Enterprise human resource intelligent recruitment and talent management system and method

Abstract

The invention discloses an enterprise human resource intelligent recruitment and talent management system, and relates to the technical field of enterprise recruitment and management. A BERT model is adopted for word segmentation, 12 types of key indexes are extracted, self-defined weights are supported, position portraits containing 6-8 core requirements are generated, resumes are converted into 300-dimensional vectors, the matching degree with the portraits is calculated, missing items are marked, and the resumes are repeatedly delivered within 30 days after shielding. Generating a comprehensive score of 0-100 points through an AI interview robot from three-dimensional evaluation of professional ability, comprehensive quality and culture matching degree; storing candidates according to six categories, generating 5-7 skill labels, and automatically pushing position updating; recording performance, training and promotion data, generating a growth curve, and predicting a promotion post; 15 indexes are counted, a visual report is generated, and improvement suggestions are triggered. The recruitment efficiency and precision are improved, talent management scientificity is enhanced, a data decision closed loop is formed, team stability is improved, and enterprise development is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise recruitment and management, in particular to an enterprise human resource intelligent recruitment and talent management system and method. BACKGROUND

[0002] In enterprise human resource management, recruitment and talent development are core links, but the traditional mode has many pain points. In the recruitment link, HR needs to manually analyze the position requirements, which is easy to cause fuzzy portrait due to subjective understanding deviation, and then receive a large number of mismatched resumes. The screening process relies on manual comparison, which not only consumes time (the average viewing time of a single resume is more than 5 minutes), but also may cause the omission of high-quality candidates due to experience differences. The interview evaluation lacks a standardized system, and the scoring dimensions of different interviewers are different, which leads to low comparability of the results, and the adaptation problem of the post after recruitment occurs frequently.

[0003] The talent management stage also faces challenges. Employee information is stored in tables or different systems, making it difficult to form a complete growth trajectory, and training recommendations are based on experience rather than data, resulting in resource waste. The resignation warning relies on the subjective judgment of managers, and often only after the employee proposes to resign is it dealt with passively, and the core talent loss cost is high. At the same time, recruitment and talent management data are fragmented, and the improvement direction of the recruitment link cannot be deduced through the resignation reason, forming a broken closed loop of "recruitment-management-optimization".

[0004] With the expansion of enterprise size and the intensification of talent competition, the efficiency bottleneck of the traditional mode becomes more and more obvious. According to statistics, the traditional recruitment cycle is more than 45 days on average, and the long waiting period of high-quality talents leads to an increase in the turnover rate; the employee training coverage rate is less than 60%, and the matching degree with the job demand is low; the annual turnover rate of core positions is more than 20%, which seriously affects business continuity. The root cause of these problems is the lack of intelligent tools to integrate recruitment whole-process and talent whole-life cycle data, making it difficult to achieve accurate matching, dynamic tracking and scientific decision-making. SUMMARY

[0005] The enterprise human resource intelligent recruitment and talent management system and method provided by the present application are used to solve the problems mentioned in the prior art.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: an enterprise human resource intelligent recruitment and talent management system, comprising:

[0007] A position requirement analysis module: a BERT pre-training model is used to perform word segmentation processing on the recruitment requirement text, extract education, skill and work experience indicators, and construct a structured position model through entity recognition technology; the system supports HR to customize the weight of each indicator, the weight adjustment step is 1%, the total sum of all indicator weights is 100%, and finally a position portrait containing 6 to 8 core requirements is generated;

[0008] Resume screening module: Based on the Word2Vec model, the resume text is converted into a 300-dimensional vector, and the matching degree with the position image is calculated by the cosine similarity algorithm. For resumes missing key skills, the system marks missing items and importance, with importance divided into high, medium and low levels; The system has a built-in duplicate delivery detection mechanism, which is checked by ID number and mobile phone number. The same resume is automatically shielded within 30 days;

[0009] Candidate evaluation module: A three-dimensional evaluation system is constructed, including five core skills in professional competence dimension, communication ability and stress resistance index in comprehensive quality dimension, and value fit index in cultural matching degree dimension. Video interviews are initiated by interview robots, with 8 preset professional questions, 4 of which are professional questions and 4 are behavioral questions. The candidate's answer time is 3 to 5 minutes, and the system generates a comprehensive score of 0 to 100 points through semantic understanding and sentiment analysis;

[0010] Talent pool management module: Store candidate information according to technical, management and functional positions, and automatically generate 5 to 7 skill tags for each resume; The system supports sorting by matching degree in descending order, with the latest 6 months of update time priority, and active degree within 30 days whether to view the position, etc. When the position requirements are updated, automatically push the position change information to the candidate who meets the historical matching degree;

[0011] Employee development tracking module: Records the quarterly performance appraisal results of employees after they join the company, uses KPI and OKR dual-dimensional scoring, records annual training records including course name, duration, assessment score and promotion path, records the position, time and reason of each promotion; Through linear regression algorithm to generate personal growth curve;

[0012] Data analysis module: Real-time statistics of recruitment cycle, statistics of resume conversion rate of each channel, including conversion rates from delivery volume to initial screening pass volume, initial screening pass volume to interview volume, interview volume to recruitment volume, and employee annual retention rate, that is, the proportion of employees who have been employed for more than 1 year, and generate line chart and pie chart visualization report.

[0013] Further, it also includes:

[0014] Interview assistance module: Integrates video interview function, supports real-time voice to text, identifies candidate's response basic expression through OpenCV library, calculates speech speed and pause frequency, generates emotion fluctuation curve, with time as horizontal axis and emotion value as vertical axis, and marks emotion abnormal period;

[0015] Salary matching module: access third-party salary database to obtain the same post salary quantile value, including P25, P50, P75 quantile value, combined with the candidate's expected salary and enterprise salary bandwidth to calculate the recommended salary range, automatically generate 3 salary schemes, including basic, standard and competitiveness, the scheme includes basic salary and performance ratio.

[0016] Training recommendation module: identify gaps through skill gap analysis, including the difference between job required skill score and employee current skill score, more than 15 points is considered to be improved, combined with career development plan to recommend matching courses from course library, support classification according to urgent promotion and long-term development; Track learning progress, video watching completion rate and post-test score of 60 points or more are considered qualified;

[0017] Early warning module of turnover risk: collect employee attendance data in the past 3 months and collect performance evaluation ranking in the past 2 consecutive quarters, the last 20% is abnormal, and internal job browsing records, calculate the turnover risk value through the logistic regression model, the risk value ranges from 0 to 100 points, 60 to 79 points for medium risk, push to department manager, 80 points or more for high risk, and push to HR director at the same time, and attach risk factor analysis.

[0018] Further, the comprehensive score formula of the candidate evaluation module is: S = 0.4A + 0.3B + 0.3C, wherein S is the comprehensive score; A is the professional ability score, which is obtained by weighting 5 core skills; B is the comprehensive quality score, which includes communication and stress resistance indicators; C is the cultural matching degree score, which includes the value of the matching degree index.

[0019] Further, the recommended salary calculation formula of the salary matching module is: P = aP1 + (1-a)P2, wherein P is the recommended salary; P1 is the candidate's expected salary, P2 is the market P50 salary of the same post, and a is the weight, which is dynamically adjusted according to the scarcity of the candidate, and a is 0.5 for scarce talents.

[0020] Further, the risk value calculation formula of the turnover risk early warning module is: R = 0.2K + 0.4J + 0.4L, wherein R is the turnover risk value, K is the attendance abnormality score, J is the performance change score, and L is the job browsing score.

[0021] Further, the new employee training effect evaluation formula is: T = 0.3C1 + 0.3E + 0.4P3, wherein T is the training effect evaluation score, C1 is the course completion degree, which is calculated according to the ratio of actual course duration to planned duration; E is the test score after training; P3 is the skill improvement degree, which is the ratio of the difference between the skill score after training and the skill score before training and the difference between the post requirement.

[0022] A method of applying the enterprise human resource intelligent recruitment and talent management system, comprising:

[0023] Demand analysis step: HR uploads the recruitment demand text in Word or PDF format, the system uses BERT model to split words and extract indicators, HR adjusts the weight of each indicator in the visual interface, confirms the position image after generating the position image, locks 6 to 8 core requirements, sets the matching threshold, and supports saving as a post template;

[0024] Resume screening step: the system synchronizes or HR uploads resumes from recruitment websites, calculates the matching degree according to Word2Vec vector, and displays the top 50 matching resumes after sorting, and marks the resumes with missing skills in red;

[0025] Evaluation interview step: HR selects 8 interview questions from the question bank, 4 of which are professional questions and 4 are behavioral questions, and the system sends an interview invitation with a two-dimensional code to the candidate; the candidate scans the code to enter the video interview, the system transcribes the text in real time and analyzes the expression and speech speed, and generates a comprehensive score within 10 minutes after the interview;

[0026] Talent storage step: qualified candidate information is automatically entered into the talent database, and the system generates 5 to 7 skill tags; set an automatic update reminder every 3 months, send it through SMS and email, and the candidate updates the resume, the system re-scores after updating, and the HR checks the resume modification record;

[0027] Employee tracking step: after the employee is employed, the system synchronizes the attendance data every month, and records the performance evaluation results every quarter, KPI accounts for 60%, OKR accounts for 40%, annual training courses, including online learning time of more than 20 hours per year and reward and punishment; update the growth curve every half year, and push the promotion evaluation application when the skill standard rate is more than 90%;

[0028] Data review step: generate the last month's recruitment report before the 5th of each month, including the conversion rate of resumes from each channel, analyze the employee retention rate every quarter, and when the department's annual retention rate is less than 70%, automatically associate the department's salary level and leadership style data to generate improvement suggestions.

[0029] Further, the professional ability score formula of the evaluation interview step is: A = Σ (a x w), wherein A is the professional ability score, a is the i-th skill score, which is scored by the evaluation system according to the answer content; w is the i-th skill weight, Σw = 1, consistent with the skill weight in the position image.

[0030] Further, a new recruitment channel effectiveness score formula is added: Q = 0.4R1 + 0.3C2 + 0.3S1, wherein Q is the recruitment channel effectiveness score, R1 is the resume quality score, calculated according to the average resume matching degree; C2 is the conversion rate, which is the ratio of the number of recruited people to the number of delivered people; S1 is the cost benefit score, calculated according to the ratio of the unit recruitment cost to the average cost.

[0031] Compared with the prior art, the present application has the following advantages:

[0032] In terms of recruitment efficiency and accuracy, the system automatically analyzes the position requirements through natural language processing, generates standardized portraits, and avoids manual interpretation bias. Intelligent screening based on semantic similarity algorithm quickly matches resumes, significantly reducing the processing time of invalid resumes. At the same time, the structured evaluation system of AI interview ensures objective and consistent scoring, significantly improves the quality of person-job matching, and reduces the risk of post-recruitment mismatch.

[0033] The scientificity and forwardness of talent management are enhanced, the system integrates employee performance, training, promotion and other data to form a complete growth curve, accurately identifies skill gaps and recommends matching training to avoid resource waste. The early warning mechanism of turnover risk monitors in real time through multi-dimensional data, discovers potential loss risk in advance, saves time for retention measures, and reduces the loss of core talent loss.

[0034] Data-driven decision-making loop effectively improves management level, data analysis module connects recruitment and talent management data, and provides basis for strategy adjustment through index correlation analysis, such as linking turnover reasons with recruitment channel quality. At the same time, the modules form a whole chain management from demand analysis to employee development, making human resource work from passive execution to active planning, adapting to the talent management needs of enterprise scale development, especially in reducing management cost and improving team stability, providing talent guarantee for sustainable development of enterprises. BRIEF DESCRIPTION OF DRAWINGS

[0035] Fig. 1 The schematic diagram of the enterprise human resource intelligent recruitment and talent management system proposed by the present application;

[0036] Fig. 2 The schematic diagram of the enterprise human resource intelligent recruitment and talent management method proposed by the present application;

[0037] Fig. 3 The schematic diagram of the enterprise human resource intelligent recruitment and talent management method proposed by the present application for comparing the resume screening effect under different matching thresholds;

[0038] Fig. 4 The schematic diagram of the enterprise human resource intelligent recruitment and talent management method proposed by the present application for the relationship between employee skill compliance rate and training effect. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0040] In the description of the present application, it should be understood that the orientations or positional relationships indicated by the terms “center”, “longitudinal”, “lateral”, “length”, “width”, “thickness”, “upper”, “lower”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inner”, “outer”, “clockwise”, “counterclockwise” and the like are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements indicated must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0041] In addition, the terms “first” and “second” are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with “first” and “second” can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of “multiple” is two or more, unless otherwise explicitly and specifically limited. In addition, the terms “mounting”, “connection” and “connection” should be broadly understood, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication between two elements. For a person of ordinary skill in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances, and the present application will be further described in detail below with reference to the drawings.

[0042] Reference Figs. 1 to 4 An enterprise human resource intelligent recruitment and talent management system, comprising:

[0043] A position demand analysis module: adopts a BERT pre-training model to perform word segmentation processing on recruitment demand text, the word segmentation accuracy is above 98%, 12 types of key indicators such as education background, skills and work experience can be extracted, a structured position model is constructed through entity recognition technology, and the F1 value of the entity recognition technology is above 95%. The system supports HR to define the weight of each indicator, the weight adjustment step is 1%, the total sum of all indicator weights is 100%, and finally a position portrait containing 6 to 8 core requirements is generated, and the portrait matching threshold can be set between 70% and 90%.

[0044] Resume intelligent screening module: based on the Word2Vec model, the resume text is converted into a 300-dimensional vector, and the matching degree with the position image is calculated by the cosine similarity algorithm. The calculation time of each resume is controlled within 0.5 seconds. The resumes with a matching degree of 80% or more are automatically screened out. For resumes lacking key skills, the system will mark the missing items and their importance, which is divided into high, medium and low levels. The system has a built-in duplicate delivery detection mechanism. Through the double verification of ID number and mobile phone number, the same resume within 30 days will be automatically shielded, and the historical delivery record will be prompted.

[0045] Candidate evaluation module: a three-dimensional evaluation system is constructed. The professional ability dimension includes 5 core skills, and the score of each skill ranges from 0 to 20 points. The comprehensive quality dimension includes 4 indicators such as communication ability and stress resistance, and the score of each indicator ranges from 0 to 25 points. The cultural matching degree dimension includes 3 indicators such as value fit, and the score of each indicator ranges from 0 to 33 points. Video interviews are initiated by AI interview robots. There are 8 pre-set job-related questions, 4 of which are professional questions and 4 of which are behavioral questions. The candidate's answer time for each question is controlled within 3 to 5 minutes. The system generates a comprehensive score of 0 to 100 points through semantic understanding and sentiment analysis, and the accuracy rate of semantic understanding is above 90%.

[0046] Talent pool management module: candidate information is stored in 6 categories such as technical positions, management positions, and functional positions. Each resume automatically generates 5 to 7 skill tags such as data analysis and project management. The system supports sorting by matching degree in descending order, priority given to updates within the last 6 months, and active degree within 30 days. When the job requirements are updated, the system automatically pushes job change information to candidates with a historical matching degree of 75% or more, with no more than 1 push per day.

[0047] Employee development tracking module: records the quarterly performance appraisal results of employees after they join the company, uses KPI and OKR dual-dimensional scoring, and records the score of each item ranging from 0 to 100 points. It also records annual training records, including course name, duration, assessment results, and promotion path, and records the position, time, and reason of each promotion. Through linear regression algorithm, the personal growth curve is generated, and the R value of the curve is above 0.85. Combined with the data of job promotion, 2 to 3 target positions that can be promoted within 2 to 3 years and the skills needed to be supplemented are predicted. 2

[0048] ​Data analysis module: Real-time statistics of recruitment cycle, i.e. the average length of time from job posting to recruitment, accurate to the day, while statistics of resume conversion rate of each channel, including the conversion rate from delivery volume to initial screening pass volume, initial screening pass volume to interview volume, interview volume to recruitment volume, and employee annual retention rate, i.e. the proportion of employees who have been employed for more than 1 year, etc. 15 indicators, generate line chart, pie chart and other visual reports, the update frequency of the report reaches once a day or more. When the recruitment rate is less than 30% or the retention rate is less than 80%, automatically trigger cause analysis and push improvement suggestions.

[0049] In the present application, it also includes:

[0050] AI interview assistance module: integrated high-definition video interview function, video resolution is 1080P, frame rate is 30fps, supports real-time voice to text, voice to text accuracy rate reaches more than 95%, delay control is within 1 second. Recognize 6 basic expressions of the candidate through OpenCV library, including smile, frown, surprise, etc., the accuracy rate of expression recognition reaches more than 85%, calculate the speech rate, the normal speech rate range is 150 to 200 words per minute, at the same time calculate the number of pauses, the pause time reaches 5 seconds and above is considered as effective pause, generate emotion fluctuation curve, the horizontal axis of the curve is time, the vertical axis is emotion value, the emotion value range is -100 to 100, mark the emotion abnormal period, such as the emotion value drops to 30 and above when answering the salary question.

[0051] Salary matching module: access to third-party salary database, database covers 34 provinces and cities nationwide, 200 industries and above, get the same position salary quantile value, including P25, P50, P75 quantile value, combined with the candidate's expected salary and the enterprise's salary bandwidth, such as 8k-15k for a middle-level engineer, calculate the recommended salary range, the salary precision is to the hundred yuan, the deviation rate is controlled within 5%. Automatically generate 3 salary schemes, including basic, standard and competitiveness, the scheme includes basic salary, performance ratio, performance ratio range is 30% to 40%, and year-end bonus coefficient, year-end bonus coefficient range is 1 to 3 times of monthly salary.

[0052] Training recommendation module: identify the gap through skill gap analysis, i.e. the difference between the job requirement skill score and the current employee skill score, the difference reaches 15 points and above is considered to be improved, combined with the career development plan, such as promotion to manager within 3 years, recommend matching courses from the course library, the course library contains 500 online courses and above, each course lasts for 2 to 8 hours, the course matching degree reaches 80% and above, supports classification according to urgent improvement and long-term development. Track learning progress, video watching completion rate and post-test score reaches 60 points and above is considered qualified, generate learning report every month, push reminders to employees who have not completed the learning plan for 2 consecutive months.

[0053] The off risk early warning module: the attendance data of the employees in the last 3 months are collected, the monthly absence of 3 days and above or the tardiness of 5 times and above are regarded as abnormal, the performance evaluation ranking of 2 consecutive quarters is collected, the last 20% is regarded as abnormal, and the internal post browsing record is collected, the monthly browsing of 10 times and above is regarded as abnormal, and 5 kinds of data are calculated by a logistic regression model to calculate the off risk value, the risk value ranges from 0 to 100 points, 60 to 79 points are medium risk, are pushed to the department manager, 80 points and above are high risk, are synchronously pushed to the HR general manager, and the risk factor analysis is attached, such as browsing the post of a competitive company 3 times in the last 1 month.

[0054] In the application, the comprehensive score formula of the candidate evaluation module is: S=0.4A+0.3B+0.3C, wherein S is the comprehensive score, the score range is 0-100; A is the professional ability score, the score range is 0-100, which is obtained by weighting 5 core skills; B is the comprehensive quality score, the score range is 0-100, which includes 4 indexes such as communication and stress resistance; C is the cultural matching degree score, the score range is 0-100, which includes 3 indexes such as value concept matching degree.

[0055] In the application, the salary matching module suggestion salary calculation formula is: P=alpha P1+(1-alpha) P2, wherein P is the suggestion salary, the unit is yuan / month; P1 is the candidate expected salary, the unit is yuan / month; P2 is the market same post P50 salary, the unit is yuan / month; alpha is the weight, the value range is 0.3-0.5, which is dynamically adjusted according to the candidate scarcity, and alpha of scarce talents is 0.5.

[0056] In the application, the risk value calculation formula of the off risk early warning module is: R=0.2K+0.4J+0.4L, wherein R is the off risk value, the score range is 0-100; K is the attendance abnormality score, the score range is 0-100, 0 points for no abnormality, and 100 points for serious abnormality; J is the performance change score, the score range is 0-100, 0 points for performance rising, and 100 points for continuous decline; L is the post browsing score, the score range is 0-100, 0 points for no browsing, and 100 points for high-frequency browsing.

[0057] In the application, the new employee training effect evaluation formula is: T=0.3C1+0.3E+0.4P3, wherein T is the training effect evaluation score, the score range is 0-100; C1 is the course completion degree, the score range is 0-100, which is calculated according to the proportion of the actual course duration and the planned duration; E is the examination score, the score range is 0-100, which is the test score after training; P3 is the skill improvement degree, the score range is 0-100, which is the proportion of the skill score difference before and after training and the post requirement difference.

[0058] The application also discloses a method of enterprise human resource intelligent recruitment and talent management system.

[0059] Demand analysis step: HR uploads the recruitment demand text in Word or PDF format, the number of words reaches 500 and above, the system uses BERT model to divide words and extract 12 types of key indicators, HR adjusts the weight of each indicator in the visual interface, such as setting Python skill to 20%, and generates position image after confirmation, locks 6 to 8 core requirements, sets the matching threshold, the default threshold is 80%, supports saving as a post template, and the template can be reused.

[0060] Resume screening step: the system synchronizes or HR uploads resumes from recruitment websites, supports PDF, Word, JPG formats, and the OCR recognition accuracy reaches 98% and above, calculates the matching degree according to the Word2Vec vector, and displays the top 50 high-matching resumes after sorting, and marks the resumes lacking key skills in red, such as lacking 'big data analysis' experience, automatically shields the resumes repeatedly delivered within 30 days and displays the historical score.

[0061] Evaluation interview step: HR selects 8 interview questions from the question bank, which contains 1000 and above questions, 4 of which are professional questions and 4 of which are behavior questions, the system sends an interview invitation with a two-dimensional code to the candidate, the invitation is valid for 48 hours. The candidate scans the code to enter the video interview, each question is limited to 3 to 5 minutes, the system transcribes the text in real time and analyzes the expression and speech speed, and generates a comprehensive score within 10 minutes after the interview, which is 70 and above. The pass, at the same time, generates a 300-word evaluation summary, which includes strengths and weaknesses.

[0062] Talent storage step: the information of qualified candidates is automatically recorded in the talent database, the system generates 5 to 7 skill tags such as machine learning and team management, and stores them according to the post classification. Set up an automatic update reminder every 3 months, send it through SMS and email, candidates can update their resumes, the system will re-score after updating, and HR can view the resume modification record, such as adding project experience.

[0063] Employee tracking step: after the employee is employed, the system synchronizes the attendance data every month, and records the performance evaluation results every quarter, KPI accounts for 60%, and OKR accounts for 40%. The annual record training courses include online learning time of 20 hours and above per year and reward and punishment. Update the growth curve every half year, when the skill standard rate reaches 90% and above, push the promotion evaluation application, and attach the average promotion time comparison of the same post.

[0064] Data review step: Generate the last month's recruitment report before the 5th of each month, including the conversion rate of each channel, such as 40% for headhunter channel and 25% for social recruitment website, and the average recruitment period, such as 35 days for technical positions and 20 days for functional positions. Analyze employee retention rate every quarter. When the annual retention rate of a certain department is less than 70%, automatically associate the department's salary level, leadership style, etc. Data to generate 3 improvement suggestions, such as increasing the performance bonus ratio of the department.

[0065] In the present application, the professional ability score formula of the evaluation interview step is: A = Σ (a x w), wherein A is the professional ability score, the score range is 0-100; a is the i th skill score, the score range is 0-100, which is scored by AI according to the answer content; w is the i th skill weight, Σ w = 1, which is consistent with the skill weight in the position image.

[0066] In the present application, the recruitment channel effectiveness score formula is added: Q = 0.4R1 + 0.3C2 + 0.3S1, wherein Q is the recruitment channel effectiveness score, the score range is 0-100; R1 is the resume quality score, the score range is 0-100, calculated according to the average value of resume matching degree; C2 is the conversion rate, the score range is 0-100, which is the ratio of the number of recruits to the number of deliveries; S1 is the cost benefit score, the score range is 0-100, calculated according to the ratio of unit recruitment cost to average cost.

[0067] Specific embodiments of the enterprise human resource intelligent recruitment and talent management system and method.

[0068] Example 1: Application of Internet Enterprise Technical Position Intelligent Recruitment and Management System

[0069] This example is aimed at the Java development engineer position (3 years of experience, need to master SpringBoot, microservice architecture) of an Internet enterprise, and deploys an intelligent recruitment and talent management system. The position demand analysis module receives the 800-word recruitment script uploaded by HR, extracts 7 core indicators such as "bachelor's degree or above" "3-5 years of development experience" "SpringBoot framework proficiency" after word segmentation by BERT model, and HR sets the weight of "microservice experience" to 25% and "database optimization" to 20% in the system interface to generate a position image with a matching threshold of 80%.

[0070] The resume intelligent screening module synchronizes the resumes of the recruitment website every day, converts the received 200 resumes into Word2Vec vectors (300 dimensions), and calculates the matching degree by cosine similarity. Among them, 150 resumes are automatically filtered because the matching degree is less than 80%, 50 resumes enter the next link, and the system marks 12 resumes missing "distributed system development" experience as "high importance missing". The repeated delivery detection is verified by ID number, and 8 resumes that have been repeatedly delivered within 30 days are shielded.

[0071] The candidate evaluation module initiates AI video interviews with 50 candidates, with 8 pre-set questions (4 technical questions such as "microservice service discovery mechanism" and 4 behavioral questions such as "team conflict handling case study"). The system transcribes the answers in real time (96% accuracy), identifies candidates' expressions such as frowning and pausing (speech speed of 180 words / minute is normal, >220 words / minute is marked as "too fast"), and calculates a comprehensive score using S = 0.4A + 0.3B + 0.3C (A is technical score, B is communication and other comprehensive qualities, and C is cultural fit). 30 candidates with a score of ≥70 are added to the talent pool.

[0072] The talent pool management module adds tags such as "microservice development" and "high-concurrency processing" to 30 people, sorting them in descending order of matching degree. When the job title is updated to "Senior Java Engineer," information is automatically pushed to 15 people with a historical matching degree of ≥75%. The employee development tracking module records quarterly KPIs (code quality, project progress) and OKRs (technical breakthrough goals) for the 10 newly hired employees, generating growth curves (R²) through linear regression. 2 =0.88), predicts that he can be promoted to "Technical Team Leader" within 2 years, but needs to supplement his "team management" skills.

[0073] The data analysis module shows that the recruitment cycle for this position is 32 days, the resume conversion rate through headhunters is 42%, and the annual retention rate is 88%, generating a visual report. When two of the recruits left within three months, the system's correlation analysis showed that their salaries were below the market P50, and a suggestion to "adjust the salary bandwidth for this position" was pushed.

[0074]

[0075] The table shows that traditional manual screening of 200 resumes takes 8 hours, while the system only takes 0.5 hours. This is significantly reduced by automatically filtering out low-match resumes and duplicate submissions. Interview evaluation, through a structured process and automated scoring, reduced the evaluation time for 50 people from 25 hours to 5 hours, with standardized scoring criteria. The recruitment cycle was shortened by 47%, thanks to the intelligent processing at each stage, avoiding human delays and demonstrating the system's significant effectiveness in improving recruitment efficiency.

[0076] Example 2: Application of Intelligent Recruitment and Management System for Management Positions in Manufacturing Enterprises

[0077] This example is designed for the production manager position in a manufacturing company (5+ years of experience, familiar with lean production). The system functions are as follows: The job requirement analysis module extracts 6 core indicators such as "lean production certification" and "management experience with a team of 100 or more people". HR sets the weight of "cost control ability" at 30%.

[0078] The AI interview assistance module identifies the candidate's expression in the video interview through OpenCV (emotional value + 50 when smiling, -30 when frowning), and marks 3 people whose emotional value drops by 40 when answering "overtime acceptance" as "need to communicate in detail". The salary matching module calculates the recommended salary as P = 0.4P1 + 0.6P2 (P1 is the candidate's expectation, P2 is the market P50), and generates a range of 23k-26k for a candidate who expects 25k / month (market P50 is 24k).

[0079] The training recommendation module analyzes the skill gap of the 5 production managers hired, finds that the "digital production management" score is 20 points different from the job requirements (≥15 points need to be improved), recommends a 40-hour online course (completion rate ≥80% qualified), and generates a learning report every month. The turnover risk warning module collects data such as attendance (monthly absence <3 days), performance (continuous two quarters rising), and job browsing (monthly <5 times), and calculates the risk value as R = 0.2K + 0.4J + 0.4L (all <60 points, no warning).

[0080] The employee tracking step calculates the skill compliance rate as D = (M / N) x 100% (M is the number of compliant skills, N is the total number of skills), and when the compliance rate of 5 people is all ≥90%, it pushes the promotion evaluation application. The data review step shows that the internal recommendation channel has a hiring rate of 60% (higher than the 35% of social recruitment), and generates a "increase internal recommendation incentives" suggestion.

[0081]

[0082] The table reflects the advantages of the system in talent management: traditional training relies on experience recommendation, and the matching degree with job requirements is low; the system accurately recommends through skill gap analysis, and improves the effectiveness of training. The turnover warning is extended from less than 3 days to more than 30 days, giving time for retention measures. The promotion accuracy rate is increased from 60% to 85% through data-based evaluation, avoiding subjective judgment bias. These improvements are due to the system's integrated analysis of employee lifecycle data, making management decisions more scientific.

[0083] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any skilled person in the art can make equivalent replacement or change according to the technical solution and inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. An enterprise human resource intelligent recruitment and talent management system, characterized in that, Comprise: Position demand analysis module: adopt BERT pre-training model to process recruitment demand text, extract education, skills, work experience indicators, and build structured position model through entity recognition technology; the system supports HR to define the weight of each indicator, the weight adjustment step is 1%, and the total weight of all indicators is 100%, finally generates a position portrait containing 6 to 8 core requirements; Resume screening module: based on Word2Vec model, convert resume text into 300-dimensional vector, calculate the matching degree with position portrait through cosine similarity algorithm, for resumes missing key skills, the system marks missing items and importance, importance is divided into high, medium and low three levels; the system has a repeated delivery detection mechanism, which is checked by ID number and mobile phone number, and the same resume within 30 days is automatically shielded; Candidate evaluation module: build a three-dimensional evaluation system, professional competence dimension contains 5 core skills, comprehensive quality dimension contains communication ability and stress resistance index, cultural matching degree dimension contains value fit index, video interview is initiated through interview robot, 8 preset professional questions, 4 of which are professional questions, and 4 are behavioral questions, candidate's answer time is 3 to 5 minutes, the system generates a comprehensive score of 0 to 100 through semantic understanding and sentiment analysis; Talent pool management module: store candidate information according to technical, management and functional positions, and automatically generate 5 to 7 skill tags for each resume; the system supports sorting by matching degree in descending order, priority given to resumes updated within the last 6 months, and active degree within 30 days; when the position requirements are updated, the system automatically pushes the position change information to the candidates who meet the historical matching degree; Employee development tracking module: records the quarterly performance evaluation results of employees after they join the company, uses KPI and OKR dual-dimensional scoring, records annual training records, including course name, duration, evaluation score and promotion path, records the position, time and reason of each promotion; generate personal growth curve through linear regression algorithm; Data analysis module: real-time statistics of recruitment cycle, statistics of resume conversion rate of each channel, including conversion rate from delivery volume to initial screening pass volume, initial screening pass volume to interview volume, interview volume to recruitment volume, and employee annual retention rate, i.e. the proportion of employees who have been employed for more than 1 year, generate line chart and pie chart visualization report.

2. The system according to claim 1, wherein, Also include: Interview assistance module: integrates video interview function, supports real-time voice to text, identifies candidate's response basic expression through OpenCV library, calculates speech speed and pause frequency, generates emotion fluctuation curve, curve horizontal axis is time, vertical axis is emotion value, marks emotion abnormal period; Salary matching module: access third-party salary database to get same position salary quantile value, including P25, P50, P75 quantile value, calculate the recommended salary range combined with candidate's expected salary and enterprise salary bandwidth, automatically generate 3 salary schemes, including basic salary, standard salary and competitiveness salary, the scheme contains basic salary and performance proportion.

3. The system for intelligent recruitment and talent management of enterprise human resources according to claim 1, characterized in that, Also include: Training recommendation module: identify gaps through skill gap analysis, including the difference between job requirement skill score and employee current skill score, more than 15 points is considered to be improved, combined with career development plan to recommend matching courses from course library, support classification according to urgent promotion and long-term development; Track learning progress, video watching completion rate and post-test score of 60 points is considered to be qualified; The risk of leaving the warning module: collect the employee's attendance data in the past 3 months, and collect the performance evaluation ranking in the past 2 consecutive quarters, the last 20% is abnormal, and the internal post browsing record is collected, the risk value of leaving is calculated by logistic regression model, the risk value range is 0 to 100 points, 60 to 79 points is medium risk, push to department manager, 80 points and above is high risk, synchronous push to HR director, and attach risk factor analysis.

4. The system for intelligent recruitment and talent management of enterprise human resources according to claim 1, characterized in that, The comprehensive score formula of the candidate evaluation module is: S = 0.4A + 0.3B + 0.3C, where S is the comprehensive score; A is the professional ability score, which is obtained by weighting 5 core skills; B is the comprehensive quality score, which includes communication and stress resistance indicators; C is the cultural matching degree score, which includes the value of the matching degree index.

5. The system for intelligent recruitment and talent management of enterprise human resources according to claim 2, characterized in that, The salary matching module suggestion salary calculation formula is: P = αP1 + (1-α)P2, where P is the suggested salary; P1 is the candidate's expected salary, P2 is the market P50 salary of the same position, and α is the weight, which is dynamically adjusted according to the scarcity of the candidate, and α of scarce talents is 0.

5.

6. The system for intelligent recruitment and talent management of enterprise human resources according to claim 3, characterized in that, The risk value calculation formula of the risk of leaving the warning module is: R = 0.2K + 0.4J + 0.4L, where R is the risk value of leaving, K is the attendance abnormal score, J is the performance change score, and L is the post browsing score.

7. The system for intelligent recruitment and talent management of enterprise human resources according to claim 1, characterized in that, New employee training effect evaluation formula: T = 0.3C1 + 0.3E + 0.4P3, where T is the training effect evaluation score, C1 is the course completion degree, which is calculated according to the ratio of actual course completion time to planned time; E is the test score after training; P3 is the skill improvement degree, which is the ratio of the difference between the skill score after training and the skill score before training and the difference between the job requirement.

8. A method for using the enterprise human resource intelligent recruitment and talent management system according to any one of claims 1-7, characterized in that, Including: Demand analysis step: HR uploads Word or PDF format recruitment demand text, system uses BERT model to split and extract indicators, HR adjusts the weight of each indicator on the visual interface, confirms the position image after generating, locks 6 to 8 core requirements, sets the matching threshold, and supports saving as a post template; Resume screening step: the system synchronizes or HR uploads resumes from recruitment websites, calculates the matching degree according to Word2Vec vector, and displays the top 50 matching resumes after sorting, and marks the resumes with missing skills in red; Evaluation interview step: HR selects 8 interview questions from the question bank, 4 of which are professional questions and 4 are behavioral questions, the system sends an interview invitation with a two-dimensional code to the candidate; the candidate scans the code to enter the video interview, the system transcribes the text in real time and analyzes the expression and speaking speed, and generates a comprehensive score within 10 minutes after the interview; Talent storage step: qualified candidate information is automatically entered into the talent pool, and the system generates 5-7 skill tags; set automatic update reminders every 3 months, send through SMS and email, candidates update their resumes, after updating, the system re-scores, HR reviews resume modification records; Employee tracking steps: after the employee is hired, the system synchronizes attendance data every month, quarterly performance review results are entered, KPI accounts for 60%, OKR accounts for 40%, annual training courses are recorded, including online learning for more than 20 hours per year and reward and punishment; update the growth curve every half year, when the skill standard rate is more than 90%, push the promotion evaluation application; Data review steps: generate last month's recruitment report by the 5th of each month, including resume conversion rate from each channel, analyze employee retention rate every quarter, when the department's annual retention rate is less than 70%, automatically associate the department's salary level and leadership style data, and generate improvement suggestions.

9. The enterprise human resource intelligent recruitment and talent management method according to claim 8, characterized in that, The professional ability score formula of the evaluation interview step is: A = Σ(a x w), where A is the professional ability score, a is the ith skill score, which is scored by the evaluation system according to the answer content; w is the weight of the ith skill, Σw = 1, consistent with the skill weight in the position image.

10. The method of claim 8, wherein, New recruitment channel effectiveness score formula: Q = 0.4R1 + 0.3C2 + 0.3S1, where Q is the recruitment channel effectiveness score, R1 is the resume quality score, calculated according to the average value of resume matching degree; C2 is the conversion rate, which is the ratio of the number of hires to the number of applicants; S1 is the cost benefit score, calculated according to the ratio of unit hiring cost to average cost.

Images