Intelligent human resource training recommendation method and system

Abstract

The invention provides an intelligent human resource training recommendation method and system, and relates to the technical field of data processing, and the method comprises the steps: obtaining post data, skill data, performance data, training records and development target data of a target, building an employee database, and constructing a knowledge graph; post data, skill levels, performance expressions, training records and individual development targets of the employees are integrated to form a comprehensive portrait; dividing all courses according to difficulty levels, selecting courses with corresponding difficulty according to post requirements and performance data of the employees, and generating a group of preliminary course lists matched with current skill levels and post requirements of the employees; semantic analysis is carried out on skill requirements of employee portraits and recommended course contents, a semantic analysis result is obtained, and according to employee portrait data, a semantic matching algorithm is used to calculate the matching degree between the employees and each training content; and according to the matching degree scores of the employees and the training contents, generating a course list subjected to semantic matching optimization.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to an intelligent human resources training recommendation method and system. Background Art

[0002] At present, in traditional human resource management, the recommendation of employee growth and development paths mostly relies on manual judgment, and lacks an efficient and personalized recommendation mechanism. Employees' capabilities, performance and training data are often not fully integrated, and it is impossible to provide employees with real-time and personalized growth paths. As companies have higher and higher requirements for talents, traditional human resource management methods can no longer meet the needs of modern management. Therefore, an intelligent, humanized and efficient recommendation system is needed to provide employees with suitable training and career development paths. Summary of the invention

[0003] The purpose of this application is to provide an intelligent human resources training recommendation method and system, which aims to utilize employees' job information, performance data, skill data, training records, and development goal data to build a comprehensive employee portrait based on knowledge graphs and semantic analysis technology, and generate personalized growth paths and training plans for employees through intelligent recommendation algorithms to improve the adaptability and capabilities of employees in their positions.

[0004] Based on this, this application provides an intelligent human resources training recommendation method and system, and its specific scheme is as follows: Obtain the target's position data, skill data, performance data, training records, and development goal data, clean the original data, remove redundant and incomplete data, and unify data in different formats into a standardized format to establish an employee database; Build a knowledge graph based on the data in the employee database; Based on the knowledge graph, employees’ job data, skill level, performance, training records and personal development goals are integrated into a comprehensive portrait; Divide all courses by difficulty level and select courses of corresponding difficulty based on employees’ job requirements and performance data to generate a preliminary list of courses that match employees’ current skill levels and job requirements; Perform semantic analysis on the skill requirements and recommended course content of employee portraits, extract keywords from them, and obtain semantic analysis results; Based on the results of semantic analysis and employee portrait data, a semantic matching algorithm is used to calculate the matching degree between employees and various training contents; Generate a course list optimized for semantic matching based on the matching scores between employees and various training contents.

[0005] The method for constructing the knowledge graph includes: Determine the core nodes in the knowledge graph, that is, create nodes such as positions, skills, performance, training, and development goals in the knowledge graph. Each type of node corresponds to the multi-dimensional attributes of employees, such as the skill requirements of the position and the indicator scores of performance data; By setting the association relationship between different nodes to form a knowledge graph structure, you can use triples to build a knowledge graph and associate all information into a networked graph structure; Vectorize each node and its attributes so that they can be directly used in subsequent semantic matching and recommendation algorithms; The method for obtaining the comprehensive portrait of the employee includes: Through the knowledge graph, skill data and performance data are combined to assess the depth of employees’ skills and mark them in the portrait; Compare job requirements and training records in the knowledge graph to obtain employee training needs; Match employees' career development goals with job requirements to form personalized growth goals; A comprehensive profile of the employee is generated, which includes the depth of skill mastery, key performance items, training needs and development goals.

[0006] The method for generating the preliminary course list is as follows: Get the set of courses to be trained based on training needs: Divide all courses in the set of courses to be trained by difficulty level, obtain recommended training courses based on the employee's skill depth and generate a preliminary course list; the preliminary course list includes specific courses and their difficulty levels; The method for obtaining the semantic analysis result includes: Conduct semantic analysis on the skill depth of employee profiles and the content of the preliminary course list to extract keywords; The semantic matching algorithm is used to calculate the matching degree between the skill depth in the employee profile and each course in the preliminary course list, and generate a matching score for each course, ensuring the accuracy and personalization of the recommended content.

[0007] Furthermore, the method further comprises: Build a personalized recommendation model based on employee portraits, performance, job requirements, semantic matching scores, etc.: By combining the course recommendation list and development goals, personalized employee growth paths and development recommendations are generated to achieve the dual goals of job requirements and personal growth.

[0008] Furthermore, the method further comprises: Record employees' learning progress and periodic performance; Dynamically adjust recommended courses and learning paths based on employees’ latest performance and learning data, and update future learning paths and recommended content to ensure that employees receive the latest learning content that suits their growth needs; Furthermore, the method further comprises: Obtain employees' learning feedback and training effect evaluation; Based on feedback data, optimize and adjust the recommendation engine to gradually improve the accuracy of recommended content; Obtain optimized recommendation strategies and personalized recommendation results; An intelligent human resources training recommendation system, comprising: Data management module: The data management module includes: Data collection unit: used to collect employees' job data, skill data, performance data, training records and development goal data, and store them in the employee database; Knowledge graph construction module: The knowledge graph construction module includes: Node determination unit: identifies and defines the core nodes in the knowledge graph; Relationship setting unit: establishes associations between nodes, defines the matching relationship between employee skills and job requirements, and the priority of training needs; Vectorized representation unit: vectorizes each node and its attributes for subsequent calculation and matching; Employee portrait construction module: The employee portrait construction module includes: Skill level integration unit: Combines employee skill data with performance data to assess the depth of skill mastery and records the results in the employee profile; Training needs analysis unit: compares employees’ job requirements with their current skill levels, generates a list of training needs, and marks the skills that need to be trained first; Development goal analysis unit: connect employees' career development goals with job requirements, form personalized growth goals and integrate them into employee portraits; Course screening module: the course screening module includes: Course Grading Unit: All courses are divided into difficulty levels; Difficulty screening unit: selects courses of corresponding difficulty based on the employee's job requirements and performance data, and generates a preliminary list of courses that match the employee's current skill level and job requirements; Semantic matching and optimization module: The semantic matching and optimization module includes: Semantic analysis unit: performs semantic analysis on the skill requirements and course descriptions in employee portraits, extracts keywords, and obtains semantic analysis results; Matching calculation unit: Based on the semantic analysis results, it calculates the matching degree between the employee's skill requirements and each course in the preliminary course list, and generates a matching score for each course; Generate a course list optimized for semantic matching based on the matching scores between employees and various training contents; Intelligent recommendation engine module: The intelligent recommendation engine module includes: Recommendation algorithm design unit: Use performance-driven screening and semantic matching to optimize the course list and generate personalized recommendation plans based on the data of various dimensions in the employee portrait; Growth path generation unit: Combines employees' career development goals and job requirements to generate personalized growth paths, ensuring that the recommended content supports the long-term development of employees. The final course recommendation list and growth path include courses suitable for employees, course difficulty and learning order; Dynamic learning path management module: The dynamic learning path management module includes: Learning progress monitoring unit: used to record employees’ learning progress and training results, and monitor employees’ growth progress in real time; Path adjustment unit: It is used to dynamically optimize the recommended course content and learning path based on the real-time learning status and latest performance of employees, ensuring that the recommended plan meets the progress needs of employees; Feedback optimization module: The feedback optimization module includes: Feedback collection unit: collects employees' training feedback and effect evaluation data.

[0009] Recommendation strategy adjustment unit: Use feedback data to optimize the recommendation system, improve the recommendation algorithm and matching strategy, so as to enhance the accuracy and applicability of the recommended content.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages: 1. Knowledge graph-driven comprehensive portrait: This method uses knowledge graph technology to integrate employees’ multi-dimensional information into a comprehensive portrait, greatly improving the data understanding depth and accuracy of the recommendation system.

[0011] 2. Application of semantic analysis and semantic matching algorithms: Semantic analysis technology provides a deep semantic understanding of employee needs and course content, providing a new dimension for matching employee skill needs and training content, breaking the traditional simple rule matching limitations.

[0012] 3. Personalized dynamic growth path: Through the dynamic learning path management module, the system can provide dynamically adjusted growth paths and training plans based on employees' real-time learning progress and latest performance, making the employees' learning process more coherent and scientific.

[0013] 4. Feedback optimization mechanism: The system obtains learning feedback in real time through the feedback optimization module, continuously adjusts the recommendation strategy, ensures dynamic optimization and updating of recommended content, and enables the recommendation system to continuously adapt to changes in employees' growth needs.

[0014] In summary, this method is based on the intelligent human resource growth recommendation system of knowledge graph, semantic analysis, dynamic adjustment and feedback optimization, which can significantly improve the recommendation accuracy and personalization. Compared with the existing technology, this system can better respond to the dynamic growth needs of employees, help employees improve their skills and career development, and has significant application value.

[0015] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0017] Figure 1 A flowchart of an intelligent human resource training recommendation method for this application; Figure 2 A schematic diagram of the structure of an intelligent human resource training recommendation system for this application; Explanation of the accompanying drawings: 11. Data management module, 12. Knowledge graph construction module, 13. Employee portrait construction module, 14. Course screening module, 15. Semantic matching and optimization module, 16. Intelligent recommendation engine module, 17. Dynamic learning path management module, 18. Feedback optimization module. DETAILED DESCRIPTION

[0018] This application provides an intelligent human resource training recommendation method and system, and an intelligent human resource growth recommendation system based on knowledge graph, semantic analysis, dynamic adjustment and feedback optimization, which can significantly improve the recommendation accuracy and personalization. Compared with the existing technology, this system can better respond to the dynamic growth needs of employees, help employees improve their skills and career development, and improve the adaptability and ability of employees in their positions.

[0019] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0020] Example 1, please refer to the attached Figure 1 The present application provides an intelligent human resource training recommendation method and system, wherein the intelligent human resource training recommendation method includes: An intelligent human resource growth recommendation method, comprising: S100 obtains the target’s job data, skill data, performance data, training records, and development goal data to establish an employee database; Its main goal is to obtain various employee data in real time or regularly from multiple data sources inside and outside the system, including job data, skill data, performance data, training records, development goal data, etc.; in: Position data: including the skill requirements and key task indicators of the employee’s current position; Skills data: including hard skills and soft skills data, usually provided by skills assessment and evaluation systems; Performance data: derived from the performance management system, covering employee performance, goal achievement, etc.; Training records: covering employees’ training experience, courses participated, assessment results, etc., from the company’s internal training system or external training platform; Development goal data: usually through questionnaires filled out by employees and interview records, reflecting employees' career development intentions and learning preferences; The data can be collected through the following means: API data interface: Establish API interface with the company's performance management system, training system, HR management system, etc. to pull data regularly or in real time.

[0021] File import: Some external data (such as records from external training platforms) can be imported through files (such as Excel, CSV, etc.).

[0022] Questionnaire and manual input: Employee development intention data can be supplemented through a questionnaire collection system or manual input by HR.

[0023] Among them, the S100 also includes: cleaning and standardizing the collected raw data to ensure the consistency and quality of the data and provide high-quality input for subsequent portrait generation and analysis: The specific steps are as follows: Data cleaning Deduplication: Delete duplicate records to avoid data redundancy caused by multiple data collection.

[0024] Missing value processing: For missing data, take methods such as deletion, filling (such as filling with the mean), interpolation or data repair.

[0025] Outlier detection and processing: Detect outliers through statistical analysis or setting thresholds, and eliminate or correct obviously unreasonable data.

[0026] Data Standardization Data format unification: Convert data from different sources into a unified format, such as date format, numerical unit, etc.

[0027] Field mapping and labeling: Map fields representing the same information in different sources, such as unifying "course name" into "training program" and adding labels to qualitative data to facilitate subsequent classification.

[0028] Data normalization: This includes normalization and scaling of numerical data (such as unifying employee ratings into a range of 1 to 5 points) to ensure that data of different dimensions are analyzed on the same scale.

[0029] It can use: ETL tools: such as Apache NiFi and Talend for data extraction, transformation and loading.

[0030] Use Python, SQL, etc. to write cleaning and standardization scripts to automatically complete the cleaning process.

[0031] Use statistical analysis, box plots, or Z-scores to detect outliers in your data.

[0032] S200 builds a knowledge graph based on the data in the employee database, that is, extracts relevant data from the employee database and organizes it into nodes and relationships in the knowledge graph, so that the employee's position, skills, performance, training and development goal data can be organically integrated and vectorized to support intelligent analysis of the recommendation engine. This graph-structured information expression method enhances the accuracy and personalization of recommendations; The method for constructing the knowledge graph includes: S210 determines the core nodes in the knowledge graph: Based on business needs and data types, determine the core node categories in the knowledge graph, including but not limited to the following categories: Position node: represents the specific information of various positions, such as position title, job description, required skills, ability requirements, etc.

[0033] Skill node: represents the various skills possessed by employees and their corresponding mastery levels.

[0034] Performance Node: represents the key performance of employees at work and is related to specific skills or job requirements.

[0035] Training course node: represents the subject, difficulty and related skills of various training courses in the system.

[0036] Development goal node: represents the employee’s long-term career goals and their relevance to job requirements.

[0037] Node attribute definition: Define core attributes for each node to ensure that the graph can cover multi-dimensional information.

[0038] Position node attributes: position title, main responsibilities, key skills, ability level, etc.

[0039] Skill node attributes: skill name, mastery level, update date, etc.

[0040] Performance node attributes: performance indicators, evaluation date, score, etc.

[0041] Training course node attributes: course name, skills involved, difficulty level, etc.

[0042] Development goal node attributes: goal name, time limit for realization, related positions and skills, etc.

[0043] S220 sets the association relationship between different nodes to form a knowledge graph structure; That is, determine the type of relationship between nodes: define the main relationship types in the knowledge graph based on the system functional requirements. For example: Position-skill relationship: The relationship between a position node and a skill node. This relationship is used to describe the core skills required for a position.

[0044] Employee-Performance Relationship: The relationship between an employee node and a performance node, indicating the employee's performance in a specific skill.

[0045] Employee-Training Relationship: The relationship between an employee node and a training course node, indicating the training courses that the employee has completed or recommended.

[0046] Employee-development goal relationship: The relationship between the employee node and the development goal node, describing the degree of match between the employee's current career development goal and related positions and skills.

[0047] Skill-Course Relationship: The association between a skill node and a training course node, indicating the skills involved in a specific course.

[0048] Relationship direction and weight: Define the direction (e.g., from position to skill) and weight for each relationship. The weight can be determined through data analysis to reflect the strength of the relationship. For example, the core skills of a position have a higher weight, while the secondary skills have a lower weight.

[0049] Based on nodes and relationships, various nodes are connected into a network structure through relationships. This structure is implemented using a graph database (such as Neo4j).

[0050] Based on the existing data in the database, build relationships in batches to ensure the efficiency and accuracy of knowledge graph construction. For example: Use SQL or graph database scripts to automatically extract the relationships between job data and skill data and add them to the graph.

[0051] The employee's performance data and training records are processed to generate the association between the employee node and the performance and training course nodes.

[0052] S230 vectorizes each node and its attributes; That is, the node information is converted into vectors, so that the nodes in the knowledge graph can participate in calculation and analysis in numerical form. The vectorized data can be used for similarity calculation and recommendation through semantic matching algorithms; For text attributes in nodes (such as job descriptions and skill names), use word embedding models (such as Word2Vec and BERT) to convert them into vector representations. This step requires training so that the model can understand semantic content such as job positions and skills; For numerical attributes (such as performance scores and skill levels), the numerical values ​​are directly retained and normalized so that they occupy the corresponding positions in the vector; Merge text embedding and numerical embedding to form a complete node vector. For example, the vector of a job node can include its job title, key skill embedding, job level and other numerical information; According to the weights of different relationships, a specific algorithm (such as distance or similarity calculation) is used to assign vector weights between different relationships. Weight vectorization can be used to adjust relevance in recommendations; Combine the node vector and the relationship weight to generate a relationship feature vector for analysis. For example, combine the position and skill node vectors in the "position-skill" relationship to form a new vector for matching analysis between positions and employee skills; In addition, the step S200 further includes: Regularly update job information, performance data, skill data, etc. from the employee database to ensure the dynamic nature of the graph; Use automated scripts to regularly clean up nodes and relationships that are no longer valid, ensuring that the data in the graph always maintains high quality. Introducing graph embedding technology to embed the entire knowledge graph into a low-dimensional vector space to improve the efficiency of large-scale data computing while retaining the network structure information of the knowledge graph; Based on the knowledge graph, S300 integrates employees’ job data, skill level, performance, training records and personal development goals into a comprehensive portrait to support subsequent intelligent recommendations and personalized growth path planning; The method for obtaining the comprehensive portrait of the employee includes: S310 combines skill data and performance data through knowledge graphs to assess the depth of employees’ skills and mark them in the portrait; That is, find the employee node and its associated skill node from the knowledge graph, and obtain the skill name, mastery level, update time, etc. Obtain performance data related to employee nodes, including performance scores, specific performance items for related skills or positions, and other information; Generate employee skill depth by calculating skill mastery, frequency (number of times used or evaluated), time dimension (time of most recent evaluation), and other information; Combined with the specific performance items of the skill, the score of the corresponding skill in the performance data is included in the profile together with the skill mastery level. For example, the performance score is used as a weighting factor for the depth of the skill.

[0053] Weighting method: skill mastery and performance score are weighted by a certain weight to obtain a comprehensive score of skill depth (e.g. skill depth = 0.6 skill mastery + 0.4 performance score); Mark the depth of employees' skills in each skill into a comprehensive profile to facilitate subsequent recommendation analysis; S320 compares job requirements and training records in the knowledge graph to obtain employee training needs; Extract the skill requirements related to the employee's current position node from the knowledge graph, including essential skills and bonus skills (divided by importance). For example, if the employee's position is "data analyst", the skill requirements may include key skills such as data analysis, programming, and statistics; At the same time, extract the training requirements corresponding to the position and obtain the training records and goals in the position standards; Extract the training course nodes associated with the employee nodes from the knowledge graph to obtain information such as completed training courses, course content, difficulty level, and performance evaluation; Combined with the learning results, learning feedback and other information in the training records, analyze the training effect and determine whether repeated training or related training to further improve the skills is needed; Compare the skill list required for the position with the employee's existing skill depth to identify skill gaps. For example, if the position requires high programming skills but the employee's programming skills are insufficient, identify the need for training in this skill; Compare the employee's completed training records with the training records required for the position to identify any necessary training courses that have not been completed in order to generate a training needs list; Add identified training needs to employee profiles to label uncompleted required training courses and skill gaps.

[0054] S330 aligns employees’ career development goals with job requirements to form personalized growth goals; Extract the career development goal node of the employee node from the knowledge graph, including target position, required time, expected skill level improvement and other information; Standardize the keywords and content in development goals to ensure they match job requirements; Compare the employee's current position data with the target position data to determine the skills and performance that the employee should improve on the basis of the existing position, so as to tailor a growth path for the employee; According to the employees' career goals, the skills, performance, and training requirements of the target positions are extracted from the knowledge graph, and these requirements are connected with the current job requirements and employee skill data to form a personalized growth path; Include a generated growth path in the portrait, including specific skill development and training tasks to be completed.

[0055] S340 generates a comprehensive employee profile, which includes the depth of skill mastery, key performance items, training needs and development goals.

[0056] Integrate the data extracted and analyzed in S310, S320, and S330 into a comprehensive portrait; Depth of Skill Mastery: Note the depth of each skill and recent performance.

[0057] Key Performance Items: Displays the employee’s main performance and scores in each performance item.

[0058] Training needs: List outstanding job training courses, as well as necessary training courses identified based on the skills gap analysis.

[0059] Development goals: Clearly mark the target positions and skills that need to be improved included in the personalized growth path.

[0060] Display comprehensive employee portraits through graphical interfaces, including skill radar charts, performance bar charts, training needs lists, and growth path charts; According to the image format, the image is output as a JSON or XML file in the system standard format to facilitate calling by other modules; Wherein, the step 300 further includes: Set up periodic updates to automatically update the profile when employee data changes (such as skill improvement, completion of new training); Further analyze other related information in the graph, such as job changes, portraits of people in similar positions, etc., to improve the personalization and accuracy of the portrait; By comparing employees’ actual performance with the recommended portrait results, we continuously adjust the weight distribution of the portrait generation algorithm and the accuracy of the matching algorithm. S400 divides all courses into difficulty levels and selects courses of corresponding difficulty based on employees’ job requirements and performance data, generating a preliminary course list that matches employees’ current skill levels and job requirements. That is, by grading all training courses by difficulty, it recommends the most suitable courses for employees to learn based on their current skill depth and job requirements, and builds a preliminary course list, laying the foundation for subsequent recommendation optimization and personalized growth paths. The method for generating the preliminary course list is as follows: S410 obtains the set of courses to be trained based on training needs: Extract basic information of all courses from the system database or knowledge graph, including course name, course content, course duration, difficulty level, applicable positions, skill objectives, etc.

[0061] Categorize courses by subject or technical area (such as programming, project management, communication skills, etc.) to facilitate subsequent screening based on skill requirements; Using the training demand data generated in step S320, extract the skills that employees need to improve and the corresponding training needs; Combine the job requirements, skill requirements and training records in the employee profile to select courses related to the employee's skill improvement needs and form a list of courses to be trained; S420 divides all courses in the set of courses to be trained into difficulty levels, obtains recommended training courses based on the skill depth of the employee, and generates a preliminary course list; Classify courses into different difficulty levels, such as beginner, intermediate, and advanced. Each difficulty level should have clear criteria so that appropriate courses can be selected based on the employee's skill depth when making recommendations.

[0062] Primary course: It is aimed at employees who have no basic knowledge or are new to this skill, and provides basic concepts and entry-level operations.

[0063] Intermediate Course: Suitable for employees who have mastered certain skills in depth. Content includes advanced skills, practical cases and advanced courses.

[0064] Advanced courses: Provide expert, innovative or interdisciplinary course content for employees who already have deep skills.

[0065] Based on the skill depth in the employee profile, select the course difficulty level that matches the employee's current level; Combine the employee's job skill requirements and the skill performance in the performance data to match the difficulty level with the employee's skill requirements to ensure that the course difficulty is suitable for the employee's current learning needs and job goals; From the screened list of courses, select courses that are suitable for the employee's skill level and job requirements to form a preliminary course list.

[0066] It can control the length of the preliminary course list according to the system settings or user needs, for example, generating 5-10 courses each time, so that employees can choose courses in a targeted manner during training; Add labels to each course to mark the skill improvement direction of the course (such as "data analysis", "project management", etc.) to facilitate subsequent recommendation analysis and screening; Output the generated course list to the system for further screening and optimization by the intelligent recommendation engine; S500 performs semantic analysis on the skill requirements of employee portraits and the content of recommended courses, extracts keywords from them, and obtains semantic analysis results, that is, it performs semantic analysis on the skill depth of employee portraits and the content in the preliminary course list, extracts keywords from them, that is, extracts keywords from employee portraits and recommended course content through semantic analysis, and makes the information such as skill requirements and ability depth implicit in the text explicit, thereby supporting subsequent semantic matching calculations; The details are as follows: Extract skill-related text description data from employees’ job requirements, skill levels, key performance items, and training needs to form an employee portrait text dataset; Extract the text data of course description from the course content, difficulty label and course objectives in the preliminary course list to form a course content text dataset; Choose appropriate natural language processing (NLP) techniques to process text data, such as: Word segmentation: Split text into keywords or phrases.

[0067] Keyword extraction: Extract keywords related to skill requirements and course content through methods such as TF-IDF (term frequency-inverse document frequency) or LDA (latent Dirichlet allocation).

[0068] Word vector model: Use pre-trained word vector models such as BERT, Word2Vec, and GloVe to convert keywords into vector representations for subsequent matching calculations.

[0069] Perform semantic analysis on the skill requirements and job requirements in employee profile data to extract keywords that represent employees’ current skill requirements and depth; Conduct semantic analysis on the course content in the preliminary course list to extract the core keywords of the courses; Use the keywords obtained from the employee portrait data analysis as employee skill requirement labels; Use the keywords obtained from the course content analysis as descriptive tags for the course; Based on the results of semantic analysis and employee profile data, S600 uses a semantic matching algorithm to calculate the matching degree between employees and various training contents. In other words, the semantic matching algorithm is used to calculate the matching degree between the skill depth in the employee profile and each course in the preliminary course list, and a matching score is generated for each course to select the most suitable training content. Select the appropriate semantic matching algorithm based on your needs: Cosine similarity: Quantify the matching degree by calculating the cosine similarity between employee profile and course content vector.

[0070] Word vector model: Use deep learning models such as Word2Vec and BERT to calculate the semantic similarity between keywords.

[0071] Similarity scoring criteria: According to the matching score range (such as between 0 and 1), set different matching scores for subsequent screening.

[0072] Convert the keyword set in the employee profile into a vector representation; Convert the set of keywords in each course content into vector representation; Calculate the similarity score between each course content and employee profile data and generate a matching score table. The specific steps are as follows: The similarity between each course content and employee profile is calculated pair by pair to generate a similarity score.

[0073] If there are multiple skill requirement keywords in the employee's profile, the matching degree is calculated comprehensively by weighted average or other methods.

[0074] Generate a matching score between the employee and each course, for example, normalize the similarity score to 0.0-1.0.

[0075] Generate a matching score table between the employee and each course, indicating the suitability of each course for the employee. This score table is used to select the best course in the subsequent steps; S700 generates a course list optimized by semantic matching based on the matching scores between employees and various training contents. That is, based on the matching scores, the most suitable course contents are screened out and an optimized course list is generated, so that the recommended training contents are more in line with the needs of employees. Set a minimum match score, for example, 0.6 or above, to be included in the final optimized course list to ensure that the recommended courses are highly relevant; Sort the courses by matching scores, filter from high to low, and give priority to recommending the course content that best matches the employee profile; Generate an optimized course list based on the screening criteria, including training courses that best meet the needs of employees; Add matching scores and core keyword tags to each course so that the system can use them in subsequent recommendation engines; The optimized course list is stored in the system database as personalized recommendation content for employees; The optimized course list is transmitted to the system's intelligent recommendation engine module to recommend the final training courses for employees; After an employee completes a course, the matching score is dynamically updated by updating the skill depth data in the portrait to adapt to the employee's growth needs.

[0076] And based on employees' learning feedback, the semantic matching threshold is adjusted in real time to ensure the accuracy of recommended content.

[0077] If new courses are added to the training system, the course vector and matching degree will be automatically updated to ensure that the recommended course content meets the latest requirements.

[0078] S800 builds a personalized recommendation model based on employee portraits, performance, job requirements, semantic matching scores, etc.: Generate personalized employee growth paths and development recommendations by combining course recommendation lists and development goals; The specific steps are as follows: Collect and integrate the following data: Employee profile data: includes information such as employee job requirements, skill depth, training needs, and career development goals.

[0079] Performance data: Obtain employees’ historical performance from their assessment records, KPI data, etc.

[0080] Job requirement data: Extract key skills and ability indicators required for the job.

[0081] Semantic matching score: The matching score between the employee profile and the course content calculated previously is used as a measure of course priority; The specific method for determining the personalized recommendation model structure is as follows; Input layer: Input characteristic data such as employee profile, performance, job requirements and matching scores.

[0082] Feature weighting processing: Set weights for each feature and assign different weights based on business needs and actual analysis (e.g. performance and job requirements have higher weights).

[0083] Deep learning / recommendation algorithm: Select appropriate recommendation algorithms (such as collaborative filtering, deep neural network models, tree-based models, etc.) to process data and output personalized recommendation lists; Generate a course recommendation list for employees based on the model output, which includes course name, difficulty, priority, etc. Combine employees' career development goals to build employee growth paths, match recommended courses with employees' development needs, and form a phased learning plan; Output a personalized course recommendation list and store it in the system database. The generated growth path can be further subdivided into short-term, medium-term and long-term goals so that employees can gradually achieve their goals according to their needs; S900 records employees’ learning progress and staged performance, that is, by dynamically tracking employees’ learning progress and staged performance, it dynamically adjusts recommended courses and learning paths based on employees’ latest performance and learning data, and updates future learning paths and recommended content, providing data support for subsequent recommendation adjustments and learning path optimization; The specific steps are as follows: Record the start time, progress and completion status of each course that employees participate in to form learning progress data; Update employees' periodic performance according to their performance appraisal data at a specific stage and generate updated performance records; Store learning progress and performance data in the database as historical learning data of employees; Based on the courses and training results completed by employees, analyze whether the employees' learning outcomes have met expectations and form interim performance data; Store learning progress and performance in the system database for subsequent reference; Transmit the latest learning data to the recommendation engine module to provide real-time data support for the dynamic adjustment of the recommendation engine; S1000 obtains employees’ learning feedback and training effect evaluation; Based on feedback data, optimize and adjust the recommendation engine to gradually improve the accuracy of recommended content; Obtain optimized recommendation strategies and personalized recommendation results; The specific steps are as follows: After employees complete each training session, collect their evaluation of the training effectiveness, including the applicability of the course content, the difficulty of learning, and the overall satisfaction with the training; Collect employee training effectiveness data through assessment or evaluation, such as course pass rate, test scores, etc., to further improve feedback data; Clean the collected feedback data, remove invalid data and outliers, and ensure the reliability of the data; Quantify employee feedback data, for example, using a scoring mechanism (e.g., 1-5 points) to record employee feedback and provide usable feedback data for the recommendation engine; Based on the feedback obtained from the analysis, adjust the algorithm parameters in the recommendation engine. You can adjust the weights of similarity calculation and recommendation priority to ensure that the system considers more employees' actual needs and experiences when making recommendations; If employees give poor feedback on a certain type of course, the priority of this type of course in the recommendation can be lowered; if the feedback score of a certain type of course is high, the priority can be increased; Output new recommendation strategies based on the optimized recommendation engine, making subsequent recommendations more personalized and accurate; Generate new course recommendation lists and growth paths to ensure that the recommendations meet the latest needs and development directions of employees; Store new recommendation strategies and personalized recommendation results in the database for reference in subsequent recommendations; Preserve the connection between employees’ historical feedback and recommended strategies to form a closed feedback loop.

[0084] Embodiment 2: An intelligent human resources training recommendation system, comprising: Data management module 11: the data management module includes: Data collection unit: used to collect employees' job data, skill data, performance data, training records and development goal data, and store them in the employee database; De-duplicate, standardize and process the collected data to obtain standardized employee positions, skills, performance, training records and development goal data, laying the foundation for subsequent knowledge graph construction and data analysis; Knowledge graph construction module 12: used to obtain the knowledge graph structure, including nodes, relationships and vectorized representations thereof, the knowledge graph construction module includes; Node determination unit: Identify and define core nodes in the knowledge graph, such as positions, skills, training content, performance, and development goals; Relationship setting unit: establishes associations between nodes, defines the matching relationship between employee skills and job requirements, and the priority of training needs; Vectorized representation unit: vectorizes each node and its attributes for subsequent calculation and matching; Employee portrait construction module 13: The employee portrait construction module includes: Skill level integration unit: Combines employee skill data with performance data to assess the depth of skill mastery and records the results in the employee profile; Training needs analysis unit: compares employees’ job requirements with their current skill levels, generates a list of training needs, and marks the skills that need to be trained first; Development Goal Analysis Unit: aligns employees' career development goals with job requirements, forms personalized growth goals and integrates them into employee profiles. The employee comprehensive profile includes skill mastery depth, key performance items, training needs and development goals; Course screening module 14: the course screening module includes; Course Grading Unit: All courses are divided into "Basic", "Intermediate" and "Advanced" according to the difficulty level, and courses of corresponding difficulty are recommended based on employee performance data; Difficulty screening unit: selects courses of corresponding difficulty based on the employee's job requirements and performance data, and generates a preliminary list of courses that match the employee's current skill level and job requirements; Semantic matching and optimization module 15: The semantic matching and optimization module includes: Semantic analysis unit: performs semantic analysis on the skill requirements and course descriptions in employee portraits, extracts keywords, and obtains semantic analysis results; Matching calculation unit: Based on the semantic analysis results, it calculates the matching degree between the employee's skill requirements and each course in the preliminary course list, and generates a matching score for each course; Generate a course list optimized for semantic matching based on the matching scores between employees and various training contents; Intelligent recommendation engine module 16: The intelligent recommendation engine module includes: Recommendation algorithm design unit: Use performance-driven screening and semantic matching to optimize the course list and generate personalized recommendation plans based on the data of various dimensions in the employee portrait; Growth path generation unit: Combines employees' career development goals and job requirements to generate personalized growth paths, ensuring that the recommended content supports the long-term development of employees. The final course recommendation list and growth path include courses suitable for employees, course difficulty and learning order; Dynamic learning path management module 17: The dynamic learning path management module includes: Learning progress monitoring unit: used to record employees’ learning progress and training results, and monitor employees’ growth progress in real time; Path adjustment unit: It is used to dynamically optimize the recommended course content and learning path based on the real-time learning status and latest performance of employees, ensuring that the recommended plan meets the progress needs of employees; Feedback optimization module 18: The feedback optimization module includes: Feedback collection unit: collects employees' training feedback and effect evaluation data.

[0085] Recommendation strategy adjustment unit: Use feedback data to optimize the recommendation system, improve the recommendation algorithm and matching strategy, so as to enhance the accuracy and applicability of the recommended content.

[0086] Any of the above methods or steps can be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs can be recognized by various types of computer processors to implement any of the above methods or steps; Secondly, the above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. An intelligent human resources training recommendation method, characterized in that: include: Obtain target position data, skill data, performance data, training records, development goal data, and establish an employee database; Build a knowledge graph based on the data in the employee database; Based on the knowledge graph, employees’ job data, skill level, performance, training records and personal development goals are integrated into a comprehensive portrait; Divide all courses into difficulty levels and select courses of corresponding difficulty based on employees’ job requirements and performance data to generate a preliminary list of courses that match employees’ current skill levels and job requirements; Perform semantic analysis on the skill requirements and recommended course content of employee portraits, extract keywords from them, and obtain semantic analysis results; Based on the results of semantic analysis and employee portrait data, a semantic matching algorithm is used to calculate the matching degree between employees and various training contents; Generate a course list optimized for semantic matching based on the matching scores between employees and various training contents.

2. The intelligent human resource training recommendation method according to claim 1, characterized in that: The method for constructing the knowledge graph includes: Identify the core nodes in the knowledge graph: Set the relationship between different nodes to form a knowledge graph structure; Each node and its attributes are vectorized.

3. The intelligent human resource training recommendation method according to claim 2, characterized in that: The method for obtaining the comprehensive portrait of the employee includes: Through the knowledge graph, skill data and performance data are combined to assess the depth of employees’ skills and mark them in the portrait; Compare job requirements and training records in the knowledge graph to obtain employee training needs; Match employees' career development goals with job requirements to form personalized growth goals; A comprehensive profile of the employee is generated, which includes the depth of skill mastery, key performance items, training needs and development goals.

4. The intelligent human resource training recommendation method according to claim 1, characterized in that: The method for generating the preliminary course list is as follows: Obtain a set of courses to be trained based on training needs; Divide all courses in the set of courses to be trained by difficulty level, obtain recommended training courses based on the employee's skill depth and generate a preliminary course list.

5. The intelligent human resource training recommendation method according to claim 1, characterized in that: Methods for obtaining semantic analysis results include: Conduct semantic analysis on the skill depth of employee profiles and the content of the preliminary course list to extract keywords; The semantic matching algorithm is used to calculate the matching degree between the skill depth in the employee profile and each course in the preliminary course list, and generate a matching score for each course.

6. The intelligent human resource training recommendation method according to claim 1, characterized in that: The method further comprises: Build a personalized recommendation model based on employee portraits, performance, job requirements, semantic matching scores, etc.: By combining the course recommendation list and development goals, personalized employee growth path and development recommendations are generated.

7. The intelligent human resource training recommendation method according to claim 1, characterized in that: The method further comprises: Record employees' learning progress and periodic performance; Dynamically adjust recommended courses and learning paths based on employees’ latest performance and learning data, and update future learning paths and recommendations.

8. The intelligent human resource training recommendation method according to claim 1, characterized in that: The method further comprises: Obtain employees' learning feedback and training effect evaluation; Based on feedback data, optimize and adjust the recommendation engine to gradually improve the accuracy of recommended content; Obtain optimized recommendation strategies and personalized recommendation results.

9. An intelligent human resources training recommendation system, the system being used to execute the method according to any one of claims 1 to 8, characterized in that: include; Data management module: The data management module includes: Data collection unit: used to collect employees' job data, skill data, performance data, training records and development goal data, and store them in the employee database; Knowledge graph construction module: The knowledge graph construction module includes: Node determination unit: identifies and defines the core nodes in the knowledge graph; Relationship setting unit: establishes associations between nodes, defines the matching relationship between employee skills and job requirements, and the priority of training needs; Vectorized representation unit: vectorizes each node and its attributes for subsequent calculation and matching; Employee portrait construction module: The employee portrait construction module includes: Skill level integration unit: Combines employee skill data with performance data to assess the depth of skill mastery and records the results in the employee profile; Training needs analysis unit: compares employees’ job requirements with their current skill levels, generates a list of training needs, and marks the skills that need to be trained first; Development goal analysis unit: connect employees' career development goals with job requirements, form personalized growth goals and integrate them into employee portraits; Course screening module: the course screening module includes: Course Grading Unit: All courses are divided into difficulty levels; Difficulty screening unit: selects courses of corresponding difficulty based on the employee's job requirements and performance data, and generates a preliminary list of courses that match the employee's current skill level and job requirements; Semantic matching and optimization module: The semantic matching and optimization module includes: Semantic analysis unit: performs semantic analysis on the skill requirements and course descriptions in employee portraits, extracts keywords, and obtains semantic analysis results; Matching calculation unit: Based on the semantic analysis results, it calculates the matching degree between the employee's skill requirements and each course in the preliminary course list, and generates a matching score for each course; Generate a course list optimized for semantic matching based on the matching scores between employees and various training contents.

Images