Quantitative evaluation method for man-post matching based on competency model

Abstract

The invention provides a person and post matching quantitative evaluation method based on a competency model. According to the method, a unified competency evaluation model containing 15 secondary indexes in total in three dimensions of core ability, general ability and professional ability is constructed, post weights are calculated by adopting an analytic hierarchy process to generate standard scores, and comprehensive employee scores are generated in combination with 360-degree evaluation data and performance adjustment coefficients. And through the cosine similarity and the Euclidean distance, a man-post matching coefficient is calculated, and accurate diagnosis and personalized development recommendation are realized. The problems that in the prior art, post evaluation and employee evaluation data standards are heterogeneous, matching results are fuzzified, and system linkage is poor are solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of human resource management and computer data processing, and particularly relates to a method for quantitatively evaluating human-post matching based on a competency model. BACKGROUND

[0002] In modern enterprise management, achieving efficient matching of employees and posts is the key to improving organizational efficiency. Existing human resource management computer systems usually regard post evaluation and employee evaluation as two independent modules.

[0003] On the one hand, the post evaluation system aims to determine the relative value of the post to the organization, and its evaluation dimensions are mainly around traditional factors such as skills, responsibilities, effort and working environment. On the other hand, the employee evaluation system focuses on evaluating the performance and personal ability of the employee, and the methods include performance appraisal, 360-degree feedback, etc.

[0004] Such a separate technical architecture has the following technical defects:

[0005] First, the data standard heterogeneity and data island problem are prominent. Post evaluation and employee evaluation use two different data models and evaluation indicators, which leads to the output data being unable to be directly compared and correlated quantitatively. This makes the human-post matching judgment at the system level rely on the manual interpretation and subjective experience of managers, and the computer system cannot automatically and objectively generate accurate matching conclusions.

[0006] Second, the fuzziness and non-quantifiability of the matching results are serious. Due to the lack of a unified quantitative scale, the matching results are usually qualitative descriptions such as "suitable" or "unsuitable", which cannot accurately measure the degree of matching and identify the specific differences between the employee's ability and the post requirements, resulting in frequent misplacement of employees.

[0007] Third, the system linkage is poor and the resource waste is obvious. Fuzzy evaluation results are difficult to directly drive subsequent talent development systems (such as training systems), and a large amount of manual intervention is needed to develop development plans, resulting in low efficiency of the entire management process and poor targeting of training, causing waste of training resources.

[0008] Therefore, there is an urgent need in the field for a new technical solution to solve the technical problems in handling heterogeneous human resource data, achieving objective quantitative matching, and automating the linkage of talent development. SUMMARY

[0009] The technical problem to be solved is to provide a method for quantitatively evaluating human-post matching based on a competency model to solve the above problems in the prior art.

[0010] TECHNICAL SOLUTION

[0011] To achieve the above object, the application provides a kind of quantitative evaluation method of person-post matching based on competency model, comprising the following steps:

[0012] Step one: build the three-level structure containing core competence dimension, general ability dimension and professional ability dimension, set five secondary indicators under each primary dimension, form unified competency evaluation model;

[0013] Step two: determine the weight value of each dimension and generate post standard score vector by judgment matrix construction, characteristic vector calculation and consistency test;

[0014] Step three: collect multi-source data of employee self-evaluation, superior evaluation, peer evaluation, subordinate evaluation and customer evaluation, generate employee ability score through weighted fusion and outlier detection processing;

[0015] Step four: adjust employee ability score through performance adjustment coefficient, calculate cosine similarity and Euclidean distance of calibrated score vector and post standard score vector, and obtain person-post matching coefficient by fusion;

[0016] Step five: generate individual development plan containing training content, training method and training cycle through ability gap identification and training resource matching.

[0017] Specifically, the step one builds unified competency evaluation model, specifically including:

[0018] The core competence dimension includes five secondary indicators of value recognition, organizational commitment, professional ethics, team cooperation and communication expression;

[0019] The general ability dimension includes five secondary indicators of learning ability, innovation ability, execution ability, decision-making ability and stress resistance;

[0020] The professional ability dimension includes five secondary indicators of professional knowledge, professional skill, work experience, qualification certification and business insight.

[0021] Specifically, the step two quantitatively processes target post based on analytic hierarchy process, specifically including: constructing judgment matrix, comparing each dimension and each secondary indicator of the unified competency evaluation model two by two to determine relative importance;

[0022] Calculate the characteristic vector of judgment matrix to obtain the initial weight of each dimension and each secondary indicator;

[0023] Consistency test is carried out, when the consistency ratio is less than the preset threshold, the final weight value is determined;

[0024] According to the most weight value and the preset standard score range, a post standard score vector of the target post is generated. Specifically, the evaluation data collected and processed in the step three specifically includes:

[0025] The employee self-evaluation, superior evaluation, peer evaluation, subordinate evaluation and customer evaluation data are collected;

[0026] The weighted fusion is performed according to the employee self-evaluation weight 0.2, superior evaluation weight 0.3, peer evaluation weight 0.2, subordinate evaluation weight 0.15 and customer evaluation weight 0.15;

[0027] The outlier detection method based on quartiles is adopted to eliminate the evaluation values exceeding the first quartile minus 1.5 times the interquartile range or the third quartile plus 1.5 times the interquartile range.

[0028] Specifically, the employee ability score is calibrated by the performance adjustment coefficient in the step four, which specifically includes:

[0029] The work performance evaluation data of the employee in the past 12 months is obtained;

[0030] The performance adjustment coefficient is calculated based on the difference between the performance completion rate and the 100% reference value;

[0031] The calibrated employee comprehensive score is obtained by multiplying the employee ability score and the performance adjustment coefficient;

[0032] The performance adjustment coefficient calculation formula is:

[0033]

[0034] In the formula, a is the performance adjustment coefficient, and the value range is [0.6, 1.4]; P actual is the actual performance completion rate, expressed in percentage; P baseline is the reference performance completion rate, and the fixed value is 100%.

[0035] Specifically, the human-post matching coefficient is calculated in the step four, which specifically includes:

[0036] The cosine similarity of the 15-dimensional employee comprehensive score vector and the post standard score vector is calculated;

[0037] The Euclidean distance between the two vectors is calculated and normalized by dividing the maximum possible distance;

[0038] The similarity and the normalized distance are weighted and fused to obtain the human-post matching coefficient;

[0039] The human-post matching coefficient calculation formula is:

[0040] In the formula, M is a human-post matching coefficient, with a value range of [0, 1]; cosθ is a cosine similarity; d is an Euclidean distance; d max is a maximum Euclidean distance, equal to w1 is a similarity weight, with a value of 0.6; w2 is a distance weight, with a value of 0.4.

[0041] Further, it further includes a context-adaptive weight adjustment step:

[0042] The current project phase identifier, team size and task type code of the employee are obtained through the project management system interface; according to a preset context-weight mapping table, the weights of the three first-level dimensions are adjusted, wherein the core competence dimension weight in the project start phase is increased by 0.1, the professional competence dimension weight in the project execution phase is increased by 0.1, and the general competence dimension weight in the project closing phase is increased by 0.1;

[0043] The post standard score vector and the human-post matching coefficient are recalculated based on the adjusted weights.

[0044] Further, it further includes a data sparsity processing step:

[0045] The missing rate of each secondary index evaluation data is counted, and a missing rate exceeding 30% is defined as data sparsity;

[0046] Employees with the same post as the target employee and an evaluation completeness exceeding 80% are selected as similar groups, the Pearson correlation coefficient is used to calculate the similarity, and the evaluation mean of the top 5 employees is selected to complete the missing data;

[0047] Based on the historical evaluation data of the post in the expert knowledge base, the expected value of the posterior probability distribution is calculated as the missing evaluation value;

[0048] The data sparsity dimension weight is multiplied by a reliability coefficient, and the reliability coefficient is equal to 1 minus the missing rate.

[0049] Further, it further includes a new employee processing step:

[0050] Employees with less than 6 months of employment time are identified as new employees;

[0051] A linear regression model is constructed based on the education level code, work experience, number of skill certificates and entry evaluation score to predict the potential performance coefficient;

[0052] The potential prediction coefficient is completely used for the first 3 months of the new employee, the potential prediction weight is decreased by 20% per month from the 4th to the 6th month, the actual performance weight is increased by 20% per month, and the actual performance adjustment coefficient is completely used after 6 months.

[0053] Further, it further includes a post opportunity scanning step:

[0054] Store the 15-dimensional standard score vector of all positions in the organization;

[0055] Calculate the matching coefficient of the employee's comprehensive score vector and the standard score vector of all positions;

[0056] Screen positions with a matching coefficient higher than the current position's matching coefficient as development opportunities;

[0057] Based on the preset position level relationship matrix and career development path diagram, recommend the target position with the highest matching coefficient and meeting the promotion conditions.

[0058] Specifically, the step five through capability gap identification and training resource matching specifically includes:

[0059] Calculate the difference between the employee's 15 secondary indicator scores and the position standard score, and screen indicators with a difference greater than 1 point as key improvement points;

[0060] Based on the index code, search for corresponding courses in the training resource library, match 3-5 relevant training courses for each indicator; multiply the gap value by the weight of the indicator to get the priority score, arrange in descending order of score to determine the training order; output a structured training plan containing course name, training method identifier, training period, and target improvement score. Further, it also includes the project role model construction step:

[0061] Determine the type of project the employee is currently participating in through the project management system identifier;

[0062] Add three temporary indicators of project management, cross-departmental collaboration, and rapid learning to the basic 15 secondary indicators;

[0063] Regroup the 18 indicators into a composite evaluation system with a basic competence weight of 0.7 and a project competence weight of 0.3;

[0064] Adjust the weight distribution of the 18 indicators in real time according to the employee's project role changes.

[0065] Further, it also includes the strategic talent discovery step:

[0066] Define the innovation ability, decision-making ability, business insight, and leadership of the 15 secondary indicators as the four key strategic competencies;

[0067] Screen employees whose scores on the four key indicators all exceed 8 points as strategic talent candidates;

[0068] Calculate the matching degree of the employee's current ability score and the strategic position standard score as the development potential index;

[0069] Sort by potential index to establish a strategic talent reserve list and match each candidate with a corresponding ability improvement training plan.

[0070] In another aspect, a person-post matching quantitative evaluation system based on competency model is provided, comprising:

[0071] A model construction module comprising a three-level hierarchical structure constructor and 15 secondary indicator definers, which constructs a unified competency evaluation model comprising core ability dimension, general ability dimension and professional ability dimension;

[0072] A post quantitative module comprising a judgment matrix constructor, an eigenvector calculator and a weight determinator, which generates a 15-dimensional standard score vector of the target post through the analytic hierarchy process;

[0073] An employee portrait module comprising a multi-source data collector, a weighted fusion device and a performance calibrator, which generates a calibrated employee 15-dimensional comprehensive score vector;

[0074] A matching calculation module comprising a similarity calculator, a distance calculator and a fusion calculator, which outputs the matching coefficient of the employee and the post;

[0075] A diagnosis recommendation module comprising a gap identifier, a resource matcher and a scheme generator, which outputs a structured individualized development scheme.

[0076] Further, it further comprises:

[0077] A data preprocessing module comprising a data cleaner, a standardization processor and an outlier detector, which normalizes the multi-source evaluation data;

[0078] An adaptive adjustment module comprising a situation recognizer, a weight adjuster and a parameter optimizer, which adjusts the evaluation parameters according to the work situation and the employee type;

[0079] A visual display module comprising a matching result renderer, a radar chart generator and a report exporter, which displays the evaluation results in the form of charts and reports;

[0080] A data storage module comprising a model database, a standard score database, an evaluation database and a result database, which stores the data required for system operation in categories.

[0081] Beneficial effects

[0082] Compared with the prior art, the present application has the following beneficial effects:

[0083] (1) Because this invention constructs a unified competency evaluation model comprising 15 secondary indicators across three dimensions—core competencies, general competencies, and professional competencies—it enables job evaluation and employee evaluation to use the same data structure and evaluation dimensions, thus eliminating the fundamental reason for inconsistent data models from a technical perspective. When the system performs person-job matching, it can directly perform mathematical operations on the 15-dimensional vector without data format conversion or manual interpretation, thereby achieving automated and accurate matching by the computer system and solving the problem of heterogeneous data standards.

[0084] (2) Because this invention uses a performance adjustment coefficient to calibrate the 360-degree evaluation data, through E final =E initial The mathematical transformation of ×α can convert subjective evaluation data into objective ability assessment results. By combining the calculations of cosine similarity and Euclidean distance, through... The formula outputs specific matching coefficient values, which ensures the accuracy and quantifiability of the matching results from a mathematical perspective, avoiding the fuzzy judgments of "suitable" or "unsuitable" in traditional methods, and improving the quantitative accuracy of the matching results. (3) Since this invention establishes an automated discrimination mechanism from matching coefficients to capability weakness identification (automatically marking a weakness when the sub-item score is less than 90% of the standard score), the system can directly trigger the training resource matching algorithm based on mathematical calculation results, generating personalized development plans without manual intervention. This automated process design based on quantitative thresholds realizes the closed-loop linkage of evaluation, diagnosis, and development recommendation from a technical architecture perspective, improving the pertinence of talent development and system operating efficiency, and realizing automated linkage of the system. Attached Figure Description

[0085] Figure 1 This is a schematic diagram of the overall process of the method of the present invention;

[0086] Figure 2 A hierarchical structure diagram for a unified competency assessment model;

[0087] Figure 3 Here is a flowchart of the weight calculation process for the Analytic Hierarchy Process (AHP).

[0088] Figure 4 A diagram illustrating the calculation of the job matching coefficient. Detailed Implementation

[0089] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0090] Example 1: Basic Person-Job Matching Assessment Method

[0091] The embodiment takes the person-post matching evaluation of a large tobacco enterprise as an example to illustrate the specific implementation process of the method. Figure 1 As shown in the figure, the whole method includes the key steps of unified competence model construction, AHP weight calculation, employee capability portrait generation, person-post matching calculation and diagnosis recommendation, etc.

[0092] In practical application, the present application has significant technical advantages compared with traditional methods:

[0093] (1) The accuracy and objectivity of data processing are improved: By introducing performance adjustment coefficient to technically calibrate subjective evaluation data, and having robust processing ability for new employees, data sparsity and other scenarios, the person-post matching coefficient output by the method can more truly reflect the comprehensive ability of employees, and the evaluation accuracy can be improved by 15-20% compared with traditional methods.

[0094] (2) The running efficiency and adaptability of computer system are improved: The work of multi-source data integration, complex analysis and scheme formulation which needs to be done manually in the past is changed into automatic data processing and recommendation process. By supporting context adaptive weight and project model, the system can flexibly adapt to dynamic and agile organizational needs, and the data processing efficiency is improved by more than 60%.

[0095] (3) The precision and automation of talent development are realized: Through precise diagnosis and automatic recommendation, the general training mode is changed into personalized development path, the training resource allocation efficiency is improved by 40%, and the training effect satisfaction is improved by 25%.

[0096] (4) The intelligence and strategic value of the system are enhanced: An integrated technical solution is provided, which integrates evaluation, diagnosis, development and discovery, not only can automatically complete the closed-loop task from data input to development suggestion, but also can prospectively find internal "hidden talents" through global scanning, providing data support for enterprise strategic talent planning. (5) The system implementation and maintenance cost is reduced: Through unified competence model and modular architecture design, the complexity of system integration is reduced, the implementation cycle is shortened by 30%, and the later maintenance cost is reduced by 50%.

[0097] Step 1: Construct a unified competence evaluation model

[0098] First, according to the characteristics of the enterprise and the industry, a competence evaluation model containing 15 secondary indicators is constructed. As shown in the figure, Figure 2 The model adopts a three-level hierarchical structure:

[0099] The core competence dimensions include: responsibility (reflecting the employees' sense of responsibility and mission), initiative (reflecting the employees' initiative and enterprising spirit), teamwork (reflecting the employees' collaboration ability), overcoming difficulties (reflecting the employees' tenacity in the face of difficulties), and innovation (reflecting the employees' innovative thinking and practical ability).

[0100] The general competence dimensions include: overall coordination ability (reflecting the employees' organizational management ability), communication and execution ability (reflecting the employees' information transmission and execution ability), learning and improvement ability (reflecting the employees' continuous learning ability), information processing ability (reflecting the employees' data analysis and processing ability), and emotional management ability (reflecting the employees' emotional control and regulation ability).

[0101] The professional competence dimensions include: professional technical knowledge (reflecting the employees' professional theoretical level), professional skill level (reflecting the employees' practical operation ability), legal and regulatory knowledge (reflecting the employees' compliance awareness and legal literacy), professional digitalization ability (reflecting the employees' digital tool application ability), and writing and expression ability (reflecting the employees' writing and expression ability).

[0102] The model adopts a hierarchical database architecture for storage, supporting multi-version management and dynamic updating. The system establishes a competence index table to store the basic information of each index, an index definition table to support multi-language description, and a weight model version table to manage different version configurations.

[0103] Step 2: Analytic Hierarchy Process weight calculation

[0104] The organization establishes a post value evaluation committee composed of business department managers and external experts, with a total of 7 members. Committee members compare the 15 secondary indicators pairwise through a Web interface and use the 1-9 scale method to evaluate importance. Figure 3 As shown in the figure, the analytic hierarchy process weight calculation process includes steps such as judgment matrix construction, geometric mean aggregation, normalization processing, characteristic vector calculation, and consistency check.

[0105] Taking the post of the rank of the official as an example, the system collects 7 15x15 judgment matrices and uses the geometric mean method for aggregation:

[0106] Let the comparison result of the kth committee member for index i and index j be P k [i][j], then the comprehensive judgment matrix element calculation formula is: The comprehensive judgment matrix is normalized by column, and then the weight vector is obtained by summing by row. The system calculates the weights of each index as follows (example part):

[0107] Responsibility: 0.0842, Initiative: 0.0756, Teamwork: 0.0698, Overcoming difficulties: 0.0731, Innovation: 0.0623, Coordination ability: 0.0789, Communication and execution ability: 0.0734, Learning and improvement ability: 0.0687, Information processing ability: 0.0645, Emotion management ability: 0.0578, Professional technical knowledge: 0.0821, Professional skill level: 0.0793, Legal knowledge: 0.0712, Professional digital ability: 0.0634, Writing ability: 0.0657.

[0108] The consistency test is performed, the consistency index CI = 0.089 is calculated, the RI = 1.59 is obtained by table lookup, and the consistency ratio CR = 0.056 < 0.1 is calculated, which passes the consistency test.

[0109] The system stores the weight vector that passes the test into the index weight table and marks the weight model as active.

[0110] Step 3: Generate post standard score Based on the calculated weights, the system generates the post standard score for the post of the post. The evaluation committee scores the standard requirements of the post on each index (using the percentage system), and the system performs weighted summation calculation: S 正科 = 0.0842 x 85 + 0.0756 x 82 + 0.0698 x 78 + …

[0111] = 82.45

[0112] At the same time, the sub-item standard score of each secondary index is generated and stored in the post standard portrait table. The sub-item score is stored in JSONB format to provide flexible data structure support.

[0113] Step 4: Employee ability portrait generation Take employee Zhang as an example, the system collects 360-degree evaluation data through a networked form:

[0114] (1) Data collection: Through the multi-source data collector of the evaluation data collection unit, 360-degree evaluation data from self-evaluation, direct superior evaluation, peer evaluation (3 people), direct subordinate evaluation (2 people), and customer evaluation is collected. Each evaluation subject evaluates Zhang's performance on 15 secondary indicators using a four-level ability rating (junior 60 points, intermediate 70 points, senior 80 points, expert 90 points).

[0115] (2) Weighted fusion: The system uses a weighted fusion device to fuse multi-source data using pre-set weights. The weight distribution strictly follows the provisions of claim 1: the employee self-evaluation weight is 0.2, the superior evaluation weight is 0.3, the peer evaluation weight is 0.2, the subordinate evaluation weight is 0.15, and the customer evaluation weight is 0.15.

[0116] Take the "responsibility" index as an example:

[0117] E 责任担当 = 0.2 x 80 + 0.3 x 85 + 0.2 x 82 + 0.15 x 78 + 0.15 x 83

[0118] = 81.6

[0119] (3) Outlier detection processing: the system uses an outlier detector for outlier processing, and uses a four-quartile-based outlier detection method to eliminate evaluation values that are outside the range of the first quartile minus 1.5 times the interquartile range or the third quartile plus 1.5 times the interquartile range. By calculating the deviation of each evaluation value from the mean, when the deviation exceeds the set range, it is marked as an outlier and processed.

[0120] (4) Comprehensive evaluation score calculation: the evaluation results of the 15 indicators are weighted and averaged to obtain the preliminary comprehensive evaluation score of Zhang, which is 81.3 points.

[0121] Step 5: Performance adjustment coefficient calculation

[0122] The performance calibrator of the matching calculation unit comes into play, and the system extracts Zhang's work performance evaluation data for the past 12 months from the enterprise performance management system through the performance data acquisition module:

[0123] Personal performance completion rate: P actual = 108.5% Base performance completion rate: P baseline = 100% (fixed value) The performance adjustment coefficient is calculated by the adjustment coefficient calculation module based on the difference between the performance completion rate and the 100% base value:

[0124] Since the calculation result 0.834 is within the set value range [0.6, 1.4], the value is directly used as the adjustment coefficient. The calibrated employee comprehensive score is obtained by multiplying the employee ability score by the performance adjustment coefficient through the score calibration module: E final = 81.3 x 0.834 = 67.8

[0125] Step 6: Human-post matching coefficient calculation

[0126] The matching calculation unit comes into play, and the calibrated 15-dimensional employee comprehensive score vector of Zhang is matched and calculated with the 15-dimensional standard score vector of the positive section level post. As shown in Figure 4 The human-post matching coefficient calculation comprehensively considers the cosine similarity and Euclidean distance two dimensions:

[0127] Employee score vector:

[0128] Post standard score vector:

[0129] (1) Similarity calculation: As shown in the similarity calculator of the matching calculation unit, the cosine similarity between the 15-dimensional employee comprehensive score vector and the post standard score vector is calculated as the direction matching degree, and the cosine similarity calculation formula is: Figure 4

[0130] (2) Distance calculation: The distance calculator calculates the Euclidean distance between the two vectors and normalizes it by dividing by the maximum possible distance, and the Euclidean distance calculation formula is:

[0131] Let The distance matching degree calculation formula is:

[0132] (3) Fusion calculation: The fusion calculator fuses the similarity and normalized distance to obtain the person-post matching coefficient, and the weight specified in claim 1 is used: similarity weight w1=0.6, distance weight w2=0.4: M=0.6×0.943+0.4×0.389

[0133] =0.566+0.156

[0134] =0.722

[0135] The person-post matching coefficient is 0.722, and the value range is in [0, 1].

[0136] Step 7: Matching diagnosis and development recommendation

[0137] The diagnostic analysis unit generates a personalized development plan containing training content, training method and training cycle through ability gap identification and training resource matching.

[0138] According to the calculated matching coefficient 0.722, the system determines that Zhang is in the improvable interval (<0.9).

[0139] (1) Precise diagnosis: The gap identifier of the diagnostic analysis unit calculates the difference between the employee's 15 secondary index scores and the post standard score, and selects the index with a difference greater than 1 point as the improvement focus:

[0140] Ability short board identification (score < standard score x 0.9):

[0141] Active: 65.8 < 82.0 x 0.9 = 73.8 (short board)

[0142] Innovation: 58.3 < 75.0 x 0.9 = 67.5 (short board) ​

[0143] Coordination ability: 63.2 < 78.5 x 0.9 = 70.7 (short board)

[0144] Professional digital ability: 55.7 < 68.0 x 0.9 = 61.2 (short board)

[0145] (2) Training resource matching: The resource matcher retrieves corresponding courses in the training resource library based on the index code, and matches 3-5 relevant training courses for each index; The priority sorter multiplies the gap value by the weight of the index to get the priority score, and arranges the training order in descending order of the score.

[0146] (3) Generation of individualized development plan: The plan generator outputs a structured training plan containing course name, training method identifier, training period and target improvement score:

[0147] For the proactive short board: Recommend the course of "goal management and execution force improvement" (40 hours); For the short board of pioneering innovation: Recommend the "innovation thinking and method" workshop (24 hours); For the short board of overall coordination ability: Recommend the "project management practice" certification training (60 hours); For the short board of professional digital ability: Recommend the "digital tool application" online course (32 hours).

[0148] The system generates a complete individualized training plan, including training content, method, cycle and expected target, and pushes it to Zhang and his superior leaders.

[0149] Embodiment two: Situation adaptive weight adjustment for dynamic role

[0150] This embodiment illustrates how the system dynamically adjusts the weight configuration according to different work situations.

[0151] A company starts a major project and needs to select project managers from existing employees. The project is in the start-up stage, the team size is small (8 people), and the task type is product development.

[0152] Step 1: Situation recognition and weight adjustment

[0153] The system identifies the current work situation: project stage: start-up; team size: small team (<10 people); task type: R&D innovation.

[0154] According to the characteristics of the situation, the system retrieves the corresponding weight configuration from the situation weight set table. As shown in the table, situation adaptive weight adjustment is an important part of the overall process. Figure 1

[0155] ​Weight adjustment in innovative R&D context: Exploitation and exploration: weight increased from 0.0623 to 0.1200; Learning and development: weight increased from 0.0687 to 0.1100; Professional technical knowledge: weight increased from 0.0821 to 0.1150; Coordination ability: weight increased from 0.0789 to 0.1050; Other indicators' weights are adjusted accordingly to ensure the sum is 1.

[0156] Step 2: Recalculate matching based on adjusted weights

[0157] Recalculate the job matching degree of candidate employee Li using the adjusted weights:

[0158] Before adjustment: matching coefficient 0.756 (in the adaptation interval) After adjustment: matching coefficient 0.892 (still in the adaptation interval, but closer to the ideal state)

[0159] Through situation adaptive adjustment, the system can more accurately identify suitable candidates in specific work environment. Embodiment Three: Adaptive Fusion Processing for Sparse Data Scenario

[0160] This embodiment illustrates how the system handles the case of incomplete evaluation data.

[0161] Employee Wang is a newly transferred employee, only has superior evaluation and self-evaluation data, lack of peer and subordinate evaluation.

[0162] Step 1: The data integrity detection system detects that Wang's evaluation data missing rate is 50%, triggering the data sparse scenario processing flow. As Figure 1 shown, data sparse processing is a key technical feature of employee portrait generation link.

[0163] Step 2: Data completion processing

[0164] (1) Similar group collaborative filtering: the system identifies a group of 10 employees similar to Wang based on job level, professional background, work experience, etc., and extracts their historical evaluation data.

[0165] (2) Bayesian inference estimation: combining the prior probability in the expert knowledge base and the evaluation distribution of the similar group, estimate the evaluation value of the missing dimensions of Wang.

[0166] Take the "team cooperation" indicator as an example: - Prior probability: the average score of this indicator is 75 points, and the standard deviation is 8 points - Similar group average: 78 points - Bayesian estimation value: points

[0167] Step 3: Weight self-adaptation Due to the use of estimated data, the system reduces the weight of relevant indicators in the final score: - Indicators with actual data: weight remains unchanged - Indicators using estimated data: weight multiplied by the reliability coefficient 0.8 Through this processing mechanism, the system can still give a relatively reliable evaluation result in the case of incomplete data. Embodiment Four: Cold Start Processing for New Employees

[0168] This embodiment illustrates how the system handles the problem of missing performance data for new employees.

[0169] New employee Zhao has been working for 3 months and does not have enough performance data to support the calculation of performance adjustment coefficient.

[0170] Step 1: New employee identification

[0171] The system detects that Zhao's tenure is 3 months, which is less than the set cold start threshold of 6 months, triggering the cold start processing flow. Referring to Figure 1 Overall process, new employee cold start processing mainly affects the calculation of performance adjustment coefficient.

[0172] Step 2: Potential prediction model construction

[0173] The system constructs a potential prediction based on Zhao's basic information: - Education background: Master's degree from a 985 university (weight 0.25, score 85) - Work experience: 2 years of experience in related positions (weight 0.20, score 75) - Skill certification: CPA qualification certification (weight 0.15, score 90) - Entry evaluation: Comprehensive evaluation of 82 points (weight 0.40, score 82)

[0174] Potential index calculation:

[0175] Set the average potential index of new employees to 78, then the adjustment coefficient calculation formula is: Limit to a reasonable range: α = min(1.2, max(0.8, 1.058)) = 1.058, 1.2 is the preset upper limit of the "cold start" performance adjustment coefficient for new employees.

[0176] Step 3: Smooth transition mechanism

[0177] As Zhao's working time increases, the system gradually transitions from potential prediction to actual performance data:

[0178] When working for 6 months, start collecting actual performance data; during 6-12 months, the adjustment coefficient is calculated by weighted average of potential coefficient and actual coefficient; after 12 months, use the actual performance adjustment coefficient completely.

[0179] Embodiment Five: Global Position Opportunity Scanning and Strategic Talent Discovery

[0180] This embodiment illustrates how the system implements global post scanning and talent discovery functions.

[0181] Step 1: Batch matching calculation

[0182] The system establishes an asynchronous task queue and performs a global scanning task once a week. As shown in Figure 1 , the global post scanning is based on the matching coefficient calculation method shown in Figure 4 . Taking 200 employees and 50 posts as an example, the system needs to calculate 200 x 50 = 10000 matching combinations.

[0183] Using a distributed computing architecture, the computing task is decomposed into multiple sub-tasks and executed in parallel: - Task 1: Calculate the matching degree of employees 1-50 with all posts - Task 2: Calculate the matching degree of employees 51-100 with all posts - Task 3: Calculate the matching degree of employees 101-150 with all posts - Task 4: Calculate the matching degree of employees 151-200 with all posts

[0184] Each sub-task is executed on an independent computing node, and the final result is merged.

[0185] Step 2: Opportunity identification and sorting

[0186] The system analyzes the calculation results and identifies high-potential matching opportunities:

[0187] Employee A's current post: Sales Manager (matching coefficient 0.78) Potential best matching post: 1. Marketing Manager (matching coefficient 1.23) 2. Product Manager (matching coefficient 1.15) 3. Training Manager (matching coefficient 1.08)

[0188] Step 3: Development path planning

[0189] Based on the organizational structure and career development path data, the system generates development suggestions:

[0190] Short-term opportunity (within 6 months): Lateral transfer to Marketing Manager position, improve matching degree by 45% Mid-term planning (1-2 years): Improve product knowledge through training, prepare for Product Manager position Long-term goal (3-5 years): Develop management skills and develop towards Marketing Director The system generates a talent flow recommendation report based on the analysis results and pushes it to HR and relevant managers through the messaging system. Through the detailed description of the above embodiments, those skilled in the art can understand the technical principles and implementation methods of the present invention, and make corresponding adjustments and optimizations according to specific application scenarios. The present invention provides a comprehensive, intelligent, and scalable job-post matching quantitative evaluation solution, providing strong technical support for modern enterprise human resource management.

Claims

1. A quantitative assessment method for person-job matching based on a competency model, characterized in that, Includes the following steps: Step 1: Construct a three-level hierarchical structure that includes core competency dimensions, general competency dimensions, and professional competency dimensions. Set five secondary indicators under each primary dimension to form a unified competency evaluation model. Step 2: By constructing the judgment matrix, calculating the eigenvectors, and performing consistency checks, determine the weight values ​​for each dimension and generate the standard score vector for the job position; Step 3: Collect multi-source data including employee self-evaluation, supervisor evaluation, peer evaluation, subordinate evaluation, and customer evaluation, and generate employee competency scores through weighted fusion and outlier detection. Step 4: Calibrate employee competency scores using performance adjustment coefficients, calculate the cosine similarity and Euclidean distance between the calibrated score vector and the job standard score vector, and fuse them to obtain the person-job matching coefficient; Step 5: Through capability gap identification and training resource matching, generate a personalized development plan that includes training content, training methods, and training cycle.

2. The method according to claim 1, characterized in that, The construction of the unified competency evaluation model in step one specifically includes: The core competency dimensions include five secondary indicators: value alignment, organizational commitment, professional ethics, teamwork, and communication. The general competence dimension includes five secondary indicators: learning ability, innovation ability, execution ability, decision-making ability, and resilience. The professional competence dimension includes five secondary indicators: professional knowledge, professional skills, work experience, qualification certification, and business insight.

3. The method according to claim 1, characterized in that, Step two involves quantifying the target job position based on the Analytic Hierarchy Process (AHP), specifically including: Construct a judgment matrix and compare each dimension and each secondary indicator of the unified competency evaluation model pairwise to determine their relative importance; Calculate the eigenvectors of the judgment matrix to obtain the initial weights of each dimension and each secondary indicator; perform a consistency check, and determine the final weight value when the consistency ratio is less than a preset threshold. Based on the final weight value and the preset standard score range, a standard score vector for the target position is generated.

4. The method according to claim 1, characterized in that, The collection and processing of evaluation data in step three specifically includes: Collect data on employee self-evaluation, supervisor evaluation, peer evaluation, subordinate evaluation, and customer evaluation; The evaluation is weighted and integrated according to the following criteria: employee self-evaluation weight 0.2, superior evaluation weight 0.3, peer evaluation weight 0.2, subordinate evaluation weight 0.15, and customer evaluation weight 0.

15. An outlier detection method based on quartiles is adopted to remove evaluation values ​​that exceed the range of the first quartile minus 1.5 times the interquartile range or the third quartile plus 1.5 times the interquartile range.

5. The method according to claim 1, characterized in that, Step four, which involves calibrating employee competency scores using performance adjustment coefficients, specifically includes: Obtain employee performance appraisal data for the past 12 months; The performance adjustment coefficient is calculated based on the difference between the performance completion rate and the 100% benchmark value; The calibrated overall employee score is obtained by multiplying the employee's competency score by the performance adjustment factor. The formula for calculating the performance adjustment coefficient is as follows: In the formula, α is the performance adjustment coefficient, with a value range of [0.6, 1.4]; P actual This represents the actual performance completion rate, expressed as a percentage; P baseline The baseline performance completion rate is 100%.

6. The method according to claim 1, characterized in that, Step four, calculating the person-job matching coefficient, specifically includes: Calculate the cosine similarity between the 15-dimensional employee comprehensive score vector and the job standard score vector; Calculate the Euclidean distance between the two vectors and normalize it by dividing by the maximum possible distance; The similarity and normalized distance are weighted and fused to obtain the person-job matching coefficient; The formula for calculating the person-job matching coefficient is as follows: In the formula, M is the human-job matching coefficient, with a value range of [0,1]; cosθ is the cosine similarity; d is the Euclidean distance; d max The maximum Euclidean distance is equal to w1 is the similarity weight, with a value of 0.6; w2 is the distance weight, with a value of 0.

4.

7. The method according to claim 1, characterized in that, It also includes a context-adaptive weight adjustment step: obtaining the employee's current project stage identifier, team size, and task type code through the project management system interface; adjusting the weights of the three primary dimensions according to the preset context-weight mapping table, where the weight of the core capability dimension in the project initiation stage is increased by 0.1, the weight of the professional capability dimension in the project execution stage is increased by 0.1, and the weight of the general capability dimension in the project closure stage is increased by 0.

1. The job standard score vector and the person-job matching coefficient are recalculated based on the adjusted weights.

8. The method according to claim 1, characterized in that, It also includes data sparsity processing steps: The missing data rate for each secondary indicator is calculated, and a missing data rate exceeding 30% is defined as data sparsity. Employees with the same job position as the target employee and whose evaluation completeness exceeds 80% are selected as the similar group. The similarity is calculated using the Pearson correlation coefficient. The evaluation mean of the top 5 employees with the highest similarity is used to fill in the missing data. A prior distribution is constructed based on the historical evaluation data of the position in the expert knowledge base, and the expected value of the posterior probability distribution is calculated in combination with the existing evaluation data as the missing evaluation value. Multiply the sparse dimension weights of the data by the reliability coefficient, and the reliability coefficient is equal to 1 minus the missing rate.

9. The method according to claim 1, characterized in that, It also includes new employee processing steps: Employees who have been with the company for less than 6 months will be identified as new employees; A linear regression model is constructed based on education level codes, years of work experience, number of skill certificates, and pre-employment assessment scores to predict potential performance coefficients. For the first 3 months of new employees, the potential prediction coefficient is used entirely. From the 4th to the 6th month, the potential prediction weight decreases by 20% each month, while the actual performance weight increases by 20% each month. After 6 months, the actual performance adjustment coefficient is used entirely.

10. The method according to claim 1, characterized in that, It also includes a job opportunity scanning step: Store the 15-dimensional standard subvectors of all positions within the organization; Calculate the matching coefficient between the employee's overall score vector and the standard score vectors for all positions; Select positions with a matching coefficient higher than that of the current position as development opportunities; Based on a pre-defined job hierarchy matrix and career development path map, the target job with the highest matching coefficient and that meets the promotion criteria is recommended.

Images