Talent data value evaluation method and system based on big data

Abstract

The invention relates to the technical field of talent value evaluation, in particular to a talent data value evaluation method and system based on big data, and the method comprises the steps: data collection: collecting user basic information and value-added information; benchmark person hypothesis: automatically complementing preset parameters for missing fields; initial value modeling: introducing a confidence coefficient to generate an initial value interval based on the reference person hypothesis parameters and in combination with the located economic region; value-added dynamic quantification: configuring a weight coefficient according to a value-added item category, and calculating a dynamic value-added value through a time attenuation factor and an age factor; outputting: fusing the initial value interval and the dynamic value-added value to generate a comprehensive value evaluation result; real-time feedback: generating an information perfection prompt and evaluation result visualization chart; the system comprises a data acquisition module, a reference person hypothesis module, an initial value modeling module, a value-added dynamic quantification module, an output module and a real-time feedback module. The method has the effect of improving the accuracy of talent value evaluation.

Description

Technical Field

[0001] The present application relates to the technical field of talent value assessment, and in particular to a talent data value assessment method and system based on big data. Background Art

[0002] Currently, in the field of human resource management and finance, accurately assessing the value of talent is of great significance for the recruitment, selection, training and salary design of enterprises or the repayment capacity assessment of banks. Traditional talent assessment methods often focus on the assessment of talent's current ability, but ignore the assessment of its future potential.

[0003] In the existing technology, talent value assessment usually relies on static parameters and single-dimensional data, such as estimation based only on academic qualifications or work experience, resulting in low accuracy in talent value assessment. Summary of the invention

[0004] In order to improve the accuracy of talent value assessment, the present application provides a talent data value assessment method and system based on big data.

[0005] In the first aspect, the present application provides a talent data value assessment method based on big data, which adopts the following technical solution: A talent data value assessment method based on big data includes the following steps: Data collection: obtain the region information mapped to the mobile phone number through the user registration interface, and parse the age and gender through the real-name authentication interface; Benchmark person assumptions: automatically complete missing fields with preset parameters, including education level, work experience, cognitive ability, and economic region mapping; Initial value modeling: Based on the benchmark person assumption parameters, combined with the economic region where it is located, and introducing the confidence coefficient to generate the initial value range; Dynamic quantification of added value: Build a value-added model, configure weight coefficients according to the value-added project categories, and calculate the dynamic value-added value through time decay factors and age factors; Output: The initial value range is integrated with the dynamic value-added value to generate the final comprehensive value assessment result; Real-time feedback: Generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

[0006] By adopting the above technical solutions, the existing technologies are mostly based on linear evaluation based on static parameters (such as education and salary), and do not integrate regional economic parameters, dynamic confidence coefficients and nonlinear quantification of value-added models; the uncertainty of missing information is reflected through dynamic confidence coefficients; the value-added model supports multi-dimensional dynamic influences such as skills and honors; real-time feedback improves user participation and result credibility; and through multi-faceted considerations, the value assessment of talents is more accurate, and can reflect the true value of talents and evaluate the value of talents in multiple dimensions.

[0007] Optionally, after the output step, a scene adaptation step is further provided; Scenario adaptation: Based on the final comprehensive value evaluation results, output financial adaptation value and / or job adaptation value and / or social contribution value.

[0008] By adopting the above technical solutions, the traditional method only outputs a single comprehensive value, without differentiated adaptation for different application scenarios (such as financial credit and job matching); it meets the needs of multiple scenarios (such as financial institutions need credit assessment and enterprises need job matching); the social contribution value is calculated independently to reflect the social influence of talents.

[0009] Optionally, the confidence coefficient is generated by the following logic: In the initial stage, the perturbation coefficient is allocated according to the number of dimensions with missing information, and the value range of the perturbation coefficient is 0.9-1.1; Each time a piece of information is improved, the range of the disturbance coefficient is reduced by 10%-15%; Adjust the final range of the confidence coefficient based on the average industry return volatility in the user's region; The value-added model supports dynamic configuration of value-added items, including: Skill certificate category: weight coefficient range is 1-1000; Social position category: weight coefficients are set hierarchically according to the position level; Honorary awards: The weight coefficient is positively correlated with the influence of the award; Property category: The weight coefficient is adjusted according to the liquidity of assets.

[0010] By adopting the above technical solution, the confidence intervals in the existing technology are mostly fixed values ​​or adjusted based on only a single factor, and the weights of value-added information are usually statically assigned; the dynamic adjustment of the confidence coefficient improves the robustness of the initial assessment; the dynamic weight configuration enhances the adaptability of the model to different career scenarios (such as technical positions focusing on certificates, and management positions focusing on positions), and through the dynamic adjustment of the confidence coefficient, the value assessment of talents is more specific, and it is possible to weight real beneficial projects, which is convenient for truly evaluating the value of talents.

[0011] Optionally, the formula for the value-added dynamic quantization step is: ; in, The time the user being assessed spends in the labor market from the year of assessment to retirement. For the year, ; For the The total time of working in a year, that is, Exp = current year - year of working; For the Year, the time when the value-added item was acquired; For the In years, the user’s age; It is the curve key number, which is used to control the curvature of the dynamic curve. The default value is 0.001. is the importance coefficient, which is used to express the influence of the value-added item on the value. The value range is 1-1000. The larger the value, the higher the weight. is the switch control coefficient, which takes a value of 0 or 1. If it is 0, the value-added item will be turned off and will no longer take effect. If it is 1, the value-added item will be turned on. is the multiplier refinement coefficient, which is used to refine the multiplier of the impact of value-added projects on income; is the theoretical annual income for that year, calculated by the Mincer equation.

[0012] Optionally, a job matching correction coefficient is introduced into the formula of the step value-added dynamic quantification step , and update the formula of the value-added dynamic quantization step as follows: ; The job matching correction factor The determination includes: Capture the target job's skill requirement keywords through crawlers; Match the overlap between skill certificates and keywords in the user's value-added items; If the overlap is ≥70%, it is considered a match and the correction factor is triggered. , when it does not match, .

[0013] By adopting the above technical solution, the traditional model uses linear superposition for value-added information, without considering the impact of time decay, age factors and job matching; the time decay factor reflects the long-term value decline of value-added projects; the job matching coefficient (z=1.2 or 0.8) improves the correlation between the evaluation results and job requirements, making the evaluation of talent value more accurate.

[0014] Optionally, the logic for generating the financial adaptation value includes: Calculate disposable income based on the user's theoretical annual income and regional average income parameters; Modify disposable income through adjustment coefficients for age, education, and marital status; Combine credit risk parameters with external credit data to generate the final range of financial adaptation value.

[0015] By adopting the above technical solutions, the existing financial assessment model ignores the dynamic impact of the blacklist, and the job assessment lacks the ability to capture industry benchmark parameters in real time; the financial adaptation value more truly reflects the user's credit risk; the job adaptation value improves accuracy based on real-time industry data; and the visualization tool enhances the user experience.

[0016] Optionally, the adjustment logic of the credit risk parameter is: If the user is blacklisted, the financial adaptation value is reduced to 20% of the original value; If there is no hit, the adaptation weight will be dynamically adjusted based on the external credit score.

[0017] Optionally, the logic for generating the position adaptation value includes: Capture industry benchmark earnings parameters of target positions through Internet big data; Match job requirements based on user's years of work experience, educational background, and skill certificates; Output the job adaptation value and its percentile level in the industry.

[0018] By adopting the above technical solutions, job adaptation value captures industry benchmark parameters and matches skills through big data, and job adaptation value improves accuracy based on real-time industry data.

[0019] Optionally, the real-time feedback step includes: Information completion progress bar: marks missing fields and indicates the priority of completion; 3D radar chart: displays the correlation between comprehensive value, financial adaptation value, and job adaptation value; Simulation Configuration Tool: Allows users to adjust the weights of value-added items and preview changes in assessment results in real time.

[0020] By adopting the above technical solution, it is easy to clearly understand the various aspects of the value information of talents through visual display. At the same time, through simulation configuration, it is convenient for users to supplement the information and strengthen the shortcomings.

[0021] Optionally, the method further comprises a dynamic optimization step: Monitor users’ newly added implicit value-added information; The implicit information and value correlation are analyzed through machine learning models, and the weight coefficient is automatically adjusted.

[0022] By adopting the above technical solution and introducing a dynamic optimization step, the real-time and accuracy of the evaluation are improved; a dynamic optimization mechanism is added so that the evaluation can be continuously updated and optimized; the logic is reasonable and can reflect the dynamic changes of user information.

[0023] Optionally, in the output step, a multi-dimensional weighted fusion is used to fuse the initial value interval and the dynamic added value, and the multi-dimensional weighted fusion includes: introducing an industry adaptation coefficient to nonlinearly amplify the incremental value, wherein the adaptation coefficient is generated by calculating the matching degree between the target industry skill requirements and the user skill data.

[0024] By adopting the above technical solutions, the industry adaptation coefficient improves the industry specificity of the evaluation results, and combines the matching degree between the target industry needs and the user skill data, making it more authentic and in line with the actual industry needs, and the talent value feedback is more accurate and targeted.

[0025] Optionally, the mapping logic includes: parsing the economic zone associated with the user IP address or registration information, matching the regional talent density coefficient and the industry concentration parameter; For cross-regional mobile users, a migration cost factor is introduced to correct the initial capacity value.

[0026] By adopting the above technical solution, the migration cost factor corrects the deviation of cross-regional talent value.

[0027] Optionally, the step of dynamically quantifying the added value further includes a dynamic weight allocation model for adjusting the configured weight coefficient, and the dynamic weight allocation model includes: Preset weight coefficients are matched based on skill categories, where technical certification skills are weighted higher than general skills; The weight of social contribution data is dynamically decayed according to its timeliness, and the decay rate is inversely proportional to the time span of the contribution event.

[0028] By adopting the above technical solution, dynamic weight allocation reflects the changing trend of skill requirements.

[0029] Optionally, the weight coefficient is dynamically adjusted in the following manner: Monitor the frequency of changes in skill requirements in target industries and increase the weight of emerging skill categories exponentially; Nonlinear gains are applied to weights based on the public impact rating of social contribution data.

[0030] In the second aspect, the present application provides a talent data value assessment system based on big data, which adopts the following technical solutions: A talent data value assessment system based on big data, including the following modules: Data collection module: used to collect user information data; Benchmark person assumption module: The input end is connected to the output end of the data acquisition module, and is used to automatically complete the preset parameters for the missing fields in the user information data; Initial value modeling module: the input end is connected to the output end of the benchmark person hypothesis module, and is used to generate an initial value interval based on the benchmark person hypothesis parameters and the confidence coefficient; Value-added dynamic quantification module: The input end is connected to the output end of the initial value modeling module, which is used to build a value-added model and calculate the dynamic value-added value; Output module: The input end is connected to the output end of the dynamic quantification module, which is used to merge the initial value range with the dynamic value-added value to generate the final comprehensive value assessment result; Real-time feedback: The input end is connected to the output end of the output module to generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

[0031] By adopting the above technical solutions, the system collects basic user data through multiple sources such as API interfaces, resume parsing tools, and third-party platforms (such as LinkedIn); covering 20+ dimensions of information such as education background, work experience, skill certificates, project achievements, etc.; in the benchmark person hypothesis stage, the industry benchmark data set (such as job competency model) is used to fill in missing values; dynamically match the characteristics of the top 30% talents in the same position as the filling parameters; initial value modeling, build a multidimensional evaluation model based on the random forest algorithm, and the confidence coefficient dynamically adjusts the value interval bandwidth according to the data completeness (0.6-1.0) ; Dynamic quantification of value-added, building an LSTM time series model to analyze career development trajectories, introducing three-dimensional dynamic factors: industry prosperity index (30%), skill scarcity (40%), and project complexity (30%); value fusion output, using Monte Carlo simulation to perform probabilistic fusion of value intervals and dynamic values, and generating a comprehensive value radar chart with probability distribution (basic capabilities, development potential, and market adaptability); real-time feedback link, generating interactive data dashboards through D3.js, and the intelligent suggestion engine recommending 3-5 value-added paths (e.g., obtaining PMP certification is expected to increase valuation by 8-15%).

[0032] The system significantly improves the objectivity and predictability of talent value assessment by building a complete closed loop of data collection → intelligent completion → dynamic modeling → value-added prediction → decision support. After verification by more than 300 enterprises, the talent retention matching degree has increased by 25%, and the rationality of the salary system has increased by 38%, fully reflecting the transformation value of human resource management driven by big data.

[0033] In summary, the present application includes at least one of the following beneficial technical effects: 1. The uncertainty of missing information is reflected through dynamic confidence coefficients; the value-added model supports dynamic influences in multiple dimensions such as skills and honors; real-time feedback improves user participation and result credibility; and through multiple considerations, the value assessment of talents is more accurate, and can reflect the true value of talents and assess the value of talents in multiple dimensions; 2. The introduction of dynamic optimization steps improves the real-time and accuracy of the evaluation; the addition of a dynamic optimization mechanism enables the evaluation to be continuously updated and optimized; the logic is reasonable and can reflect the dynamic changes of user information; 3. The industry adaptation coefficient improves the industry specificity of the assessment results, and combines the matching degree between the target industry needs and the user skill data, making it more authentic and in line with the actual industry needs, and the talent value feedback is more accurate and targeted. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a flow chart of a talent data value assessment method in an embodiment of the present application; Figure 2 It is a flow chart of obtaining the initial value in the initial value modeling step in the embodiment of the present application. DETAILED DESCRIPTION

[0035] The following combination Figure 1 to Figure 2 This application is described in further detail.

[0036] This embodiment discloses a talent data value assessment method based on big data.

[0037] Reference Figure 1 and Figure 2 ,The talent data value assessment method based on big data includes the following steps: Data collection: Obtain basic information and value-added information, where basic information includes regional information, age, gender and other information. The regional information mapped to the mobile phone number is obtained through the user registration interface, and the age and gender are parsed through the real-name authentication interface. Other information is obtained through resumes or user input; the user's basic information and value-added information are obtained through the information collection system; the added information includes at least one of skill certificates, social positions, honorary awards, and intellectual property rights; the information collection system automatically completes the unfilled fields using the benchmark person assumption and marks them as items to be corrected.

[0038] Specifically, the basic information includes: mobile phone number, ID number, gender, education background, graduation school, year of employment, year of joining the current company, ethnicity, marital information, position (job category, industry), household registration type, place of work, parents' education level, whether English is used at work, English proficiency, whether computers are used at work, type of work unit, monthly income, and health status.

[0039] Benchmark person assumptions: automatically complete missing fields with preset parameters, including education level, work experience, cognitive ability, and economic region mapping; Specifically, the so-called benchmark person assumption is to make an average or slightly lower-than-average assumption for the various variable parameters required for the calculation of the JF lifetime income model; the specific assumptions are: college degree, ordinary college; female; 38 years old, started working at the age of 25, and joined the current company this year; tertiary industry, other personnel position; urban household registration, non-party member; both parents have not received high school education or above; do not know English and do not need to use English at work; need to use computers at work; cognitive ability is at the average level, that is: 1.934808850288391; the work area is the area corresponding to the location of the mobile phone number; the salary level is the average salary level corresponding to the location of the mobile phone number; Obtain the user's actual information, complete the actual information through the benchmark person hypothesis, and mark the completed items.

[0040] Initial value modeling: Based on the benchmark person assumption parameters, combined with the economic region where it is located, and introducing the confidence coefficient to generate the initial value range; Specifically, a confidence coefficient is dynamically generated based on the degree of information loss to adjust the uncertainty range of the initial value assessment result; the generation logic of the confidence coefficient includes: Assign an initial confidence interval based on the number of dimensions where user information is missing; Based on the progress of subsequent user information improvement, dynamically narrow the confidence interval and update the confidence coefficient; The confidence coefficient is generated in the following way: In the initial stage, a random perturbation coefficient is assigned according to the number of missing dimensions of user information, and the value of the random perturbation coefficient is 0.9-1.1; Each time a piece of information is improved, the range of the disturbance coefficient is reduced by 10%-15%; Adjust the final range of the confidence coefficient based on the average industry return volatility in the user's region; The value-added model supports dynamic configuration of value-added items, including: Skill certificate category: weight coefficient range is 1-1000; Social position category: weight coefficients are set hierarchically according to the position level; Honorary awards: The weight coefficient is positively correlated with the influence of the award; Property category: The weight coefficient is adjusted according to the liquidity of assets.

[0041] Specifically, the initial evaluation step adopts a future income discount model, which specifically includes: Calculate the user's baseline theoretical annual income through the Mincer equation; Estimate future promotion probability and income growth rate; The discount result is interval-corrected in combination with the confidence coefficient to generate the initial value assessment result; the input parameters of the Mincer equation include: At least five of the following: education level, years of work experience, geographical distribution, gender, cognitive ability, and family environment; The geographical distribution is mapped to large economic regions through user mobile phone numbers and associated with regional average revenue parameters.

[0042] The original version of the JF Lifetime Income Method is mainly used to estimate the macro stock of human capital in a region. The estimation accuracy of specific individuals, especially heterogeneous human capital, is insufficient. In addition, the original version of the JF Lifetime Income Method only evaluates the impact of education investment on human capital, which is not comprehensive enough. In order to solve this problem, in a research project in cooperation with Central Finance, Professor Li Haizheng made some improvements and expansions to the method, adding a comprehensive assessment of the dimensions of "cognitive ability", "work experience", "gender", "ethnicity", "political outlook", "work area", and "family environment" of talents.

[0043] The JF Lifetime Income Method is a method based on the discount of future income. It calculates the total income that the person being assessed can obtain during his / her stay in the labor market in the future. The assessment process is divided into five steps: Step 1: Obtain information about the user's education level, years of work experience, location, gender, ethnicity, cognitive level, and family environment, and then use the Mincer equation to estimate the "theoretical annual income" of the person being evaluated in the year of evaluation.

[0044] Step 2: Estimate the probability of promotion of the person being evaluated in the next five years by using the years of work and the coefficient given by the central finance department, and then take the average value to convert it into the probability of promotion for the current year; Step 3: Estimate how many times the income will increase after career advancement in different years, as well as the normal growth rate of wages in different years; Step 4: Based on the "theoretical annual income in the year of assessment" obtained from the Mincer equation, multiply it by the different promotion probabilities and promotion income growth rates in future years to calculate the theoretical annual income of the person being assessed in future years before retirement; Step 5: Add up the income of each year in the future to get the "lifetime income" of the person being evaluated, that is, the net worth. Then combine it with the average net worth of the benchmark person, calculate the ratio of the evaluator's net worth to the average net worth of the benchmark person, and convert it into the basic value, that is, the initial value.

[0045] Dynamic quantification of added value: Build a value-added model, configure weight coefficients according to the value-added project categories, and calculate the dynamic value-added value through time decay factors and age factors; The formula for the value-added dynamic quantification step is: ; in, The time the user being assessed spends in the labor market from the year of assessment to retirement. For the year, ; For the The total time of working in a year, that is, Exp = current year - year of working; For the Year, the time when the value-added item was acquired; For the In years, the user’s age; It is the curve key number, which is used to control the curvature of the dynamic curve. The default value is 0.001. is the importance coefficient, which is used to express the influence of the value-added item on the value. The value range is 1-1000. The larger the value, the higher the weight. is the switch control coefficient, which takes a value of 0 or 1. If it is 0, the value-added item will be turned off and will no longer take effect. If it is 1, the value-added item will be turned on. is the multiplier refinement coefficient, which is used to refine the multiplier of the impact of value-added projects on income; is the theoretical annual income for that year, calculated by the Mincer equation.

[0046] Output: The initial value range is integrated with the dynamic value-added value to generate the final comprehensive value assessment result; Specifically, the ratio of the initial value obtained to the value obtained by the Mincer equation of the average level of the benchmark person is calculated, and then the ratio of the added value to the value obtained by the Mincer equation of the average level of the benchmark person is calculated. The sum of the two ratios is calculated to obtain the comprehensive value.

[0047] Real-time feedback: Generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

[0048] The method also includes a dynamic correction step for monitoring the progress of user information improvement and updating basic information and value-added information in real time; Re-perform the initial assessment and value-added assessment steps based on the updated data; Output the revised comprehensive value assessment results.

[0049] Specifically, data acquisition and benchmark person assumptions: users complete registration by entering their mobile phone numbers through the registration interface, and the system automatically resolves the location of the mobile phone number. According to Table 1: Mapping table of regions and major economic regions, the user's place of work is mapped to the preset major economic region (such as the eastern coastal area, central China, western China, or northeastern China). For basic information fields that are not filled in (such as education level and work experience), the system uses the benchmark person assumption to automatically complete the default parameters: Education level: college degree from an ordinary college; Work experience: slightly lower than the industry average (the default age is 25 to start working, and 38 to join the current company); Geographical distribution: mapped to major economic regions based on the location of the mobile phone number; Cognitive ability: preset benchmark value 1.934808850288391; Missing fields are marked as items to be corrected (such as "Educational background-to be supplemented"), and the priority of improvement is prompted through a progress bar in the user interface.

[0050] Table 1 Mapping table of regions and economic regions Eastern and coastal areas Northeast Region Western Region Central Region Beijing Heilongjiang Shaanxi Inner Mongolia Tianjin Jilin Gansu Shanxi Shanghai Liaoning Ningxia Henan Hebei / Qinghai Anhui Shandong / Xinjiang Jiangxi Jiangsu / Tibet Hubei Zhejiang / Sichuan Hunan Fujian / Chongqing / Guangdong / Guizhou / Hainan / Yunnan / / / Guangxi / Initial assessment: Future income discount model: The initial assessment module calls the JF lifetime income method. The specific process is as follows: Mincer equation calculates the benchmark theoretical annual income: Input parameters include education level (X1_1~X1_7, see Table 2 below), years of work experience (X2_1), geographical distribution (X10_1~X10_4), gender (X3), cognitive ability (X15), etc.; Table 2 Variables and codes of the lifetime income method model

[0051] The regional parameters are mapped to large economic regions through mobile phone numbers and are associated with the regional income parameters in the following Table 3: Comparison of average wages in various regions.

[0052] Table 3 Comparison of average wages in various regions (2018-2019)

[0053] Example: A user maps to the eastern coastal region and calls the average revenue parameter of the region as the benchmark input.

[0054] Estimate future promotion probability and income growth rate: Generate a promotion probability curve based on the user's years of work (e.g. 10 years) and the industry's average promotion cycle (e.g. 5 years); Adjust the future revenue growth rate based on the industry fluctuation parameters of large economic regions (such as the revenue growth rate coefficient of the eastern region is 1.2-1.5 times).

[0055] Confidence coefficient correction: In the initial stage, a random perturbation coefficient (0.9-1.1) is assigned according to the number of missing information dimensions to generate an initial value assessment range (e.g. 3 million to 4 million); If the user only fills in 50% of the basic information, the confidence interval is ±15%; after completing it to 80%, the interval is reduced to ±5%.

[0056] Based on the complete parameters of the preset benchmark person, the benchmark level of the average value in each information dimension required for the JF lifetime income method assessment is substituted into the Mincer equation to calculate the benchmark person's worth. Then, combined with the estimated initial worth of the user, the proportion of the user's initial worth to the benchmark person's worth at the average benchmark level is calculated to obtain the user's initial value.

[0057] Value-added assessment: dynamic quantitative analysis: If the user adds value-added information (such as "Professional English Level 8 Certificate"), the value-added assessment module performs the following operations: Dynamically configure weight coefficients: match preset weights according to value-added item categories (e.g. skill certificate im=60, social position im=100); Call the value-added project configuration table (see Table 4) and determine the parameters (r=0.001, k=1).

[0058] Since the total number of value-added items available for evaluation in reality is uncertain and changes all the time, a dynamic configuration method will be used to control the value-added items. The specific configuration method is as follows: Table 4 Configuration of value-added items Value-added item ID Name of value-added project Value-added categories Curve coefficient r Importance coefficient im Refinement coefficient k Switch c 100001 Professional English Level 8 Skills Certificate 0.001 60 1 1 100002 Broker Certificate Skills Certificate 0.001 35 1 1 200001 Municipal CPPCC Member Social Position 0.001 100 1 1 300001 Municipal Model Worker Meritorious Service 0.001 50 1 1 400001 Nobel Prize in Physics Awards 0.001 320 1 1 400002 Fields Medal in Mathematics Awards 0.001 280 1 1 500001 Real Estate real estate 0.001 120 1 1 The values ​​in the table are all examples. The specific values ​​need to be verified and studied according to the actual situation before they can be finally determined. In the future, as the value-added formula is expanded, the configuration items in the configuration table will also increase. Substitute into the value-added model formula: ; Through the above formula, the original added value is calculated, and then the original added value is converted. By calculating the ratio of the original added value to the average net worth of the benchmark person, the final added value is obtained.

[0059] Comprehensive evaluation: Calculate the final talent value by calculating the sum of basic value and added value; In other implementations, since the basic value is calculated by the user's core long-term information using the JF lifetime income method, the result is relatively stable and has a high weight; the added value is calculated by calculating some of the user's added value items, and the added value based on the core value is converted, the result is more dynamic and has a lower weight, and nonlinear weighted integration can be used: The comprehensive assessment module nonlinearly integrates the initial value assessment results with the added value: Weight distribution: basic value weight 70%, added value weight 30%; Non-linear correction: If the value-added project is highly matched with the target position (such as IT certificate matching technical position), the multiplier coefficient is triggered Improve the added value weight and finally output the comprehensive value assessment results.

[0060] Dynamic correction and scene adaptation: Dynamic correction: When the user updates his education level to a postgraduate degree, the system re-executes the Mincer equation calculation and corrects the theoretical annual income; The confidence interval narrows from ±10% to ±5% as the information completeness increases; The update log records the changes in evaluation results before and after the correction.

[0061] In other embodiments, a job matching correction coefficient is introduced into the formula of the step value-added dynamic quantification step. , and update the formula of the value-added dynamic quantization step as follows: ; The job matching correction factor The determination includes: Capture the target job's skill requirement keywords through crawlers; Match the overlap between skill certificates and keywords in the user's value-added items; If the overlap is ≥70%, it is considered a match and the correction factor is triggered.

[0062] When it does not match, .

[0063] In other implementations, after the output step, a scene adaptation step is further provided; Scenario adaptation: Based on the final comprehensive value evaluation results, output financial adaptation value and / or job adaptation value and / or social contribution value.

[0064] Scenario adaptation output: Financial adaptation value: Based on the disposable income model, combined with the age coefficient (1.2 for 31-40 years old), marital status (1.2 for married) and external credit data (Lingxi score / 700), a credit adaptation interval is generated; Among them, based on the comprehensive income assessment of the user, combined with the user's reasonable income adjustment coefficient, minus the basic expenditure, the user's disposable income can be calculated; Income adjustment coefficient (see Table 5-1, Table 5-2, Table 5-3, Table 5-4): Table 5-1 Income Adjustment Coefficients of Age

[0065] Table 5-2 Income adjustment coefficient of education

[0066] Table 5-3 Income Adjustment Coefficients of Marital Status

[0067] Table 5-4 Income adjustment coefficients for real estate conditions

[0068] According to financial industry practice, 70% of disposable income can be considered as the user's debt repayment ability; According to the financial market, ordinary individuals can apply for a 10-year equal principal and interest credit loan product with an average annualized rate of 10%. The monthly repayment amount for a principal of 100,000 yuan is 2,124 yuan; User's debt repayment ability / 2124 yuan, multiplied by 100,000 yuan is the customer's financial value; Taking into account the specificity of high-level talents, their financial value is increased by the following corresponding values ​​(see Table 6): Table 6 Talent Rating Added Value Talent Rating Added value Category A 10 million Category B 5 million Category C 1 million Category D 500,000 Category E 200,000 Considering that there is a chance that the financial value is higher than the net worth, the minimum value between the financial value and the net worth*0.7 is taken; Finally, the financial value is further processed by introducing the external Lingxi score and blacklist. The blacklist hit takes 0.2 of the financial value, and the miss takes 1. Referring to the previous Lingxi score application, the Lingxi score divided by 700 is used as the adjustment coefficient. Financial value assessment model: Financial value: the user's reasonable credit debt tolerance User income = (job value / 12)*0.5+local average monthly salary*0.2+customer reported monthly income*0.3; User income adjustment coefficient = age coefficient + education coefficient + marriage coefficient + property coefficient; User average monthly basic expenditure = 2000; User monthly disposable income = income*coefficient-average basic expenditure; Monthly disposable income less than 0 is counted as 0; User monthly repayment ability = monthly disposable income*0.7; Combined with the financial loan market, ordinary individuals can apply for a five-year personal consumption loan with an average annualized interest rate of 10%, and the monthly repayment amount for every 100,000 yuan in equal principal and interest repayment is 2124 yuan; Financial value 1 = (((income*coefficient-average basic expenditure)*0.7) / 2124)*100000; Taking into account the specificity of high-level talents, their financial value is increased accordingly; Considering that there is a chance that the financial value is higher than the net worth, the minimum value between the financial value and the net worth*0.7 is taken; Finally, the financial value is introduced into the external Lingxi score and blacklist for further processing. A blacklist hit is taken as 0.2 of the financial value, and a miss is taken as 1.

[0069] Refer to the previous Lingxi score application and divide the Lingxi score by 700 as the adjustment factor.

[0070] Position adaptation value: The generation logic of the position adaptation value includes: Capture industry benchmark earnings parameters of target positions through Internet big data; Match job requirements based on user's years of work experience, educational background, and skill certificates; Output the job adaptation value and its percentile level in the industry.

[0071] Job value is the evaluation value of the user's "doing things" application scenario. We define job value as: Different users have different annual salary levels corresponding to different positions.

[0072] The job value will be captured through the Internet big data to establish different job salary databases for jobs in different regions, different work experience, and different industries. Finally, it will be mapped to different job values ​​based on user information.

[0073] Capture the industry benchmark parameters of the target position (such as "senior engineer"), match the user's skill certificates with years of work experience, and output the percentile level (such as Top 20%).

[0074] Visual feedback and optimization suggestions: Information completeness interface: the progress bar shows the current completeness (e.g. 80%), and highlights items to be revised (e.g. “Mother’s education level – to be supplemented”); Prompt users to complete high-priority information (such as education background, work experience).

[0075] Evaluation result display: The radar chart shows the correlation between comprehensive value, financial adaptation value, and job adaptation value; the heat map shows the contribution of value-added items to the evaluation results (such as certificates accounting for 40% and honors accounting for 30%).

[0076] Configuration suggestion tool: simulate the change in evaluation results after the user adds "real estate" (im=120) (such as a 15% increase in comprehensive value); recommend high-weighted value-added projects (such as "provincial awards im=200").

[0077] Specifically, the comprehensive value evaluation refers to the process in which, after becoming a platform user, users gradually improve their information in various dimensions as they use the platform products more deeply, thereby continuously revising their worth and obtaining more accurate comprehensive worth assessment results.

[0078] The improvement of information mainly includes two parts: Improvement of basic information: The basic information will be used to evaluate the user's basic net worth in the JF lifetime income model. When the user just registers, the basic information of each dimension uses the benchmark person assumption. As the user information is improved, the default values ​​of each dimension in the benchmark person assumption will be revised. While obtaining a more accurate user portrait, a more accurate comprehensive net worth assessment result will also be obtained.

[0079] Improvement of value-added information: Value-added information will be used in the value-added model to calculate the value-added of the user's worth. Users can add new value-added items at any time. After the value-added model calculates the item, it will be converted into value and added to the user's comprehensive value.

[0080] This embodiment ensures that the evaluation results gradually converge to the true value as the information is improved through confidence coefficients and dynamic correction mechanisms; regional economic parameters and value-added models (non-linear formulas) improve regional and scenario adaptability. The benchmark person assumptions cover multiple dimensions such as education, experience, and region to reduce information missing bias; financial and job adaptation values ​​meet the needs of multiple scenarios (job hunting, credit); the visual interface guides users to complete information and enhances participation; real-time feedback and simulation tools help users formulate value enhancement strategies.

[0081] In other implementations, there are differences in the value representation of men and women, and there are also differences in benefits. In order to better calculate the value of talents accurately, the calculation model of the updated value-added value is as follows: ; Among them, s is gender, the index of male is greater than that of female, the value of s is 0-1, and the value of the index is determined according to actual needs.

[0082] In other implementations, the talent data value assessment method based on big data further includes a dynamic optimization step: Monitor users’ newly added implicit value-added information; The implicit information and value correlation are analyzed through machine learning models, and the weight coefficient is automatically adjusted.

[0083] Specifically, implicit value-added information collection Data source: User behavior log: monitor user operation records within the platform (such as participation in training courses, project collaboration records); External data interface: access to the enterprise HR system or third-party platforms (such as LinkedIn) to obtain users' project results and patent application records; Natural Language Processing (NLP): Parse documents uploaded by users (such as project reports and award certificates) and extract keywords (such as "machine learning project leader" and "provincial science and technology progress award").

[0084] Information type: Project participation: quantified based on project duration, role importance (person in charge / member), and impact of results (number of patents and paper citations); Training records: graded by training duration and the authority of the certification body (e.g. Coursera certification vs. in-house training); Social contribution: duration of public welfare, frequency of media coverage, and level of social position (such as municipal / provincial CPPCC member).

[0085] Machine Learning Model Training and Weight Adjustment: Feature Engineering: Convert implicit information into feature vectors: Project participation → Feature value = Project duration × Role weight (Person in charge = 1.2, member = 0.8); Training record → Feature value = Certification agency level (1-5 stars) × Training duration; Social contribution → Feature value = Public welfare duration × Position level coefficient (Municipal level = 1, Provincial level = 1.5).

[0086] Model construction: Use the random forest model to input feature vectors and user historical evaluation results (such as comprehensive value change trends) to predict the relevance of implicit information to value; Output weight adjustment factor , dynamically update the weight parameters in the value-added model: ; Example: The user adds a "Provincial Science and Technology Progress Award" (original im=200). After the model analyzes its correlation with the comprehensive value, Δim=0.1→new weight im=220.

[0087] In other embodiments, in the output step, a multi-dimensional weighted fusion is used to fuse the initial value interval and the dynamic added value, and the multi-dimensional weighted fusion includes: introducing an industry adaptation coefficient to nonlinearly amplify the incremental value, wherein the adaptation coefficient is generated by calculating the matching degree between the target industry skill requirements and the user skill data.

[0088] Specifically, the industry adaptation coefficient is calculated as follows: Industry skill demand capture: Use a crawler to crawl the target job descriptions on recruitment websites (such as Zhaopin.com) and extract skill keywords (such as "Python" and "cloud computing"); Build an industry skill demand matrix and sort the skills by frequency of occurrence (e.g. the top 5 skills in the IT industry: Python, machine learning, big data, Java, AWS).

[0089] User skill matching assessment: Parse the skill certificates in the user's value-added items (such as "AWS Certification" and "Python Senior Engineer") to generate the user's skill vector; Calculate the cosine similarity between the user skill vector and the industry demand matrix:

[0090] Non-linear amplification logic: If the matching degree is ≥70%, the industry adaptation coefficient ; If 30%≤matching degree<70%, ; If the matching degree is less than 30%, .

[0091] Fusion formula: ; and Represent the weights corresponding to the initial value and the added value respectively.

[0092] In other embodiments, the mapping logic includes: parsing the economic zone associated with the user's IP address or registration information, matching the regional talent density coefficient with the industry concentration parameter; For cross-regional mobile users, a migration cost factor is introduced to correct the initial capacity value.

[0093] Specifically, economic zoning mapping: Data source: User IP address: the province to which the user's login IP belongs is analyzed through the GeoIP database; Registration information: permanent residence or place of work filled in by the user.

[0094] Economic zoning rules: See Table 1: Mapping table of regions and economic regions, which classifies users into four major economic regions (eastern coastal, central, western, and northeastern); Calling regional parameters: Talent density coefficient: East = 1.2 (high competition), West = 0.9 (low competition); Industrial concentration: Concentration of high-tech industries in the east = 1.5, concentration of manufacturing industries in the central region = 1.0.

[0095] Migration cost factor correction: Cross-regional user identification: If the user's current location and historical work location do not belong to the same economic zone, they are marked as cross-regional mobile users.

[0096] Migration cost calculation: , minimum 0.7; Example: Users migrate from the west (original region) to the east (new region), across 5 provinces → β = 1 − 5 / 10 = 0.5, but the lower limit is 0.7 → the final β = 0.7.

[0097] Initial value correction: .

[0098] In other embodiments, the step of dynamically quantifying the added value further includes a dynamic weight allocation model for adjusting the configured weight coefficient, and the dynamic weight allocation model includes: Preset weight coefficients are matched based on skill categories, where technical certification skills are weighted higher than general skills; The weight of social contribution data is dynamically decayed according to its timeliness, and the decay rate is inversely proportional to the time span of the contribution event.

[0099] The weight coefficient is dynamically adjusted in the following way: Monitor the frequency of changes in skill requirements in target industries and increase the weight of emerging skill categories exponentially; Nonlinear gains are applied to weights based on the public impact rating of social contribution data.

[0100] Specifically, the dynamic weight allocation model: skill category weight allocation: Technical certification skills (such as AWS certification, PMP certificate): basic weight im=100-500; General skills (such as Office operation, communication skills): basic weight im=10-50; Social contribution category (such as charity hours, CPPCC members): basic weight im=50-200, decaying according to timeliness.

[0101] Time-dependent decay algorithm: ; in, : Decay rate, default is 0.05 (time span is in years); : The time from when the social contribution event occurred to now (years).

[0102] Example: Provincial awards from 5 years ago (im=200), , → .

[0103] Emerging skills weighting: Monitoring changes in skill requirements: Crawl skill keyword frequencies on recruitment websites and calculate monthly growth rates; If the growth rate is ≥20%, it is judged as an emerging skill (such as "blockchain" and "metaverse").

[0104] Exponential improvement rules:

[0105] Example: Growth rate = 30%, original im = 100 → .

[0106] Social contribution impact gain: Public impact rating: score (1-5 stars) based on the level of media coverage (local / national level) and the scale of the public welfare project (number of participants ≥ 1,000); Nonlinear gain formula: ); Example: Rating 5 stars, original im=100→ .

[0107] The present application also provides a talent value assessment system, including the following modules for implementing the talent value assessment method of the previous embodiment: Includes the following modules: Data collection module: used to collect user information data; Benchmark person assumption module: The input end is connected to the output end of the data acquisition module, and is used to automatically complete the preset parameters for the missing fields in the user information data; Initial value modeling module: the input end is connected to the output end of the benchmark person hypothesis module, and is used to generate an initial value interval based on the benchmark person hypothesis parameters and the confidence coefficient; Value-added dynamic quantification module: The input end is connected to the output end of the initial value modeling module, which is used to build a value-added model and calculate the dynamic value-added value; Output module: The input end is connected to the output end of the dynamic quantification module, which is used to merge the initial value range with the dynamic value-added value to generate the final comprehensive value assessment result; Real-time feedback: The input end is connected to the output end of the output module to generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

[0108] Dynamic optimization module: The input end is connected to the output end of the output module, and the output end is connected to the input end of the value-added dynamic quantification module. It is used to monitor the user's newly added implicit value-added information, and analyze the correlation between implicit information and value through a machine learning model to automatically adjust the weight coefficient.

[0109] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A talent data value assessment method based on big data, characterized by: The following steps are involved: Data collection: Collect basic user information and value-added information, including region information, age, gender and other information. Obtain the region information mapped to the mobile phone number through the user registration interface, and parse the age and gender through the real-name authentication interface; Benchmark person assumptions: automatically complete missing fields with preset parameters, including education level, work experience, cognitive ability, and economic region mapping; Initial value modeling: Based on the benchmark person assumption parameters, combined with the economic region where it is located, and introducing the confidence coefficient to generate the initial value range; Dynamic quantification of added value: Build a value-added model, configure weight coefficients according to the value-added project categories, and calculate the dynamic value-added value through time decay factors and age factors; Output: The initial value range is integrated with the dynamic value-added value to generate the final comprehensive value assessment result; Real-time feedback: Generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

2. The talent data value assessment method based on big data according to claim 1 is characterized by: After the output step, a scene adaptation step is also provided; Scenario adaptation: Based on the final comprehensive value evaluation results, output financial adaptation value and / or job adaptation value and / or social contribution value.

3. The talent data value assessment method based on big data according to claim 2 is characterized by: The confidence coefficient is generated by the following logic: In the initial stage, the perturbation coefficient is allocated according to the number of dimensions with missing information, and the value range of the perturbation coefficient is 0.9-1.1; Each time a piece of information is improved, the range of the disturbance coefficient is reduced by 10%-15%; Adjust the final range of the confidence coefficient based on the average industry return volatility in the user's region; The value-added model supports dynamic configuration of value-added items, including: Skill certificate category: weight coefficient range is 1-1000; Social position category: weight coefficients are set hierarchically according to the position level; Honorary awards: The weight coefficient is positively correlated with the influence of the award; Property category: The weight coefficient is adjusted according to the liquidity of assets.

4. The talent data value assessment method based on big data according to claim 3 is characterized by: The formula for the value-added dynamic quantification step is: ； in, The time the user being assessed spends in the labor market from the year of assessment to retirement. For the year, ; For the The total time of working in a year, that is, Exp = current year - year of working; For the Year, the time when the value-added item was acquired; For the In years, the user’s age; It is the curve key number, which is used to control the curvature of the dynamic curve. The default value is 0.

001. is the importance coefficient, which is used to express the influence of the value-added item on the value. The value range is 1-1000. The larger the value, the higher the weight. is the switch control coefficient, which takes a value of 0 or 1. If it is 0, the value-added item will be turned off and will no longer take effect. If it is 1, the value-added item will be turned on. is the multiplier refinement coefficient, which is used to refine the multiplier of the impact of value-added projects on income; is the theoretical annual income for that year, calculated by the Mincer equation.

5. The talent data value assessment method based on big data according to claim 4 is characterized by: Introducing the job matching correction coefficient into the formula of the step value-added dynamic quantification step , and update the formula of the value-added dynamic quantization step as follows: ； The job matching correction factor The determination includes: Capture the target job's skill requirement keywords through crawlers; Match the skill certificates and keywords in the user's value-added items; if the overlap is ≥ 70%, it is considered a match and the correction factor is triggered. , when it does not match, .

6. The talent data value assessment method based on big data according to claim 2 is characterized by: The logic for generating the financial adaptation value includes: Calculate disposable income based on the user's theoretical annual income and regional average income parameters; Modify disposable income through adjustment coefficients for age, education, and marital status; Combine credit risk parameters with external credit data to generate the final range of financial adaptation value; The adjustment logic of the credit risk parameter is: If the user is blacklisted, the financial adaptation value is reduced to 20% of the original value; If there is no hit, the adaptation weight will be dynamically adjusted based on the external credit score.

7. The talent data value assessment method based on big data according to claim 5 is characterized by: The generation logic of the position adaptation value includes: Capture industry benchmark earnings parameters of target positions through Internet big data; Match job requirements based on user's years of work experience, educational background, and skill certificates; Output the job adaptation value and its percentile level in the industry.

8. The talent data value assessment method based on big data according to claim 7 is characterized by: The real-time feedback step includes: Information completion progress bar: marks missing fields and indicates the priority of completion; 3D radar chart: displays the correlation between comprehensive value, financial adaptation value, and job adaptation value; Simulation Configuration Tool: Allows users to adjust the weights of value-added items and preview changes in assessment results in real time.

9. The talent data value assessment method based on big data according to claim 1 is characterized by: The method also includes a dynamic optimization step: Monitor users’ newly added implicit value-added information; The implicit information and value correlation are analyzed through machine learning models, and the weight coefficient is automatically adjusted.

10. A system for implementing the talent data value assessment method based on big data as described in any one of claims 1 to 9, characterized in that: Includes the following modules: Data collection module: used to collect user information data; Benchmark person assumption module: The input end is connected to the output end of the data acquisition module, and is used to automatically complete the preset parameters for the missing fields in the user information data; Initial value modeling module: the input end is connected to the output end of the benchmark person hypothesis module, and is used to generate an initial value interval based on the benchmark person hypothesis parameters and the confidence coefficient; Value-added dynamic quantification module: The input end is connected to the output end of the initial value modeling module, which is used to build a value-added model and calculate the dynamic value-added value; Output module: The input end is connected to the output end of the dynamic quantification module, which is used to merge the initial value range with the dynamic value-added value to generate the final comprehensive value assessment result; Real-time feedback module: The input end is connected to the output end of the output module to generate information completeness prompts and visual charts of evaluation results, and provide value-added configuration optimization suggestions.

Images