the matching process, and is unable to effectively adapt to rapidly changing market demands and skill development trends. Traditional systems do not have a mechanism to collect and analyze employee feedback to optimize matching strategies, lack the ability to be targeted and self-optimized, and make employee career development plans and skill improvement plans inaccurate or time-sensitive, making it difficult to meet the actual needs of employees and companies. This limits the potential for employee career development and also affects the efficiency of talent management for companies. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a system for automatically matching function evaluation with employee positions.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a function evaluation and employee position automatic matching system, the system includes a skill map construction module, a skill demand analysis module, an employee skill evaluation module, a matching degree analysis module, a career development planning module, a feedback collection and analysis module, a matching strategy optimization module, and a continuous learning and model updating module;

[0007] The skill map construction module uses a graph neural network based on real-time market data to represent skills in a node-based manner, encode and connect skills, analyze the relationships between skills, update the map in real time and reflect market changes, and generate a dynamic skill map.

[0008] The skill demand analysis module is based on a dynamic skill map and uses natural language processing technology to extract skill keywords from job descriptions, analyze skill occurrence frequency and relevance, and update skill nodes in the map to match skill requirements and generate skill demand data;

[0009] The employee skill assessment module uses a graph convolutional network based on skill demand data to vectorize the employee's skills, analyze the correlation between the employee's skills and the skill nodes in the graph, assess the employee's skill level, and generate an employee skill assessment record;

[0010] The matching analysis module uses the BERT model based on the employee skill assessment records to perform feature learning on the employee skill set and job skill requirements, analyze the correlation and compatibility between employee skills and job requirements, and generate matching analysis data;

[0011] The career development planning module uses a decision tree algorithm based on matching analysis data to analyze employee skill paths and future job development trends according to employee skill levels, and provides guidance on career development and skill planning to generate career and skill improvement plans;

[0012] The feedback collection and analysis module analyzes employee feedback content based on the career and skill improvement plan by using association rule learning, identifies potential improvement points in the matching strategy by mining frequent item sets and association rules in employee feedback data, and generates feedback optimization results;

[0013] The matching strategy optimization module uses a reinforcement learning algorithm based on feedback optimization results to refine and adjust the existing employee and position matching strategy, optimizes strategy parameters and weight distribution through a reward mechanism in a simulated environment, and generates an optimized matching strategy;

[0014] The continuous learning and model updating module is based on the optimized matching strategy, adopts online learning and incremental learning algorithms, continuously receives market data and feedback information, iteratively updates the model, and matches skill requirements and workplace environment changes to generate an updated model.

[0015] The improvements of the present invention are that the skill requirement data includes skill keywords, skill frequency and skill correlation, the employee skill assessment record includes skill level score, skill potential index and skill development plan, the matching analysis data includes similarity score, compatibility evaluation and matching priority, the optimized matching strategy includes weight adjustment value, parameter update list and matching logic improvement measures, and the updated model includes performance improvement indicators, efficiency optimization measures and skill matching capabilities.

[0016] The present invention is improved in that the skill map construction module includes a market data analysis submodule, a job description analysis submodule, and a map update submodule;

[0017] The market data analysis submodule uses time series analysis based on real-time market data, uses the Pandas library to perform sliding window processing on the data, sets the window size to 30 days, calculates the frequency of occurrence of skills within the window period, sorts the frequency of occurrence, identifies the skills with the best market demand growth, and generates skill demand trend analysis;

[0018] The job description analysis submodule is based on skill demand trend analysis, adopts natural language processing technology, uses NLTK library and regular expressions to perform text processing, including word segmentation, stop word removal and stem extraction, and uses TF-IDF algorithm to calculate the weight of skill keywords and generate a job skill keyword list;

[0019] The graph update submodule is based on the job skill keyword list, uses a graph neural network, builds a graph model through the PyG library, defines a node for the skill, uses edges to connect related nodes according to the correlation in the skill keyword list, sets the node features to the frequency and weight of the skill, updates the graph and includes the identified skills and relationships, and generates a dynamic skill graph.

[0020] The present invention is improved in that the skill demand analysis module includes a skill demand identification submodule, a key skill analysis submodule, and a scarce skill identification submodule;

[0021] The skill demand identification submodule is based on the dynamic skill map and uses the k-means clustering algorithm of the Scikit-learn library. The number of clusters is set to 5. According to the characteristics of skill nodes, including frequency of occurrence and market trends, skills are clustered to identify skill groups at different levels of market demand and generate skill demand identification results.

[0022] The key skill analysis submodule adopts association rule learning based on the skill requirement identification results, uses the Apriori algorithm of the MLxtend library, sets the minimum support to 0.01 and the minimum confidence to 0.5, analyzes the association rules between different skill groups, identifies the key skills and interdependencies in the industry, and generates key skill analysis records;

[0023] The in-demand skills identification submodule adopts a decision tree algorithm based on key skill analysis records, uses the Scikit-learn library to build a decision tree model, sets the partitioning criterion to information gain, and the maximum depth to 5, analyzes the relationship between key skills and market demand, identifies skills that are in short supply in the current market, and generates skill demand data.

[0024] The present invention is improved in that the employee skill assessment module includes a work experience analysis submodule, an education background comparison submodule, and a skill self-assessment analysis submodule;

[0025] The work experience analysis submodule uses natural language processing technology to extract keywords from job descriptions based on skill demand data, segment words and remove stop words through the NLTK library, apply the TF-IDF algorithm, calculate keyword weights, and analyze employees' practical experience in the skill field to generate work experience analysis data;

[0026] The education background comparison submodule uses the cosine similarity function of the Scikit-learn library based on work experience analysis data to compare and analyze education background and market skill requirements, calculate the course and skill requirement data in the employee's education background, and analyze the matching degree with skill keywords to generate background matching analysis results;

[0027] The skill self-assessment analysis submodule analyzes the employee's self-assessed skill information based on the background matching analysis results, integrates the employee's self-assessment data with the work experience analysis data through the weighted average method, determines the employee's cognition of the degree of their own skill mastery, and generates an employee skill assessment record.

[0028] The present invention is improved in that the matching degree analysis module includes a skill matching analysis submodule, a job requirement matching submodule, and a matching degree calculation submodule;

[0029] The skill matching analysis submodule loads the pre-trained BERT model through the Transformers library based on the employee skill assessment records, inputs the employee skill set and job description text into the model for feature extraction, analyzes the matching degree between skills and job requirements, and generates a preliminary match between skills and jobs;

[0030] The job demand matching submodule uses the K-means clustering algorithm based on the preliminary matching of skills and jobs, sets three cluster centers, classifies job requirements, highlights the criticality of optimal skills, and assigns differentiated weights to skill groups to generate job demand matching analysis results;

[0031] The matching degree calculation submodule uses the BERT model to perform feature learning based on the job requirement matching analysis results, compares the correlation between employee skills and job requirements, adjusts the learning rate to 2e-5 through fine-tuning of the model, selects Adam as the optimizer, selects cross entropy as the loss function, performs matching degree calculation, and generates matching degree analysis data.

[0032] The present invention is improved in that the career development planning module includes a career path planning submodule, a skill improvement planning submodule, and a career interest matching submodule;

[0033] The career path planning submodule is based on matching analysis data, uses the Scikit-learn library, adopts a decision tree algorithm, defines the decision tree parameter criterion as entropy, and the maximum depth is 4. By analyzing the employee's skill data and market trend data, the career path of the matching employee is identified and the career path planning result is generated;

[0034] The skill improvement planning submodule is based on the career path planning results, uses the Scikit-learn library, adopts a support vector machine, sets the kernel function to linear, C to 1.0, balances the error and model complexity, analyzes the difference between the employee's existing skills and the skills required for the target career path, and generates skill improvement planning measures;

[0035] The career interest matching submodule is based on skill improvement planning measures, uses the Scikit-learn library, applies the K nearest neighbor algorithm, sets the number of neighbors n_neighbors to 5, determines the classification decision basis, and assigns the optimal weight to the distance. It also matches the personal interests of employees with potential career paths and generates career and skill improvement plans.

[0036] The present invention is improved in that the feedback collection and analysis module includes a feedback collection submodule, a feedback analysis submodule, and an optimization scheme submodule;

[0037] The feedback collection submodule collects employee feedback based on the career and skill improvement plan, designs a questionnaire using an online questionnaire tool, collects employee feedback on the plan, and counts the feedback frequency and theme to generate an employee feedback data set;

[0038] The feedback analysis submodule is based on the employee feedback data set, uses the MLxtend library, adopts the association rule learning algorithm, sets the minimum support to 0.02 and the minimum confidence to 0.5, analyzes the patterns and trends in employee feedback, identifies key issues and employee concerns in career planning and skill improvement plans, and generates feedback analysis results;

[0039] The optimization program submodule formulates the optimization program based on the feedback analysis results and adopts the linear regression model to analyze the relationship between employee satisfaction and the plan, adjust the plan to meet employee needs and market development trends, and generate feedback optimization results.

[0040] The present invention is improved in that the matching strategy optimization module includes a strategy adjustment submodule, a reinforcement learning application submodule, and an accuracy improvement submodule;

[0041] The strategy adjustment submodule is based on the feedback optimization results. By analyzing the feedback data, the deficiencies in the matching strategy are identified. The Pandas library is used for data processing, including calculating multiple feedback frequencies, applying the decision tree model, setting the maximum depth to 3, adjusting the parameters and weights in the matching strategy, and generating strategy adjustment analysis data.

[0042] The reinforcement learning application submodule uses a deep Q network to optimize the strategy based on the strategy adjustment analysis data, builds a neural network model through the TensorFlow framework, sets the learning rate to 0.01, and the discount factor to 0.95. By interacting with the simulation environment, the strategy parameters are adjusted according to the reward signal to generate an optimization strategy draft;

[0043] The accuracy improvement submodule uses K-fold cross validation to verify the effectiveness of the matching strategy based on the optimization strategy draft, sets 5-fold cross validation, evaluates the model accuracy after the strategy adjustment, adjusts the parameters and weight configuration in the matching strategy according to the verification results, and generates an optimized matching strategy.

[0044] The present invention is improved in that the continuous learning and model updating module includes an online learning submodule, an incremental learning submodule, and a model updating submodule;

[0045] The online learning submodule uses an online gradient descent method based on the optimized matching strategy, sets the step size to 0.02, updates the model in real time, reflects the received market data and employee feedback information, performs numerical calculations through the NumPy library, gradually adjusts the model parameters, and generates online learning update records;

[0046] The incremental learning submodule uses a micro-batch processing method to process new data based on online learning update records, sets the size of each batch of data to 50, uses TensorFlow and Keras to adjust the model, matches the data distribution, determines that the model matches the workplace environment and skill requirements, and generates incremental learning update results;

[0047] The model update submodule records update information, including update time and update content, based on incremental learning update results, uses Git to manage model versions, verifies that updates can be recorded and tracked, and generates an updated model.

[0048] Compared with the prior art, the advantages and positive effects of the present invention are:

[0049] In the present invention, through the application of graph neural networks and graph convolutional networks, the system can effectively analyze the complex relationship between skills and realize the construction of a dynamic skill map, while the use of natural language processing technology optimizes the process of extracting skill keywords from job descriptions. The BERT model further enhances the analysis of the correlation and compatibility between employee skill sets and job requirements, ensuring that the deep learning and feature extraction of matching results are more accurate. Decision tree algorithms and association rule learning provide in-depth insights into employee skill development and feedback optimization, while reinforcement learning algorithms allow the system to continuously optimize itself through reward mechanisms in simulated environments. The combination of online learning and incremental learning algorithms ensures that the system can continuously receive and adapt to changes in market data and feedback information, making the entire system highly adaptable and capable of continuous learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A module diagram of a function evaluation and employee position automatic matching system is proposed for the present invention;

[0051] Figure 2 A system framework diagram of a function evaluation and employee position automatic matching system is proposed for the present invention;

[0052] Figure 3 The present invention proposes a flow chart of a skill map construction module in a function evaluation and employee position automatic matching system;

[0053] Figure 4 The present invention proposes a flow chart of a skill demand analysis module in a function evaluation and employee position automatic matching system;

[0054] Figure 5 The present invention proposes a flow chart of an employee skill assessment module in a function assessment and employee position automatic matching system;

[0055] Figure 6The present invention proposes a flow chart of a matching degree analysis module in a function evaluation and employee position automatic matching system;

[0056] Figure 7 The present invention proposes a flow chart of a career development planning module in a system for automatically matching function evaluation with employee positions;

[0057] Figure 8 The present invention proposes a flow chart of a feedback collection and analysis module in a function evaluation and employee position automatic matching system;

[0058] Fig. 9 The present invention proposes a flow chart of a matching strategy optimization module in a function evaluation and employee position automatic matching system;

[0059] Fig.10 The present invention proposes a flow chart of a continuous learning and model updating module in a function evaluation and employee position automatic matching system. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0061] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating positions or positional relationships, are based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0062] Example

[0063] See also Figure 1 ,The present invention provides a technical solution: a function assessment and employee position automatic matching system includes a skill map construction module, a skill demand analysis module, an employee skill assessment module, a matching degree analysis module, a career development planning module, a feedback collection and analysis module, a matching strategy optimization module, and a continuous learning and model updating module;

[0064] The skill map construction module uses graph neural networks based on real-time market data to represent skills in a node-based manner, encode and connect skills, and analyze the relationships between skills. It updates the map in real time and reflects market changes to generate a dynamic skill map.

[0065] The skill demand analysis module is based on a dynamic skill map and uses natural language processing technology to extract skill keywords from job descriptions, analyze skill occurrence frequency and relevance, and update skill nodes in the map to match skill requirements and generate skill demand data.

[0066] The employee skill assessment module uses graph convolutional networks to vectorize employee skills based on skill demand data, analyze the correlation between employee skills and skill nodes in the graph, assess employee skill levels, and generate employee skill assessment records;

[0067] The matching analysis module uses the BERT model to learn the characteristics of employee skill sets and job skill requirements based on employee skill assessment records, analyze the correlation and compatibility between employee skills and job requirements, and generate matching analysis data;

[0068] The career development planning module uses a decision tree algorithm based on matching analysis data to analyze employee skill paths and future job development trends according to employee skill levels, and provides guidance on career development and skill planning to generate career and skill improvement plans;

[0069] The feedback collection and analysis module uses association rule learning to analyze employee feedback content based on the career and skill improvement plan. By mining frequent item sets and association rules in employee feedback data, it identifies potential improvement points in the matching strategy and generates feedback optimization results.

[0070] The matching strategy optimization module uses the reinforcement learning algorithm based on the feedback optimization results to refine the existing employee and position matching strategy, optimize the strategy parameters and weight distribution through the reward mechanism in the simulation environment, and generate the optimized matching strategy;

[0071] The continuous learning and model updating module is based on the optimized matching strategy, adopts online learning and incremental learning algorithms, continuously receives market data and feedback information, iteratively updates the model, and matches skill requirements and changes in the workplace environment to generate an updated model.

[0072] Skill requirement data includes skill keywords, skill frequency and skill correlation. Employee skill assessment records include skill level scores, skill potential index and skill development plans. Matching analysis data includes similarity scores, compatibility evaluations and matching priorities. The optimized matching strategies include weight adjustment values, parameter update lists and matching logic improvement measures. The updated models include performance improvement indicators, efficiency optimization measures and skill matching capabilities.

[0073] In the skill graph construction module, real-time market data is analyzed through graph neural networks, and skills are encoded and connected using a node-based representation method. Specifically, the module first collects skill data on the market, and then uses the node representation learning technology of graph neural networks to regard each skill as a node in the graph, and constructs edges between skills by learning the interactions and connections between skills. This process involves feature vectorization of skill nodes and uses the propagation rules of graph neural networks to update node states, thereby capturing the complex relationships between skills. Through continuous iterative learning, this module achieves dynamic updates of the skill graph to reflect market changes. The generated dynamic skill graph can provide real-time and accurate skill relationship data for subsequent modules, promoting more effective skill demand analysis and employee skill assessment.

[0074] In the skill demand analysis module, natural language processing technology is used to extract and analyze skill keywords in job descriptions based on dynamic skill graphs. This module uses text mining and keyword extraction algorithms to identify key skill vocabulary from job description texts, and uses the term frequency-inverse document frequency (TF-IDF) method to measure the importance of skill keywords. By calculating the frequency of occurrence of skill keywords and their distribution in different job descriptions, the correlation between skills is analyzed. Then, this module updates the node information in the skill graph to reflect the current market demand for various skills. The generated skill demand data provides a basis for employee skill evaluation and matching analysis, ensuring the accuracy and efficiency of job matching.

[0075] In the employee skill assessment module, the employee's skills are vectorized and evaluated based on the skill requirement data through the graph convolutional network. This module uses the graph convolutional network to process the nodes in the skill graph, maps and compares the employee's personal skills with the nodes in the skill graph, and uses the hierarchical feature extraction capability of the graph convolutional network to evaluate the match between employee skills and job requirements. By constructing the employee's skill vector and comparing it with the skill nodes in the graph, the correlation between the employee's skill level and each skill node is calculated. The generated employee skill assessment record records in detail the employee's skill level score, skill potential index, etc., providing a basis for accurate job matching and career development planning.

[0076] In the matching analysis module, the BERT model is used to perform feature learning and matching analysis on employee skill sets and job skill requirements. The deep bidirectional representation capability of the BERT model is used to deeply understand the semantic content of employee skill descriptions and job requirement texts, thereby accurately evaluating the similarity and compatibility between employee skills and job requirements. This module trains the BERT model to learn the feature representations in employee skill texts and job description texts, and then calculates the matching between employees and jobs. The generated matching analysis data includes similarity scores, compatibility evaluations, and matching priorities.

[0077] In the career development planning module, the decision tree algorithm is used to analyze employees' skill paths and future job development trends based on matching analysis data. This module uses a decision tree algorithm to classify and make decisions on employee skill levels and job skill requirements, and guide employees in their career development and skill planning. By analyzing the gap between employees' current skill levels and market demand, this module proposes specific skill improvement plans and career development directions. The generated career and skill improvement plans are designed to help employees clarify their future development goals and develop practical and feasible skill improvement strategies.

[0078] In the feedback collection and analysis module, the employee feedback content is deeply analyzed through association rule learning. This module adopts the association rule learning method to mine frequent item sets and association rules from the employee feedback data, and identify potential improvement points in matching strategies and career development plans. By analyzing the employee feedback on skill matching and career development plans, this module continuously optimizes matching strategies and development suggestions to better meet the needs and expectations of employees. The generated feedback optimization results provide valuable improvement suggestions for matching strategies and career development plans.

[0079] In the matching strategy optimization module, the matching strategy is optimized and adjusted through the reinforcement learning algorithm. The module uses the reward mechanism of the reinforcement learning algorithm to adjust the parameters and weight distribution of the matching strategy according to the feedback optimization results to achieve a better matching effect between employees and positions. Through the interaction process in the simulation environment, the module continuously adjusts the strategy parameters, optimizes the matching logic and weight distribution to improve the matching accuracy and efficiency. The generated optimized matching strategy provides employees and enterprises with more efficient and accurate talent matching services.

[0080] In the continuous learning and model updating module, the model is continuously updated through online learning and incremental learning algorithms to adapt to changes in the market and workplace environment. This module uses online learning and incremental learning algorithms to process the latest market data and employee feedback received to ensure that the model can quickly adapt to new skill requirements and matching strategies. Through the continuous learning and updating process, the module maintains the flexibility and adaptability of the system, ensuring the long-term effectiveness and accuracy of the skill matching system. The generated updated model not only improves performance and efficiency, but also enhances the system's responsiveness to emerging skills and workplace trends.

[0081] See also Figure 2 and Figure 3 ,The skill map building module includes a market data analysis submodule, a job description analysis submodule, and a map updating submodule;

[0082] The market data analysis submodule uses time series analysis based on real-time market data and the Pandas library to perform sliding window processing on the data. The window size is set to 30 days. The frequency of occurrence of skills within the window period is calculated and sorted. The skills with the best market demand growth are identified and skill demand trend analysis is generated.

[0083] The job description analysis submodule is based on skill demand trend analysis and uses natural language processing technology, NLTK library and regular expressions to perform text processing, including word segmentation, stop word removal and stem extraction. It also uses the TF-IDF algorithm to calculate the weight of skill keywords and generate a job skill keyword list.

[0084] The graph update submodule is based on the job skill keyword list and uses graph neural networks to build a graph model through the PyG library. It defines a node for the skill, uses edges to connect related nodes according to the correlation in the skill keyword list, sets the node features to the frequency and weight of the skill, updates the graph and includes the identified skills and relationships, and generates a dynamic skill graph.

[0085] In the market data analysis submodule, the real-time market data is processed by sliding windows through time series analysis combined with the Pandas library to finely identify and analyze the dynamic changes in skill demand. The data format used is mainly time series data, which contains the number of occurrences of each skill and the corresponding timestamp. The specific operation process involves setting a 30-day sliding window. For each window period, the frequency of occurrence of each skill is calculated and recorded. This is achieved through the sliding window function of the Pandas library. The .rolling(window=30).mean() method is used to calculate the average frequency of occurrence within the window period, and then the frequencies are sorted to identify the skills with the fastest growing demand. Such analysis not only reveals which skills are on the rise in the market, but also provides data support for the update of the skill map. The generated skill demand trend analysis file lists in detail the changes in demand for each skill in the market, providing important input for the job description analysis submodule and ensuring the real-time and accuracy of the skill map.

[0086] In the job description analysis submodule, the job description text is deeply analyzed through natural language processing technology combined with the NLTK library and regular expressions to extract key skill vocabulary. Text processing operations include word segmentation, stop word removal and stem extraction, which aim to clean and standardize text data and improve the accuracy of subsequent analysis. Non-text content is removed through regular expressions, and the NLTK library is used for word segmentation and stop word removal, followed by stem extraction to streamline the vocabulary to its root form. Next, the TF-IDF algorithm is used to calculate the weight of each skill keyword in the job description, highlighting those skill vocabulary that are both common and specific to certain positions. This step is implemented through the NLTK library, and the TfidfVectorizer class is used to convert and analyze text data. The generated job skill keyword list shows in detail the importance of the skills required for each position, providing accurate input data for the graph update submodule to ensure that the skill graph can accurately reflect current job requirements.

[0087] In the graph update submodule, a dynamic skill graph is constructed and updated through graph neural network and PyG library, and the job skill keyword list is converted into nodes and edges in the graph. Each skill is defined as a node in the graph, and the edges between nodes represent the correlation between skills. Node features include the frequency of skill occurrence and TF-IDF weight. Through the PyG library, a graph data structure is constructed, and the learning ability of the graph neural network is used to analyze and update the relationship between skills. When setting node features, the market frequency of skills and the TF-IDF weight in the job description are used as node attributes. This not only takes into account the importance of skills, but also the actual demand for skills in the market. By training the graph neural network model, the connection between skills and node features are optimized to ensure that the skill graph can accurately reflect the latest relationship between skills and market dynamics. The generated dynamic skill graph provides an updated and accurate skill relationship network for the functional assessment and employee position automatic matching system, supporting the system to efficiently and accurately conduct employee skill assessment and position matching.

[0088] Assume that the market data contains the frequency data of the skills "Python programming", "data analysis" and "machine learning", among which the demand for "machine learning" has increased significantly in the past 30 days. The job description analysis identifies the skill keywords and calculates the TF-IDF algorithm to obtain the highest weight for "machine learning", indicating that it is extremely important for certain positions in the current market. During the graph update process, skills are added to the skill graph, among which "machine learning" becomes an important node in the graph due to its high frequency and high weight. It is connected to other skills such as "Python programming" and "data analysis" through edges, indicating the close relationship between them. The generated dynamic skill graph reflects the market demand and mutual correlation of skills in detail, providing valuable input data for the functional assessment and employee position automatic matching system, and promoting more accurate matching between employees and positions.

[0089] See also Figure 2 and Figure 4 ,The skill demand analysis module includes the skill demand identification submodule, the key skill analysis submodule, and the ,scarce skill identification submodule;

[0090] The skill demand identification submodule is based on the dynamic skill graph and uses the k-means clustering algorithm of the Scikit-learn library. The number of clusters is set to 5. According to the characteristics of skill nodes, including frequency of occurrence and market trends, skills are clustered to identify skill groups at different levels of market demand and generate skill demand identification results.

[0091] The key skills analysis submodule adopts association rule learning based on the results of skill requirement identification, uses the Apriori algorithm of the MLxtend library, sets the minimum support to 0.01 and the minimum confidence to 0.5, analyzes the association rules between different skill groups, identifies key skills and interdependencies in the industry, and generates key skills analysis records;

[0092] The in-demand skills identification submodule is based on key skills analysis records, adopts a decision tree algorithm, uses the Scikit-learn library to build a decision tree model, sets the partitioning criterion to information gain, and sets the maximum depth to 5. It analyzes the relationship between key skills and market demand, identifies skills that are in short supply in the current market, and generates skill demand data.

[0093] In the skill demand identification submodule, the skills in the dynamic skill map are clustered by the k-means clustering algorithm of the Scikit-learn library. The input data format of this submodule is the feature set of skill nodes, including the frequency of occurrence of each skill and market trend data. During the operation, the feature data of the skill is first standardized to ensure that the features of different dimensions have the same weight in the cluster analysis. Then, the number of clusters is set to 5, and the k-means algorithm is used to cluster the features of the skill nodes. The algorithm iteratively updates the cluster centers until the optimal distribution is achieved. In Scikit-learn, the KMeans(n_clusters=5) function is used to implement clustering, where the n_clusters parameter is set to 5, indicating that the data set is expected to be divided into 5 groups. The clustering results reveal the skill groups with different levels of demand in the market. The generated skill demand identification results list the classification of skills in each group in detail, which helps to understand the differences in market demand for different skills and provides a basis for subsequent key skill analysis and identification of scarce skills.

[0094] In the key skill analysis submodule, based on the skill requirement identification results, the Apriori algorithm of the MLxtend library is used to learn association rules and analyze the interdependence between different skill groups. The input data is the result of skill clustering, and the data format is converted into a transaction record format. Each record represents a skill set in a skill group. By setting the minimum support to 0.01 and the minimum confidence to 0.5, the Apriori algorithm is used to find frequent item sets and construct association rules. In MLxtend, this is implemented through the apriori and association_rules functions, which are used to find frequent item sets and derive association rules, respectively. This step reveals the key connections and dependencies between skills in the industry, and the generated key skill analysis records illustrate the correlation between skills.

[0095] In the submodule of identifying scarce skills, a decision tree model is constructed based on the key skill analysis records using the Scikit-learn library to analyze the relationship between key skills and market demand and identify skills that are currently in short supply in the market. The input data is the result of the key skill analysis, including the relevance of skills and market demand indicators. By setting information gain as the division criterion and the maximum depth to 5, a decision tree model is constructed to analyze the relationship between skills and market demand. In Scikit-learn, the decision tree model (criterion = entropy, max_depth = 5) function is used to implement model construction, where criterion = entropy specifies the use of information gain as the criterion for node division, and max_depth = 5 limits the maximum depth of the tree. By analyzing the criticality and market demand of skills, the most scarce skills are identified. The generated skill demand data provides important input for the functional evaluation and employee position automatic matching system, helping the system to accurately match employee skills with position requirements.

[0096] Assume that the dynamic skill map contains the occurrence frequency and market trend data of the skills "Python programming", "data analysis", "machine learning", "artificial intelligence" and "big data processing". Through the cluster analysis of the skill demand identification sub-module, the skills are divided into different groups. For example, "machine learning" and "artificial intelligence" are clustered together, indicating that the market demand for these skills is similar and growing rapidly. The key skill analysis sub-module finds through association rule learning that the skills of "machine learning" and "big data processing" often appear together, indicating that the two skills have a close relationship in the industry. The scarce skill identification sub-module further analyzes and determines that "data analysis" is the most scarce skill in the current market. Finally, the generated skill demand data details the market demand for the "data analysis" skill, providing an accurate matching basis for the functional assessment and employee position automatic matching system.

[0097] See also Figure 2 and Figure 5 ,The employee skill assessment module includes the work experience analysis submodule, the education background comparison submodule, and the skill self-assessment analysis submodule;

[0098] The work experience analysis submodule uses natural language processing technology to extract keywords from job descriptions based on skill demand data, segment words and remove stop words through the NLTK library, apply the TF-IDF algorithm, calculate keyword weights, and analyze employees' practical experience in the skill field to generate work experience analysis data;

[0099] The education background comparison submodule uses the cosine similarity function of the Scikit-learn library to analyze the education background and market skill requirements based on work experience analysis data, calculate the course and skill requirement data in the employee's education background, and analyze the matching degree with skill keywords to generate background matching analysis results;

[0100] The skill self-assessment analysis submodule analyzes the self-assessed skill information of employees based on the background matching analysis results. It integrates the employee self-assessment data with the work experience analysis data through the weighted average method, determines the employee's cognition of their own skill mastery, and generates employee skill assessment records.

[0101] In the work experience analysis submodule, natural language processing technology is used, especially the use of the NLTK library for text processing, to extract keywords from the job description text provided by the employee in order to analyze the employee's practical experience in specific skill areas. This submodule first preprocesses the job description text, including using the word segmentation function of the NLTK library to split the text into words, and then removes stop words to filter out common but insignificant words in the text. Next, the TF-IDF algorithm is used to calculate the weight of each word in the text, highlighting those words that appear frequently in the employee's job description but are relatively rare in the entire document collection. The words are often key skills or professional terms. Through this series of operations, the important skill keywords in the employee's description can be accurately identified, and the depth and breadth of their practical experience can be analyzed accordingly. The generated work experience analysis data lists in detail the employee's key skills and their related experience levels, providing basic data for further educational background comparison and skill self-assessment analysis.

[0102] In the education background comparison submodule, the cosine similarity function of the Scikit-learn library is used to compare and analyze the employees' education background with the market skill requirements. This submodule uses the key skills in the work experience analysis data as a benchmark and compares it with the skill keywords mentioned in the course content of the employee's education background. By calculating the cosine similarity between the education background text and the skill requirement data, it can quantify the match between the employee's education background and the current market skill requirements. The cosine similarity calculation represents the text as a vector in the vector space model and then calculates the angle between the vectors to obtain the similarity. The generated background matching analysis results detail the similarity between the employee's education background and the market skill requirements, providing a valuable reference for the employee's skill development and job matching.

[0103] In the skill self-assessment analysis submodule, a comprehensive analysis is conducted on the self-assessed skill information of employees, and the employee self-assessment data and work experience analysis data are integrated through the weighted average method. This submodule takes into account the employee's self-evaluation of his or her skill mastery and the practical experience level obtained through work experience analysis, and uses the weighted average method to integrate these two aspects of information, thereby obtaining a more accurate and objective employee skill assessment result. When performing weighted calculations, the weights can be adjusted according to the degree of importance that different organizations or industries attach to practical experience and self-assessment information, so as to ensure the applicability and accuracy of the assessment results. The generated employee skill assessment record lists in detail the employee's various skills and their mastery level, providing important input for the functional assessment and employee job automatic matching system, ensuring that the system can accurately identify and match employee skills with job requirements.

[0104] Suppose an employee provides a detailed job description, including keywords such as "Python programming, data analysis, machine learning project experience", as well as his educational background in the field of data science, including participation in relevant courses and training camps. In the work experience analysis sub-module, after word segmentation, removal of stop words and application of the TF-IDF algorithm, "Python programming", "data analysis" and "machine learning" are successfully identified as key skills. The education background comparison sub-module finds that the employee's educational background is highly matched with the market skill demand through cosine similarity analysis. The skill self-assessment analysis sub-module combines the employee's self-assessed skills and work experience analysis data, and ultimately determines that the employee has a high level of skill mastery in "data analysis" and "machine learning". The generated employee skill assessment record shows the employee's skill rating in detail, provides clear guidance for his career development, and provides accurate talent matching information for enterprises.

[0105] See also Figure 2 and Figure 6 ,The matching degree analysis module includes the skill matching analysis submodule, the job requirement matching submodule, and the matching degree calculation submodule;

[0106] The skill matching analysis submodule is based on the employee skill assessment records. It loads the pre-trained BERT model through the Transformers library, inputs the employee skill set and job description text into the model for feature extraction, analyzes the matching degree between skills and job requirements, and generates a preliminary match between skills and jobs.

[0107] The job demand matching submodule uses the K-means clustering algorithm based on the preliminary matching of skills and jobs, sets three cluster centers, classifies job requirements, highlights the criticality of optimal skills, and assigns differentiated weights to skill groups to generate job demand matching analysis results;

[0108] The matching degree calculation submodule uses the BERT model to perform feature learning based on the job requirement matching analysis results, compares the correlation between employee skills and job requirements, adjusts the learning rate to 2e-5 through model fine-tuning, selects Adam as the optimizer, and selects cross entropy as the loss function to perform matching degree calculation and generate matching degree analysis data.

[0109] In the skill matching analysis submodule, the pre-trained BERT model is loaded through the Transformers library to extract features of employee skill sets and job description texts. The input data format of this submodule includes the text data of employee skill evaluation records and the text data of job descriptions. Using the powerful natural language processing capabilities of the BERT model, the text data is first tokenized to convert the text into a format that the model can understand, including adding necessary special tags such as [CLS] and [SEP]. Then, the pre-processed text data is input into the BERT model. The model learns the context-related features of the text data through the self-attention mechanism, thereby extracting the deep semantic representation of employee skills and job requirements. By comparing the feature representation of the employee skill set and the feature representation of the job description, the matching degree between the two is analyzed, and the preliminary matching results of skills and positions are generated. This result provides a basis for subsequent job requirement matching analysis and helps identify which employees' skill sets match specific job requirements.

[0110] In the job requirement matching submodule, based on the preliminary matching results of skills and jobs, the K-means clustering algorithm is used to perform classification analysis of job requirements. During the operation of this submodule, the preliminary matching data set is first vectorized according to the similarity distribution characteristics of skills and job requirements, and then the number of cluster centers is set to 3. The K-means clustering algorithm is used to classify the vectorized data to identify different categories of job requirements. During the clustering process, the algorithm calculates the distance between the data point and the cluster center, iteratively updates the cluster center until the clustering result is stable. Such classification analysis helps to highlight the differences in skill requirements for different jobs. By assigning differentiated weights to different skill groups, the matching relationship between skills and job requirements is further refined. The generated job requirement matching analysis results reveal the preference of various jobs for specific skill groups and provide input data for matching calculations.

[0111] In the matching calculation submodule, based on the job requirement matching analysis results, the BERT model is used again for in-depth feature learning and matching calculation. The operation process of this submodule involves taking the matching relationship between employee skills and job requirements as input, adjusting the learning rate to 2e-5 through fine-tuning of the model, selecting Adam as the optimizer, and using cross entropy as the loss function to optimize the performance of the model and calculate the matching degree more accurately. During the fine-tuning process, the model parameters are updated through the back propagation algorithm to minimize the difference between the predicted results and the actual results, thereby improving the model's prediction accuracy for the matching degree between employee skills and job requirements. The generated matching analysis data lists in detail the matching score between each employee and each job requirement, providing the final matching recommendation for the functional evaluation and employee job automatic matching system, helping enterprises and employees to achieve more effective talent allocation and career development planning.

[0112] Suppose an employee's skill assessment record includes "Python programming", "data analysis" and "machine learning" skills, and the job description text emphasizes the need for "data analysis" and "machine learning" experience. Through the skill matching analysis sub-module, the BERT model is used to identify the highly correlated features between the employee's skill set and the job description. Subsequently, the job requirement matching sub-module classifies this job as a group with a high demand for "data science" skills. Finally, the matching calculation sub-module calculates through the fine-tuning BERT model that the matching degree between the employee and this job is very high. The generated matching analysis data recommends this job for the employee, demonstrating the efficiency and accuracy of the functional assessment and employee job automatic matching system in practical applications.

[0113] See also Figure 2 and Figure 7 ,The career development planning module includes the career path planning submodule, the skill improvement planning submodule, and the career interest matching submodule;

[0114] The career path planning submodule is based on matching analysis data, using the Scikit-learn library and the decision tree algorithm. The decision tree parameter criterion is defined as entropy, and the maximum depth is 4. By analyzing the employee's skill data and market trend data, the matching employee's career path is identified and the career path planning results are generated.

[0115] The skill improvement planning submodule is based on the career path planning results, uses the Scikit-learn library, adopts the support vector machine, sets the kernel function to linear, C to 1.0, balances the error and model complexity, analyzes the difference between the employees' existing skills and the skills required for the target career path, and generates skill improvement planning measures;

[0116] The career interest matching submodule is based on skill improvement planning measures, uses the Scikit-learn library, applies the K nearest neighbor algorithm, sets the number of neighbors n_neighbors to 5, determines the classification decision basis, and assigns the optimal weight to the distance. It also matches the personal interests of employees with potential career paths and generates career and skill improvement plans.

[0117] In the career path planning submodule, the decision tree algorithm is used through the Scikit-learn library to analyze and identify career paths based on the employee's skill data and market trend data. The data input format of this submodule includes the employee's skill evaluation record, which lists in detail the employee's various skills and their mastery level, as well as market trend data, reflecting the changes in demand for different skills in the market. The operation process first involves the initialization of the decision tree model and setting the parameters of the decision tree, where criterion = entropy is used to measure the split quality, and max_depth = 4 limits the maximum depth of the tree to avoid overfitting. Then, the model is trained based on the provided skill data and market trend data. By analyzing the relationship between skills and the association between skills and market demand, career paths with a high degree of match with employee skills are identified. The generated career path planning results detail the career development suggestions for employees, including recommended career directions and potential development space under current market trends, providing personalized career development guidance for employees.

[0118] In the skill improvement planning submodule, the Scikit-learn library is used to adopt the support vector machine (SVM) algorithm. Based on the career path planning results, the difference between the employees' existing skills and the skills required for the target career path is analyzed. The input data format is the employee's skill set and the skill set required for the corresponding career path. The kernel function is set to linear, C = 1.0 to balance the error and complexity of the model. During the operation, the SVM model analyzes the gaps and overlaps between the existing skills and the expected skills, and identifies the key skills that employees need to improve or learn. In this way, the model can provide employees with targeted skill improvement plans, including recommended courses, training programs or practice opportunities. The generated skill improvement planning measures list in detail the specific skill improvement suggestions required to achieve the target career path, helping employees to plan their personal skill development in a targeted manner and improve their professional competitiveness.

[0119] In the career interest matching submodule, the K nearest neighbor algorithm is applied through the Scikit-learn library to match employees' personal interests with potential career paths based on skill improvement planning measures. The input data format is the employee's personal interest description and career path planning results. Set n_neighbors=5 to determine the number of neighbors for classification decisions, and weights=distance to give closer neighbors higher weights. During the operation, the K nearest neighbor algorithm calculates the distance between the employee's interest point and the career path to identify the career path that best matches the employee's interest. This process takes into account the employee's preferences to ensure that the recommended career path is not only in line with the employee's skill improvement plan, but also consistent with his or her career interests. The generated career and skill improvement plan lists in detail the employee's career interest assessment and recommended matching career paths.

[0120] Assuming that an employee has basic skills in data analysis and Python programming and has a strong interest in the field of data science, the career path planning sub-module identifies data scientist as a potential career path for the employee. The skill improvement planning sub-module analysis finds that the employee needs to improve his or her machine learning and big data processing capabilities to meet the skill requirements of a data scientist. The career interest matching sub-module further confirms that the data scientist career path is highly matched with the employee's career interests. Finally, the generated career and skill improvement plan details how the employee should improve his or her skills by taking specific courses and participating in related projects to achieve his or her career goal of becoming a data scientist.

[0121] See also Figure 2 and Figure 8 ,The feedback collection and analysis module includes a feedback collection submodule, a feedback analysis submodule, and an optimization scheme submodule;

[0122] The feedback collection submodule collects employee feedback based on the career and skills improvement plan. It uses online questionnaire tools to design questionnaires, collects employee feedback on the plan, and counts the feedback frequency and theme to generate an employee feedback data set.

[0123] The feedback analysis submodule is based on the employee feedback dataset, uses the MLxtend library, adopts the association rule learning algorithm, sets the minimum support to 0.02, and the minimum confidence to 0.5, analyzes the patterns and trends in employee feedback, identifies key issues and employee concerns in career planning and skill improvement plans, and generates feedback analysis results;

[0124] The optimization plan submodule formulates optimization plans based on the feedback analysis results and adopts a linear regression model to analyze the relationship between employee satisfaction and the plan, adjust the plan to meet employee needs and market development trends, and generate feedback optimization results.

[0125] In the feedback collection submodule, questionnaires are designed through online questionnaire tools to collect employees' feedback on career path planning and skill improvement plans. The data input format of this submodule is the response content of the online questionnaire, which includes employees' opinions, suggestions and evaluation of their satisfaction with the plan. The operation process first involves designing the questionnaire to ensure that the questionnaire content covers all aspects of career planning, including satisfaction with career path selection, evaluation of the practicality of skill improvement measures and any additional suggestions or comments. After the questionnaire is designed, it is distributed to employees via email or internal communication system, and a response deadline is set. The collected feedback is automatically summarized through the online questionnaire tool, and the frequency and theme of each feedback are counted to generate an employee feedback data set. This data set records in detail the employees' views on the current career planning and skill improvement plan, providing basic data for further feedback analysis and optimization plan formulation.

[0126] In the feedback analysis submodule, the MLxtend library is used to adopt the association rule learning algorithm, based on the employee feedback data set, to analyze the patterns and trends in employee feedback. The data input format of this submodule is the item set of the employee feedback data set. During the operation, the data is first preprocessed to convert the text feedback into an analyzable item set format. Then, the minimum support is set to 0.02 and the minimum confidence is set to 0.5, and the association rule learning algorithm is used to analyze the item set to identify frequently occurring feedback patterns and trends. Through this analysis, the core issues and common opinions of employees can be revealed, and the deficiencies in career planning and skill improvement plans can be identified. The generated feedback analysis results list the key issues and focus of employee concerns in detail, providing decision support for the formulation of optimization plans.

[0127] In the optimization plan submodule, a linear regression model is used to analyze the relationship between employee satisfaction and the plan, and the plan is adjusted accordingly to meet employee needs and market development trends. The data input format of this submodule is the employee satisfaction score and the characteristic value of the plan. During the operation, the satisfaction score in the employee feedback and the relevant characteristics of the plan are first quantified, and then the linear regression model is used to analyze the data to identify the relationship between satisfaction and plan characteristics. Through model analysis, it is determined which plan characteristics have the greatest impact on employee satisfaction, and career path planning and skill improvement measures are adjusted accordingly. The generated feedback optimization results detail the adjusted plan, which aims to improve employee satisfaction and better meet market trends and employee needs.

[0128] Assume that there is an employee feedback data set that contains extensive feedback on a skill improvement plan. Through the feedback analysis sub-module, it is identified that employees generally believe that insufficient practice opportunities are a key problem. After the optimization plan sub-module is analyzed through a linear regression model, it is confirmed that increasing the number and quality of practice projects will significantly improve employee satisfaction. Finally, the generated feedback optimization results recommend adjusting the skill improvement plan, including cooperating with enterprises to carry out practical projects, providing employees with more practice opportunities, and incorporating the adjustments into career path planning and skill improvement plans to better meet the development needs of employees and improve the effectiveness of the overall plan.

[0129] See also Figure 2 and Fig. 9 ,The matching strategy optimization module includes a strategy adjustment submodule, a reinforcement learning application submodule, and an accuracy improvement submodule;

[0130] The strategy adjustment submodule analyzes feedback data based on feedback optimization results, identifies deficiencies in the matching strategy, and uses the Pandas library for data processing, including calculating multiple feedback frequencies and applying a decision tree model with a maximum depth of 3. It adjusts the parameters and weights in the matching strategy and generates strategy adjustment analysis data.

[0131] The reinforcement learning application submodule uses deep Q network for policy optimization based on policy adjustment analysis data, builds a neural network model through the TensorFlow framework, sets the learning rate to 0.01 and the discount factor to 0.95, and generates an optimization policy draft by adjusting policy parameters according to reward signals through interaction with the simulation environment;

[0132] The accuracy improvement submodule uses K-fold cross validation to verify the effectiveness of the matching strategy based on the optimization strategy draft, sets 5-fold cross validation, evaluates the model accuracy after the strategy adjustment, and adjusts the parameters and weight configuration in the matching strategy according to the verification results to generate the optimized matching strategy.

[0133] In the strategy adjustment submodule, data processing is performed through the Pandas library to analyze employee feedback data in the feedback optimization results and identify deficiencies in the matching strategy. The data input format of this submodule is an employee feedback data set, including evaluations, suggestions and satisfaction scores for career path planning and skill improvement plans. The operation process first involves using Pandas for data cleaning and preprocessing to convert text feedback into analyzable numerical data. For example, the distribution of employee opinions is quantified by calculating the frequency of different feedback topics. Then, the decision tree model is applied to analyze the processed data, and the model parameters criterion = entropy and max_depth = 3 are set. According to the problems and suggestions in the employee feedback, the decision tree model helps identify which aspects of the matching strategy need to be adjusted. In this way, the model can make specific strategy adjustment suggestions based on actual feedback, such as adjustment of parameters and weights. The generated strategy adjustment analysis data records the required adjustment measures in detail, including the optimization of parameters in the process of specific skill assessment and job matching, providing a basis for achieving more efficient and accurate matching.

[0134] In the reinforcement learning application sub-module, based on the policy adjustment analysis data, the deep Q network (DQN) is used for policy optimization. The deep neural network model is built using the TensorFlow framework, and the learning rate is set to 0.01 and the discount factor is set to 0.95. Through interactive learning with the simulation environment, the policy parameters are adjusted according to the obtained reward signal. During the operation, the model first receives the policy adjustment analysis data as input. The data reflects the direction and goal of the policy adjustment. Then, the model continuously interacts with the simulation environment, performs actions and receives feedback rewards. According to the reward signal, the DQN algorithm is used to update the policy parameters and optimize the matching strategy. In this process, the setting of the learning rate and discount factor ensures that the model can effectively learn and gradually improve the performance of the strategy. The generated optimization strategy draft details the adjusted policy parameters and weights. The purpose is to improve the overall effect of the matching strategy and ensure that the matching between employees and positions is more accurate and efficient.

[0135] In the accuracy improvement submodule, based on the draft optimization strategy, the K-fold cross-validation method is used to verify the effectiveness of the matching strategy. The cross-validation is set to 5-fold, which means that the data set is evenly divided into 5 parts, each part is used as a test set in turn, and the rest are used as training sets to evaluate the accuracy of the model after the strategy adjustment. During the operation, the model calculates the accuracy in each training and testing iteration, and evaluates the overall effectiveness of the strategy by the average accuracy. This method can not only verify the effect of the strategy adjustment, but also help identify the overfitting problem of the model in certain cases. According to the results of the 5-fold cross-validation, the parameters and weight configuration in the matching strategy are further adjusted to achieve the best matching effect. The generated optimized matching strategy records the verified and adjusted strategy parameters in detail, providing a more accurate and effective matching mechanism for the function evaluation and employee position automatic matching system.

[0136] Suppose there is a data set of employee feedback indicating that some skills are overestimated in the skill matching process, resulting in a deviation between the matching results and employee expectations. Through the analysis of the strategy adjustment sub-module, the skill weight parameters that need to be adjusted are identified. The reinforcement learning application sub-module proposes a new skill weight distribution scheme through training and optimization of the DQN model. The accuracy improvement sub-module confirms through 5-fold cross-validation that the new strategy can significantly improve the matching accuracy. Finally, the generated optimized matching strategy adjusts the weight distribution of relevant skills to better reflect the actual skill level and job requirements of employees, thereby improving the matching efficiency and accuracy of the system.

[0137] See also Figure 2 and Fig.10 ,The continuous learning and model updating module includes an online learning submodule, an incremental learning submodule, and a model updating submodule;

[0138] The online learning submodule uses the online gradient descent method based on the optimized matching strategy. By setting the step size to 0.02, the model is updated in real time to reflect the received market data and employee feedback information. The model parameters are gradually adjusted through numerical calculations using the NumPy library to generate online learning update records.

[0139] The incremental learning submodule uses a micro-batch processing method to process new data based on online learning update records, sets the size of each batch of data to 50, uses TensorFlow and Keras to adjust the model, matches the data distribution, determines that the model matches the workplace environment and changes in skill requirements, and generates incremental learning update results;

[0140] The model update submodule updates the results based on incremental learning, records update information, including update time and update content, uses the Git version control tool to manage model versions, and verifies that updates can be recorded and tracked to generate an updated model.

[0141] In the online learning submodule, the online gradient descent method is used to update the matching strategy model in real time to reflect the latest market data and employee feedback information received. The data input format of this submodule is the continuously flowing market data and employee feedback. The data is provided to the model in real time to ensure that the model can adapt to changes in market and employee needs in a timely manner. During the operation, the step size (learning rate) is first set to 0.02. This parameter controls the amplitude of the model parameter update. Numerical calculations are performed through the NumPy library. For each received data, the gradient of the loss function is calculated, and the model parameters are updated according to the step size. This online learning method allows the model to be gradually adjusted while continuously receiving new data to optimize the matching strategy. The generated online learning update record records in detail the changes in model parameters after each data update, providing a basis for the continuous optimization and adaptability of the model.

[0142] In the incremental learning submodule, based on the online learning update record, the micro-batch processing method is used to process the new data to ensure that the model can efficiently adapt to changes in data distribution. The data input format of this submodule is batches of newly collected market data and employee feedback, and the size of each batch of data is set to 50. During the operation, TensorFlow and Keras are used to build and adjust the model. For each micro-batch of data, the model calculates the prediction results through forward propagation, and then updates the model's weights and biases through back propagation. This incremental learning method enables the model to quickly adapt to the characteristics and trends of new data without retraining the entire data set. The generated incremental learning update results record in detail the adjustments to the model after each micro-batch data processing, ensuring that the model is continuously updated to match the latest changes in the workplace environment and skill requirements.

[0143] In the model update submodule, based on the incremental learning update results, the model version update is managed and the update is ensured to be recorded and tracked. During the operation of this submodule, the Git version control tool is used to record the information of each model update, including the update time and update content. This method provides the model with the ability of version management and change tracking, so that each update has a clear record, which is convenient for subsequent analysis and backtracking. Through the Git tool, different versions of the model can be easily managed, the differences between different updates can be compared, and the previous version can be rolled back when necessary. The generated updated model records the detailed information of all update operations.

[0144] Assuming that a batch of new market data and employee feedback indicates that the accuracy of specific skill matching needs to be improved, the online learning sub-module gradually reduces the matching error by updating the model parameters in real time. The incremental learning sub-module further processes the micro-batch data collected subsequently to refine the model's response to new trends. Finally, the model update sub-module records the series of update operations, manages the model version through the Git tool, and generates an optimized matching strategy model that includes the latest market adaptability adjustments. The update ensures that the functional assessment and employee position automatic matching system can respond to market changes and employee needs more accurately and efficiently, thereby improving the overall performance of the system and user satisfaction.

[0145] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A system for automatically matching job functions with employee positions, characterized by: The system includes a skill map construction module, a skill demand analysis module, an employee skill assessment module, a matching degree analysis module, a career development planning module, a feedback collection and analysis module, a matching strategy optimization module, and a continuous learning and model updating module; The skill map construction module uses a graph neural network based on real-time market data to represent skills in a node-based manner, encode and connect skills, analyze the relationships between skills, update the map in real time and reflect market changes, and generate a dynamic skill map. The skill demand analysis module is based on a dynamic skill map and uses natural language processing technology to extract skill keywords from job descriptions, analyze skill occurrence frequency and relevance, and update skill nodes in the map to match skill requirements and generate skill demand data; The employee skill assessment module uses a graph convolutional network based on skill demand data to vectorize the employee's skills, analyze the correlation between the employee's skills and the skill nodes in the graph, assess the employee's skill level, and generate an employee skill assessment record; The matching analysis module uses the BERT model based on the employee skill assessment records to perform feature learning on the employee skill set and job skill requirements, analyze the correlation and compatibility between employee skills and job requirements, and generate matching analysis data; The career development planning module uses a decision tree algorithm based on matching analysis data to analyze employee skill paths and future job development trends according to employee skill levels, and provides guidance on career development and skill planning to generate career and skill improvement plans; The feedback collection and analysis module analyzes employee feedback content based on the career and skill improvement plan by using association rule learning, identifies potential improvement points in the matching strategy by mining frequent item sets and association rules in employee feedback data, and generates feedback optimization results; The matching strategy optimization module uses a reinforcement learning algorithm based on feedback optimization results to refine and adjust the existing employee and position matching strategy, optimizes strategy parameters and weight distribution through a reward mechanism in a simulated environment, and generates an optimized matching strategy; The continuous learning and model updating module adopts online learning and incremental learning algorithms based on the optimized matching strategy, continuously receives market data and feedback information, iteratively updates the model, and matches skill requirements and workplace environment changes to generate an updated model; The employee skills assessment module includes a work experience analysis submodule, an education background comparison submodule, and a skills self-assessment analysis submodule; The work experience analysis submodule uses natural language processing technology to extract keywords from job descriptions based on skill demand data, segment words and remove stop words through the NLTK library, apply the TF-IDF algorithm, calculate keyword weights, and analyze employees' practical experience in the skill field to generate work experience analysis data; The education background comparison submodule uses the cosine similarity function of the Scikit-learn library based on work experience analysis data to compare and analyze education background and market skill requirements, calculate the course and skill requirement data in the employee's education background, and analyze the matching degree with skill keywords to generate background matching analysis results; The skill self-assessment analysis submodule analyzes the employee's self-assessed skill information based on the background matching analysis results, integrates the employee's self-assessment data with the work experience analysis data through the weighted average method, determines the employee's cognition of his or her own skill mastery, and generates an employee skill assessment record; The matching degree analysis module includes a skill matching analysis submodule, a job requirement matching submodule, and a matching degree calculation submodule; The skill matching analysis submodule loads the pre-trained BERT model through the Transformers library based on the employee skill assessment records, inputs the employee skill set and job description text into the model for feature extraction, analyzes the matching degree between skills and job requirements, and generates a preliminary match between skills and jobs; The job demand matching submodule uses the K-means clustering algorithm based on the preliminary matching of skills and jobs, sets three cluster centers, classifies job requirements, highlights the criticality of optimal skills, and assigns differentiated weights to skill groups to generate job demand matching analysis results; The matching degree calculation submodule uses the BERT model to perform feature learning based on the job requirement matching analysis results, compares the correlation between employee skills and job requirements, adjusts the learning rate to 2e-5 through fine-tuning of the model, selects Adam as the optimizer, selects cross entropy as the loss function, performs matching degree calculation, and generates matching degree analysis data.

2. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The skill requirement data includes skill keywords, skill frequency and skill correlation; the employee skill assessment record includes skill level score, skill potential index and skill development plan; the matching analysis data includes similarity score, compatibility evaluation and matching priority; the optimized matching strategy includes weight adjustment value, parameter update list and matching logic improvement measures; the updated model includes performance improvement indicators, efficiency optimization measures and skill matching capabilities.

3. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The skill map construction module includes a market data analysis submodule, a job description analysis submodule, and a map update submodule; The market data analysis submodule uses time series analysis based on real-time market data, uses the Pandas library to perform sliding window processing on the data, sets the window size to 30 days, calculates the frequency of occurrence of skills within the window period, sorts the frequency of occurrence, identifies the skills with the best market demand growth, and generates skill demand trend analysis; The job description analysis submodule is based on skill demand trend analysis, adopts natural language processing technology, uses NLTK library and regular expressions to perform text processing, including word segmentation, stop word removal and stem extraction, and uses TF-IDF algorithm to calculate the weight of skill keywords and generate a job skill keyword list; The graph update submodule is based on the job skill keyword list, uses a graph neural network, builds a graph model through the PyG library, defines a node for the skill, uses edges to connect related nodes according to the correlation in the skill keyword list, sets the node features to the frequency and weight of the skill, updates the graph and includes the identified skills and relationships, and generates a dynamic skill graph.

4. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The skill demand analysis module includes a skill demand identification submodule, a key skill analysis submodule, and a scarce skill identification submodule; The skill demand identification submodule is based on the dynamic skill map and uses the k-means clustering algorithm of the Scikit-learn library. The number of clusters is set to 5. According to the characteristics of skill nodes, including frequency of occurrence and market trends, skills are clustered to identify skill groups at different levels of market demand and generate skill demand identification results. The key skill analysis submodule adopts association rule learning based on the skill requirement identification results, uses the Apriori algorithm of the MLxtend library, sets the minimum support to 0.01 and the minimum confidence to 0.5, analyzes the association rules between different skill groups, identifies the key skills and interdependencies in the industry, and generates key skill analysis records; The in-demand skills identification submodule adopts a decision tree algorithm based on key skill analysis records, uses the Scikit-learn library to build a decision tree model, sets the partitioning criterion to information gain, and the maximum depth to 5, analyzes the relationship between key skills and market demand, identifies skills that are in short supply in the current market, and generates skill demand data.

5. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The career development planning module includes a career path planning submodule, a skill improvement planning submodule, and a career interest matching submodule; The career path planning submodule is based on matching analysis data, uses the Scikit-learn library, adopts a decision tree algorithm, defines the decision tree parameter criterion as entropy, and the maximum depth is 4. By analyzing the employee's skill data and market trend data, the career path of the matching employee is identified and the career path planning result is generated; The skill improvement planning submodule is based on the career path planning results, uses the Scikit-learn library, adopts a support vector machine, sets the kernel function to linear, C to 1.0, balances the error and model complexity, analyzes the difference between the employee's existing skills and the skills required for the target career path, and generates skill improvement planning measures; The career interest matching submodule is based on skill improvement planning measures, uses the Scikit-learn library, applies the K nearest neighbor algorithm, sets the number of neighbors n_neighbors to 5, determines the classification decision basis, and assigns the optimal weight to the distance. It also matches the personal interests of employees with potential career paths and generates career and skill improvement plans.

6. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The feedback collection and analysis module includes a feedback collection submodule, a feedback analysis submodule, and an optimization solution submodule; The feedback collection submodule collects employee feedback based on the career and skill improvement plan, designs a questionnaire using an online questionnaire tool, collects employee feedback on the plan, and counts the feedback frequency and theme to generate an employee feedback data set; The feedback analysis submodule is based on the employee feedback data set, uses the MLxtend library, adopts the association rule learning algorithm, sets the minimum support to 0.02 and the minimum confidence to 0.5, analyzes the patterns and trends in employee feedback, identifies key issues and employee concerns in career planning and skill improvement plans, and generates feedback analysis results; The optimization program submodule formulates the optimization program based on the feedback analysis results and adopts the linear regression model to analyze the relationship between employee satisfaction and the plan, adjust the plan to meet employee needs and market development trends, and generate feedback optimization results.

7. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The matching strategy optimization module includes a strategy adjustment submodule, an enhanced learning application submodule, and an accuracy improvement submodule; The strategy adjustment submodule is based on the feedback optimization results. By analyzing the feedback data, the deficiencies in the matching strategy are identified. The Pandas library is used for data processing, including calculating multiple feedback frequencies, applying the decision tree model, setting the maximum depth to 3, adjusting the parameters and weights in the matching strategy, and generating strategy adjustment analysis data. The reinforcement learning application submodule uses a deep Q network to optimize the strategy based on the strategy adjustment analysis data, builds a neural network model through the TensorFlow framework, sets the learning rate to 0.01, and the discount factor to 0.

95. By interacting with the simulation environment, the strategy parameters are adjusted according to the reward signal to generate an optimization strategy draft; The accuracy improvement submodule uses K-fold cross validation to verify the effectiveness of the matching strategy based on the optimization strategy draft, sets 5-fold cross validation, evaluates the model accuracy after the strategy adjustment, adjusts the parameters and weight configuration in the matching strategy according to the verification results, and generates an optimized matching strategy.

8. The function evaluation and employee position automatic matching system according to claim 1 is characterized by: The continuous learning and model updating module includes an online learning submodule, an incremental learning submodule, and a model updating submodule; The online learning submodule uses an online gradient descent method based on the optimized matching strategy, sets the step size to 0.02, updates the model in real time, reflects the received market data and employee feedback information, performs numerical calculations through the NumPy library, gradually adjusts the model parameters, and generates online learning update records; The incremental learning submodule uses a micro-batch processing method to process new data based on online learning update records, sets the size of each batch of data to 50, uses TensorFlow and Keras to adjust the model, matches the data distribution, determines that the model matches the workplace environment and skill requirements, and generates incremental learning update results; The model update submodule records update information, including update time and update content, based on incremental learning update results, uses Git to manage model versions, verifies that updates can be recorded and tracked, and generates an updated model.

Images