Post grading intelligent import and matching data consistency processing system based on variable enterprise architecture

Abstract

The application discloses a post grading intelligent import and matching data consistency processing system based on a variable enterprise architecture, and relates to the technical field of enterprise management data acquisition and processing. A data consistency processing system comprising a data acquisition post cluster determination module, a weight calculation module, a self-adaptive grading module and the like is constructed to acquire enterprise organization architecture data and post description information to be processed, fuse and calculate the weights of each post cluster to be graded of an enterprise organization unit, and finally obtain an optimized post grading result for post value matching and salary mapping. The application realizes intelligent post clustering and sequence division by introducing an enterprise architecture meta-model and a post attribute feature vector, effectively solving the evaluation standard confusion problem caused by mixed evaluation of different sequence posts in traditional methods. By constructing a responsibility complexity calculation model and an interference coefficient evaluation mechanism, the post cluster to be graded with fuzzy responsibility definition and unobvious value characteristics can be accurately identified, and efficient, transparent and reliable processing of massive post data is realized.

Description

Technical Field

[0001] This invention relates to the field of enterprise management data acquisition and processing technology, specifically to a system for intelligent import and matching data consistency processing based on job classification under a variable enterprise architecture. Background Technology

[0002] Job value assessment is a fundamental task in building a corporate compensation management system. It systematically measures and ranks the relative value of different positions within an organization to ensure the internal fairness and external competitiveness of the compensation system. Currently, commonly used job value assessment methods mainly include ranking, classification, factor comparison, factor point method, Hay job evaluation method, and artificial intelligence methods.

[0003] Among existing technologies, the factor point method and the Hay evaluation method are the most widely used. The factor point method determines job value by quantifying and scoring multiple pay factors. Its advantage is its clear differentiation between jobs at the same level and category. However, it cannot effectively compare the value of jobs at different levels and categories, and the evaluation process relies on subjective judgment, making it susceptible to human interference. The Hay evaluation method focuses on three general factors: knowledge and skills, problem-solving ability, and job responsibility. It is effective in differentiating between jobs at different levels and categories, but it suffers from problems such as a complex evaluation process, high computational resource costs, difficulty in adjustment, and long computation time. Furthermore, it is not effective in differentiating between jobs at the same level and category. While artificial intelligence methods have improved deployment convenience in recent years, their deployment costs in government and enterprise are high, and the opaque reasoning process makes it difficult for analysts to intuitively interpret and backtrack the data. There are also data leakage and security risks. Traditional job classification methods have significant shortcomings in ultra-large-scale, multi-level organizational structures. They are inefficient in processing massive amounts of job data, slow in classification speed, and struggle to ensure consistency when integrating job description data from multiple sources. They also exhibit poor matching accuracy when dealing with heterogeneous data sources. When the organizational structure is dynamically adjusted, inconsistencies in the classification results and data distortion are likely to occur. Furthermore, the overall processing relies on human experience, the logic is not transparent, key nodes cannot be quantified, and problems are difficult to trace back, making it impossible to adapt to the complex and ever-changing enterprise management needs.

[0004] Existing methods fail to adequately consider the impact of the stability of job hierarchical relationships on value assessment, leading to difficulties in accurately classifying complex jobs with ambiguous responsibilities and frequent cross-departmental collaborations. Current technologies often employ uniform assessment parameters, failing to adaptively adjust for subsets of jobs with high responsibilities, resulting in discrepancies between assessment results and actual job value. Furthermore, existing methods lack dual analysis of the distribution characteristics of job value assessment dimensions and organizational affiliation indicators, making it difficult to identify and correct the impact of changes in job hierarchical relationships caused by organizational restructuring (such as business reorganization or departmental mergers) on classification accuracy. Therefore, this paper proposes an intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture. Summary of the Invention

[0005] The purpose of this invention is to provide a system for intelligent import and matching data consistency processing of job classification based on a variable enterprise architecture, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A system for intelligent import and matching data consistency processing of job classification under a variable enterprise architecture, characterized by comprising: The data acquisition module is used to acquire the enterprise's organizational structure data and job description information to be processed; The job cluster determination module is used to take each organizational unit in the enterprise organizational structure data as the target organizational unit, and determine each job cluster to be assigned to the target organizational unit based on the attribute feature vector of each job in the target organizational unit and the enterprise architecture meta-model. The weight calculation module is used to determine the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension and the affiliation and association indicators in the organizational hierarchy, and to merge the distribution characteristic indicators and the affiliation and association indicators to obtain the influence weight. The adaptive grading module uses the influence weights to adaptively adjust the grading parameters and obtain optimized job grading results.

[0007] As another optional technical solution, based on the intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture, the system acquires the enterprise organizational structure data and job description information to be processed, takes each organizational unit in the enterprise organizational structure data as the target organizational unit, and determines each job cluster to be classified in the target organizational unit according to the attribute feature vector of each job in the target organizational unit and the enterprise architecture meta-model; the job cluster to be classified is a set of jobs with similar value characteristics. The distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension and the affiliation and association indicators in the organizational hierarchy are determined. The distribution characteristic indicators and the affiliation and association indicators are integrated to obtain the influence weight of each job cluster on the accuracy of the enterprise's job classification. The classification parameters of the corresponding job clusters to be classified in the target organizational unit are adaptively adjusted using the influence weights to obtain optimized job classification parameters. Then, the optimized job classification results of each organizational unit are obtained to perform intelligent matching and salary mapping of enterprise job value.

[0008] As a preferred technical solution, the enterprise architecture metamodel is dynamically updated. The system analyzes historical organizational structure change logs and job adjustment records, using graph representation learning techniques to learn implicit vector representations for different types of organizational units (such as functional departments, business units, and project teams). When new organizational units or job types emerge, the system can map them to approximate categories in the metamodel based on their relationships and descriptive text, enabling the metamodel to adaptively expand to reflect organizational evolution. This solves the technical problem of traditional methods failing to evaluate models due to organizational structure changes.

[0009] As a specific solution to the technical solution of this application, the step of determining the various job clusters to be assigned in the target organizational unit based on the attribute feature vectors of each position in the target organizational unit and the enterprise architecture meta-model includes: The target organizational unit is divided into several job subsets. Based on the attribute feature vector of each job in each job subset, the degree of responsibility difference between any two job subsets is analyzed to determine the job subsets corresponding to each job sequence. Each benchmark job in each job subset and its matching degree are determined, where the matching degree is the minimum value among all attribute feature differences between two job subsets of the same job sequence, and the benchmark jobs are the two jobs corresponding to the minimum value. Based on the benchmark jobs and their matching degrees, the responsibility complexity of each job subset is determined. Based on the scores of each job in each job subset on the job value assessment elements, the average element score of each job subset is determined. Based on the responsibility complexity and the average element score of each job subset, the interference coefficient of each job subset on the accuracy of job classification is determined. Job subsets with interference coefficients not less than a preset interference threshold are designated as job clusters to be classified.

[0010] As a specific solution to the technical solution of this application, the step of analyzing the degree of difference in responsibilities between every two job subsets based on the attribute feature vector of each job in each job subset, and determining the job subsets corresponding to each job sequence, includes: Calculate the difference between the attribute feature vector of the target position in each job subset and the attribute feature vector of each position in the second job subset, and record it as the feature difference value. Select the minimum feature difference value. Take the average of the minimum feature difference values ​​corresponding to all positions in the first job subset as the degree of responsibility difference between the first job subset and the second job subset. The first job subset and the second job subset are two different job subsets in the target organizational unit, and the target position is any position in the first job subset. Based on the degree of responsibility difference between each pair of job subsets, perform clustering on all job subsets in the target organizational unit to obtain each job cluster. Determine each job subset in the same job cluster as each job subset corresponding to the same job sequence.

[0011] As a specific solution to the technical solution of this application, the step of determining the responsibility complexity of each job subset based on the benchmark jobs and their matching degrees of each job subset includes: For each job subset, the benchmark jobs that appear repeatedly in the benchmark job set corresponding to the job subset are identified as key jobs. The number of benchmark jobs corresponding to each key job, as well as the number of benchmark jobs and key jobs within the job subset, are counted. The responsibility complexity of a job subset is determined based on the number of benchmark jobs and minimum matching degree corresponding to each key job, and the first ratio of the number of key jobs to the number of benchmark jobs in the job subset. The responsibility complexity is negatively correlated with the number of benchmark jobs and the first ratio, and positively correlated with the minimum matching degree.

[0012] As a specific solution to the technical solution of this application, the step of determining the interference coefficient of each job subset on the accuracy of job classification based on the responsibility complexity and the average score of each element of each job subset includes: For each job subset, obtain the degree of responsibility difference between every two job subsets corresponding to the job sequence to which the job subset belongs, and select the minimum degree of responsibility difference; A fusion analysis is performed on the responsibility complexity, the mean score of the elements, and the degree of difference in the minimum responsibility of the job subset to determine the interference coefficient of the job subset on the accuracy of job classification. The interference coefficient is positively correlated with the complexity of the duties, and negatively correlated with the mean score of the elements and the degree of difference in the minimum duties.

[0013] As a specific solution to the technical solution of this application, determining the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension and the affiliation and association indicators in the organizational hierarchy includes: Obtain each key position in each job cluster to be classified in the target organizational unit, and then determine the deviation vector of each key position relative to the enterprise standard job level in terms of job value assessment elements. Determine the value gap and grade span between each key position and its corresponding standard position level to form a value assessment vector. Based on the value assessment vector and position code of each key position in each position cluster to be classified, determine the distribution characteristic indicators of each position cluster to be classified in the position value assessment dimension. A graph analysis is performed on the hierarchical relationships of each key position in the organizational hierarchy to obtain the organizational affiliation depth index for each key position in each position cluster to be assigned to. Based on the organizational affiliation depth index and the proportion of positions in each position cluster to be assigned to in the organizational hierarchy, the affiliation relationship index of each position cluster to be assigned to in the organizational hierarchy is determined.

[0014] As a specific solution to the technical solution of this application, the step of determining the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension based on the value assessment vector and job code of each key job in each job cluster to be classified includes: For each job cluster to be classified, the value gap modulus, grade span angle, and code distance between each two key jobs are determined based on the value assessment vector and job code of each key job in the job cluster. The cluster similarity between each pair of key positions is determined based on the value gap modulus, the grade span angle, and the coding distance. The cluster similarity is then used to perform hierarchical clustering of all key positions in the position cluster to be classified, resulting in each value grade class. Based on the angle between the grade spans of every two key positions in each value grade category and the value gap modulus of each key position, the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension are determined.

[0015] As a specific solution to the technical solution of this application, the step of determining the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension based on the angle between the grade spans of every two key positions in each value grade category and the value gap modulus of each key position includes: Calculate the average value of the span angle of all levels in each value level class and the variance of the value gap modulus to obtain the value consistency index of each value level class. The level class with the value consistency index greater than the preset consistency threshold is used as the reference level class. The value consistency index is negatively correlated with the average value and the variance. Based on the number of grade categories corresponding to the job clusters to be classified, the value consistency index of each reference grade category, and the number of key positions, the distribution characteristic index of the job clusters to be classified in the job value assessment dimension is determined. The distribution characteristic index is negatively correlated with the number of grade categories and positively correlated with both the value consistency index and the number of key positions.

[0016] As a specific solution to the technical solution of this application, the step of determining the affiliation index of each job cluster to be assigned in the organizational hierarchy based on the organizational affiliation depth index and the proportion of the number of jobs in each job cluster to be assigned in the organizational hierarchy includes: Obtain the number of positions in each job cluster to be assigned in the organizational hierarchy, and use the second ratio of the number of positions in the job cluster to the total number of positions in the target organizational unit as the hierarchy weight. The organizational affiliation depth index is weighted and summed based on the hierarchical weight of the job clusters to be classified in each organizational level to obtain the degree of influence of each organizational level on the accuracy of job classification. For each key position, the organizational affiliation index is determined based on the degree of influence of each organizational level on the accuracy of position classification and the third ratio of the organizational affiliation depth index to the total organizational levels. The organizational affiliation index is negatively correlated with the degree of influence and positively correlated with the third ratio. A comprehensive analysis of the organizational affiliation indicators of all key positions within the same job cluster to be assigned to in the target organizational unit is conducted to obtain the affiliation association indicators of each job cluster to be assigned to in the organizational hierarchy.

[0017] As a specific solution to the technical solution of this application, the step of integrating the distribution characteristic index and the membership association index to obtain the influence weight of each job cluster to be classified on the accuracy of enterprise job classification includes: For each job cluster to be classified, calculate the product of the distribution characteristic index and the membership association index of the job cluster to be classified. The product is subjected to negative correlation normalization to obtain the influence weight of the job cluster to be classified on the accuracy of enterprise classification.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This application achieves intelligent clustering and sequence division of positions by introducing an enterprise architecture meta-model and job attribute feature vectors, effectively solving the problem of confusion in evaluation standards caused by mixed evaluation of positions with different sequences in traditional methods. By constructing a responsibility complexity calculation model and an interference coefficient evaluation mechanism, it can accurately identify job clusters with ambiguous responsibility definitions and unclear value characteristics, providing a basis for subsequent refined classification processing.

[0019] Meanwhile, this invention constructs a weighted calculation model with dual influence by integrating the distribution characteristic indicators of job value assessment dimensions and the affiliation and association indicators in the organizational hierarchy. This effectively quantifies the degree of influence of job clusters to be classified on the accuracy of job classification in enterprises. Furthermore, by optimizing job classification parameters through an adaptive parameter adjustment mechanism, it significantly improves the objectivity and accuracy of job value assessment, ensuring the internal fairness and external competitiveness of the enterprise's compensation system. It is particularly suitable for job classification management in the scenario of dynamic adjustment of organizational structure in large enterprise groups.

[0020] Furthermore, this invention utilizes interference coefficients to quantify the impact of job subsets on classification accuracy, enabling automatic filtering and priority sorting of massive job data. This solves the problems of low efficiency in large-scale data processing and poor matching accuracy of heterogeneous data sources. In addition, combined with an adaptive classification parameter adjustment mechanism, it can respond to dynamic changes in organizational structure, ensuring data consistency of job classification results. Moreover, through a highly reliable calculation process, it achieves transparency and traceability of the processing, significantly improving the system's adaptability to complex organizational structures. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the process of the intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture according to the present invention; Figure 2 This is a flowchart illustrating the process of determining the various job clusters to be assigned to the target organizational unit according to the present invention. Detailed Implementation

[0022] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the technical solution proposed according to the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form.

[0023] Unless otherwise defined, the technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] The application scenarios targeted by this invention can be: In corporate human resource management practice, as organizations expand and business structures become more complex, companies need to establish a scientific job evaluation system to achieve internal fairness in compensation. Traditional job evaluation methods (such as factor point method and Hay method) mainly rely on expert manual scoring, which suffers from problems such as inconsistent evaluation standards, strong subjectivity, and low efficiency. Especially in large enterprise groups, where there are numerous positions and complex hierarchical relationships, manual evaluation is difficult to cope with the dynamic adjustment needs of the organizational structure.

[0025] To ensure the accuracy and consistency of job classification and improve the automation level of job value assessment, one embodiment of the present invention provides a job classification intelligent import and matching data consistency processing system based on a variable enterprise architecture, such as... Figure 1 As shown, it includes the following steps: S1: Obtain the organizational structure data and job description information of the enterprise to be processed.

[0026] In one embodiment, organizational structure data of the enterprise to be processed is extracted through an Enterprise Resource Planning (ERP) system or a Human Capital Management (HCM) system. This includes information such as departmental hierarchy, job positions, job descriptions, and qualification requirements. The extracted data is then cleaned and standardized to obtain structured enterprise organizational structure data.

[0027] Data cleaning includes, but is not limited to: removing null values, standardizing job naming conventions, standardizing department codes, and eliminating redundant job information; standardization processing includes converting job description information into attribute feature vectors, and implementers can set feature extraction dimensions according to specific circumstances, such as job contribution, knowledge and skills requirements, communication and coordination abilities, and problem-solving complexity.

[0028] In a preferred embodiment, unstructured or semi-structured data, such as job descriptions, requirements, performance goals, and reporting relationships, are collected from enterprise ERP systems, HRMS systems, project management systems, and internal document libraries. Due to inconsistencies in data models across different systems, this system pre-constructs an enterprise human resources data body, defining core concepts such as "job title," "skills," "responsibilities," and "organizational units," as well as the relationships between them. The system utilizes semantic mapping technology to automatically or semi-automatically map data fields from different sources onto this ontology, resolving issues related to synonyms (such as "position" and "job post") and structural differences, forming a preliminary standardized job-organization relationship data graph.

[0029] In one embodiment, natural language processing technology is used to perform semantic parsing on the job description text, extract quantitative scores of job value assessment elements (such as knowledge and skills level, problem-solving ability, scope of responsibilities, etc.), and construct attribute feature vectors for each job.

[0030] The dimensions of the attribute feature vector can correspond to the three compensation elements of the Hay job evaluation method (knowledge and skills level, problem-solving ability, and scope of responsibility), or to multiple evaluation dimensions of the factor point method. Implementers can configure them according to the evaluation model actually used by the company.

[0031] In a preferred embodiment, for unstructured job description text, a pre-trained language model (such as BERT or ERNIE) can be used for sequence labeling and relation extraction. Specifically, the system identifies and extracts entities in the text related to job value assessment elements, such as 'required skills' (corresponding to 'knowledge and ability level'), 'projects under responsibility' (corresponding to 'scope of responsibility'), and 'problem-solving type' (corresponding to 'problem-solving ability'). Then, an attention mechanism is used to calculate the weights of these entities in the description text, and combined with their entity types, a multi-dimensional, dense attribute feature vector is generated. This method can better capture semantic information and solve the problem of "polysemy" in traditional text analysis.

[0032] In a preferred embodiment, the organizational structure data can also be denoised to remove abnormal data such as virtual positions and duplicate positions caused by historical reasons, making the organizational structure data more standardized and providing an accurate data foundation for subsequent position classification analysis.

[0033] It is worth noting that current manual job evaluation methods mainly rely on the subjective judgment of the evaluator. For complex positions with ambiguous job descriptions and frequent cross-departmental collaboration, traditional evaluation methods struggle to accurately determine their value level. Therefore, this paper proposes an intelligent clustering analysis based on the enterprise architecture meta-model and job attribute feature vectors, adaptively adjusting parameters for the job clusters to be classified, to ensure the objectivity and accuracy of job classification.

[0034] Thus, this embodiment has obtained the organizational structure data and job description information of the enterprise to be processed.

[0035] S2 takes each organizational unit in the enterprise organizational structure data as the target organizational unit, and determines the various job clusters to be assigned to the target organizational unit based on the attribute feature vector of each position in the target organizational unit and the enterprise architecture meta-model.

[0036] Here, the job cluster to be classified is a set of jobs with similar value characteristics within the target organizational unit.

[0037] In actual corporate management, organizational structures often exhibit multi-level and multi-sequence characteristics. Within the same department, different career development paths, such as management, professional, and technical tracks, may exist simultaneously. Mixing positions from different tracks together for value assessment can lead to confusion in assessment standards and affect the accuracy of job classification. This is especially true in large group companies, where job types are diverse and responsibilities are complex and overlapping, necessitating the use of intelligent clustering methods to identify groups of positions with similar value characteristics.

[0038] Each organizational unit in the enterprise organizational structure data needs to undergo job cluster analysis. However, for ease of understanding, this article takes any organizational unit as an example to perform job classification analysis. Any organizational unit in the enterprise organizational structure data can be used as the target organizational unit.

[0039] As an exemplary implementation, the various job clusters to be assigned to the target organizational unit are determined as follows: Figure 2 As shown, it includes: S21 divides the target organizational unit into several job subsets.

[0040] In organizational structure data, some positions may be difficult to categorize directly due to unclear job descriptions or cross-functional collaboration. In such cases, analyzing positions with significantly different responsibilities together with those with clearly defined responsibilities could lead to biased job valuations. Therefore, clustering positions with similar job characteristics within an organizational unit based on their attribute features avoids interference from differences in responsibilities and improves the accuracy of job classification.

[0041] Among them, the attribute characteristics of a position refer to the quantitative performance of the position in various dimensions of value assessment. For example, management positions usually score higher in the scope of responsibility dimension, while technical positions score higher in the knowledge and skills level dimension.

[0042] In one embodiment, density-based clustering (DBSCAN) or K-means clustering is used to divide the target organizational unit into several job groups with similar attribute characteristics, denoted as job subsets.

[0043] In this approach, each subset of job positions corresponds to a preliminary job category, and the clustering algorithm can be implemented in various forms. Implementers can also use other partitioning methods, such as hierarchical clustering algorithms or rule-based classification methods; no specific limitations are imposed here.

[0044] S22, based on the attribute feature vector of each job in each job subset, analyze the degree of difference in responsibilities between every two job subsets, and determine the job subsets corresponding to each job sequence.

[0045] Here, defining each job subset as corresponding to each job sequence refers to each job subset belonging to the same job sequence type. This facilitates the analysis of job classification interference among job subsets of the same job sequence type.

[0046] In one embodiment, the Euclidean distance or cosine similarity between the feature vectors of job attributes is calculated to obtain the feature difference value between every two jobs in each job subset.

[0047] Among them, the distance calculation between attribute feature vectors can reflect the degree of difference in the value assessment dimension of positions. Since positions in different sequences have significant differences in specific dimensions (such as the significant difference between management positions and professional positions in the dimension of responsibility scope), feature difference analysis can effectively distinguish the types of position sequences.

[0048] As an exemplary implementation, determining each subset of positions corresponding to each job sequence includes: The first step is to calculate the difference between the attribute feature vector of the target position in the first job subset and the attribute feature vector of each position in the second job subset, and record it as the feature difference value. Select the minimum feature difference value, and take the average of the minimum feature difference values ​​corresponding to all positions in the first job subset as the degree of difference in responsibilities between the first job subset and the second job subset.

[0049] Here, the degree of difference in responsibilities refers to the situation where the responsibilities of two job subsets are not similar. The first job subset and the second job subset are two different job subsets in the target organizational unit, and the target job is any job in the first job subset.

[0050] In one embodiment, the first job subset is denoted as the q-th subset, the second job subset is denoted as the p-th subset, and the target job is denoted as the k-th job, where q, p, and k are all positive integers. The feature difference value between the attribute feature vector of the k-th job in the q-th subset and the attribute feature vector of all jobs in the p-th subset is obtained. The minimum feature difference value is taken as the responsibility difference factor of the k-th job in the q-th subset. Then, the average value of the responsibility difference factors of all jobs in the q-th subset is taken as the degree of responsibility difference between the q-th subset and the p-th subset. The feature difference value is the Euclidean distance or Manhattan distance between the attribute feature vectors of the two jobs.

[0051] The second step is to cluster all job subsets in the target organizational unit according to the degree of difference in responsibilities between each pair of job subsets to obtain each job cluster, and then identify each job subset in the same job cluster as the job subset corresponding to the same job sequence.

[0052] In one embodiment, the degree of difference in responsibilities is used as the clustering distance. By using the clustering distance between every two job subsets to cluster all job subsets in the target organizational unit, job clusters can be obtained, and each job cluster represents the same job sequence.

[0053] It is worth noting that, since the job subsets of each job cluster have a high degree of consistency in value characteristics, they can effectively avoid interference from value differences between different job sequences and improve the accuracy of identifying job clusters to be classified.

[0054] S23, determine the baseline positions and their matching degree in each position subset.

[0055] Here, the matching degree is the minimum value among all feature differences between two subsets of the same job sequence, and the benchmark jobs are the two jobs corresponding to the minimum value. The larger the matching degree, the more obvious the difference in responsibilities between the two benchmark jobs.

[0056] In one embodiment, for the j-th job sequence, the minimum value among the attribute feature vectors of all jobs in the q-th subset and the feature differences of all jobs in the p-th subset is obtained. The two jobs corresponding to the minimum value are denoted as a benchmark job in the q-th subset and a benchmark job in the p-th subset. The minimum value is used as the matching degree between the benchmark jobs in the q-th subset and the p-th subset, where j is a positive integer.

[0057] By referring to a benchmark position in the q-th subset and a benchmark position in the p-th subset, and the method for determining their matching degree, we can obtain each benchmark position and its matching degree in each position subset.

[0058] S24. Based on the baseline positions and their matching degree of each position subset, determine the responsibility complexity of each position subset. Based on the scores of each position in each position subset on the position value assessment elements, determine the average element score of each position subset.

[0059] Here, responsibility complexity refers to the clarity of the responsibility definition of a subset of positions. The greater the responsibility complexity, the greater the interference of the subset of positions with the classification of positions.

[0060] As an exemplary implementation, the responsibility complexity of each subset of job positions is determined, including: The first step is to identify the key positions as those that appear repeatedly in the benchmark position set corresponding to each job subset, and to count the number of benchmark positions corresponding to each key position, as well as the number of benchmark positions and key positions within each job subset.

[0061] Here, a key position refers to a position in a job subset whose minimum characteristic difference value is formed by a benchmark position in one job subset and a benchmark position in multiple job subsets of the same job sequence. The more key positions there are in a job subset, the clearer the definition of responsibilities of the job subset is, and conversely, the lower the responsibilities complexity will be.

[0062] The second step is to determine the responsibility complexity of the job subset based on the number of baseline jobs and minimum matching degree corresponding to each key job, as well as the first ratio of the number of key jobs to the number of baseline jobs within the job subset.

[0063] Here, the minimum matching degree refers to the minimum matching degree among all benchmark positions corresponding to the key position. If a minimum value exists, any matching degree is selected as the minimum matching degree. The complexity of responsibilities is negatively correlated with the number of benchmark positions and the first ratio, and positively correlated with the minimum matching degree. Negative correlation means that the larger the independent variable, the smaller the dependent variable, while positive correlation means that the larger the independent variable, the larger the dependent variable.

[0064] As an example, the formula for calculating the responsibility complexity of the q-th job subset can be:

[0065] in, This represents the degree of responsibility for the q-th job subset. This represents an exponential function with the natural constant as its base. This is used to implement negative correlation processing, where L represents the number of key positions in the q-th job subset, and N represents the number of baseline positions within the q-th job subset. This represents the first ratio of the number of key positions to the number of baseline positions within the q-th position subset. Indicates the serial number of the key position. This indicates the position in the q-th job subset. The number of benchmark positions corresponding to each key position This indicates the position in the q-th job subset. The minimum matching degree for each key position.

[0066] In the formula for calculating the complexity of job duties, the ratio of the number of key positions to the number of baseline positions is referred to as the first ratio to distinguish it from the other ratios in this embodiment, namely the second and third ratios. The smaller the value, the fewer the number of key positions matched in the q-th job subset, the less clear the responsibilities of the q-th job subset, the greater the complexity of the responsibilities, and the fewer the base positions corresponding to the key positions. The larger the value, the higher the clarity of responsibilities and the lower the complexity of responsibilities in the q-th job subset, and the lower the minimum matching degree corresponding to the key positions. The smaller the value, the better the benchmark position in the q-th job subset matches the benchmark positions in other subsets of the same job sequence, and the clearer the responsibilities of the q-th job subset are defined.

[0067] As an exemplary implementation, determining the mean element score for each subset of job positions includes: The first step is to obtain quantitative scores for all positions within each job subset in the target organizational unit on job value assessment elements (such as knowledge and skills level, problem-solving ability, and scope of responsibility).

[0068] In one embodiment, the Hay job evaluation method or the factor point method is used to score the elements of each job.

[0069] The second step is to calculate the average element score of all positions in the same job subset based on the element scores of all positions within each job subset.

[0070] In this embodiment, the smaller the average score of the elements, the lower the overall value level of the corresponding job subset, and the more likely it is to belong to the basic operation category. The higher the average score of the elements, the higher the overall value level of the job subset, and the more likely it is to belong to the senior management or core technology category.

[0071] S25. Based on the responsibility complexity and element score mean of each job subset, determine the interference coefficient of each job subset on the accuracy of job classification.

[0072] Here, the interference coefficient refers to the impact of the characteristic attributes of a job subset on the job classification process.

[0073] As an exemplary implementation, the interference coefficient of each job subset on the accuracy of job classification is determined, including: The first step is to obtain the degree of responsibility difference between every two job subsets corresponding to the job sequence of each job subset, and select the minimum degree of responsibility difference.

[0074] In this embodiment, the greater the degree of minimum responsibility difference, the smaller the similarity of responsibility features among all job subsets corresponding to the job sequence of the job subset. This further indicates that the clustering results when determining the job sequence are less reliable, and the lower the confidence level of the responsibility complexity and the mean score of the job subset, the less trustworthy it is.

[0075] The second step involves a fusion analysis of the responsibility complexity, mean element score, and minimum responsibility difference of the job subsets to determine the interference coefficient of the job subsets on the accuracy of job classification.

[0076] Here, the interference coefficient is positively correlated with the complexity of duties, and negatively correlated with the mean score of elements and the degree of difference in minimum duties.

[0077] When integrating heterogeneous job description data from different sources such as ERP and HCM, the challenge of consistency often arises due to inconsistent data standards and semantic conflicts. The interference coefficient, by fusing multi-dimensional features such as job complexity and the mean score of elements, constructs a unified metric across data sources. When data from different sources conflict in semantic mapping (e.g., a job is defined as a management position in system A and a technical position in system B), its attribute feature vectors will exhibit high dispersion, directly leading to an abnormally high calculated interference coefficient. Based on this, the system can automatically identify anomalous clusters of inconsistent data and mark them as job clusters to be reclassified for specialized cleaning or manual review, rather than blindly reclassifying them. This anomaly detection mechanism based on the interference coefficient effectively prevents low-quality heterogeneous data from contaminating the overall classification model, ensuring matching accuracy in complex data environments.

[0078] As an example, the formula for calculating the interference coefficient of the q-th job subset of the j-th job sequence on the accuracy of job classification can be:

[0079] In the formula, This represents the interference coefficient of the q-th job subset of the j-th job sequence in the Z-th organizational unit on the accuracy of job classification. Represents the normalization function. This represents the complexity of the responsibilities of the q-th job subset. This represents the minimum degree of responsibility difference corresponding to the j-th job sequence in the Z-th organizational unit. Let represent the mean element score of the q-th job subset of the j-th job sequence in the Z-th organizational unit. This represents a non-zero constant (an empirical value of 0.001 is acceptable).

[0080] In the formula for calculating the interference coefficient, and Generally, it is impossible for the value to be zero, but in order to avoid extreme cases, a non-zero constant is added to the denominator of the fraction, such as taking an empirical value of 0.001.

[0081] S26. The subset of positions with interference deflection coefficients not less than the preset interference threshold is designated as the position cluster to be classified.

[0082] Here, the larger the interference coefficient, the greater the likelihood that the job subset is a job cluster to be classified, indicating that the job subset has an unclear definition of responsibilities or unclear value characteristics, and needs to be given special attention and refined classification.

[0083] In one embodiment, the interference coefficient ranges from 0 to 1, and the preset interference threshold can be set to 0.7. The subset of positions with an interference coefficient not less than 0.7 is taken as the position cluster to be classified, that is, the subset of positions with an interference coefficient greater than or equal to 0.7 is taken as the position cluster to be classified, thus obtaining each position cluster to be classified in the target resistive unit. The preset interference threshold can be set by the implementer according to the specific actual situation, and is not specifically limited here.

[0084] Thus, this embodiment has obtained the clusters of job positions to be assigned in the target organizational unit.

[0085] To address the issues of heavy computational load and slow classification processing speed when handling massive, multi-level organizational structure data, the interference coefficient plays a crucial role in intelligent task allocation within the system. Traditional methods often perform indiscriminate complex calculations on all positions, leading to system sluggish response when data volume surges. This application introduces an interference coefficient to pre-scan and quantify a subset of massive positions. The system only needs to initiate high-dimensional deep computation processes (such as the fusion analysis of distribution characteristic indicators and affiliation association indicators) for position clusters to be classified whose interference coefficient exceeds a preset threshold (i.e., complex positions with ambiguous responsibilities and unclear characteristics). For regular positions with lower interference coefficients, the standard classification path is directly used. This mechanism concentrates computational resources on the most demanding "difficult samples," avoiding ineffective computational waste on massive, clearly defined positions, thereby significantly improving the efficiency and speed of automatic classification of massive organizational structure position data.

[0086] To address the issues of inconsistent data and a black-box processing method that makes backtracking difficult during dynamic organizational restructuring (such as departmental mergers and business reorganizations), the interference coefficient provides an objective judgment logic based on data characteristics. During organizational changes, job responsibility boundaries are often temporarily blurred, and traditional methods are prone to drastic fluctuations in the ranking results. This invention utilizes the interference coefficient to accurately capture these changes in responsibility brought about by organizational restructuring. Specifically, when the difference in responsibility between a subset of jobs and its surrounding sequences narrows or the complexity increases, the interference coefficient dynamically responds and triggers an adaptive parameter adjustment mechanism. The system no longer relies on rigid static rules but dynamically introduces structural stability indicators such as organizational affiliation depth for correction based on the magnitude of the interference coefficient. This process is entirely based on a quantifiable mathematical model. Each ranking result corresponds to a clear interference coefficient threshold and weight calculation path, making the data processing transparent, quantifiable, and backtrackable. Therefore, even in scenarios with frequent organizational restructuring, the objective consistency and logical self-consistency of the job ranking results can still be guaranteed.

[0087] S3. Determine the distribution characteristic indicators of each job cluster to be classified in the target organizational unit in terms of job value assessment dimension and the affiliation and association indicators in the organizational hierarchy. Integrate the distribution characteristic indicators and affiliation and association indicators to obtain the influence weight of each job cluster to be classified on the accuracy of job classification in the enterprise.

[0088] During organizational restructuring, changes in job affiliations may occur due to business reorganization and departmental mergers, potentially blurring the boundaries of job responsibilities. Directly assessing the value of job clusters in this situation could lead to the incorrect categorization of positions with unclear responsibilities, impacting the internal fairness of the compensation system.

[0089] The stability of job responsibilities is manifested in their relative fixedness within the organizational structure. Within an organization, the more stable the hierarchical relationships of a cluster of jobs, the clearer their value hierarchy typically is. Therefore, the more dispersed the distribution of job clusters to be classified in terms of value assessment dimensions, or the more unstable their hierarchical relationships within the organization, the greater the impact of that job cluster on the accuracy of job classification within the company.

[0090] In one embodiment, based on the company's existing standard job grade system (such as a job grade and job level system), the standard grade benchmark value corresponding to each key job is determined, and the deviation between the score of the key job and the standard grade in each evaluation element dimension is calculated.

[0091] As a specific example, key positions that do not have a clear standard level correspondence will not be included in subsequent calculations and analyses.

[0092] S32, determine the value gap and grade span between each key position and its corresponding standard position grade, and construct a value assessment vector. Based on the value assessment vector and position code of each key position in each position cluster to be classified, determine the distribution characteristic indicators of the position value assessment dimension of each position cluster to be classified.

[0093] In one embodiment, each key position within the job cluster to be classified in the Zth organizational unit is mapped to the enterprise standard job grade system, and the value gap (vector of score difference of each element) and grade span (job grade difference) between each key position and the corresponding standard grade are determined, which are respectively used as the value gap and grade span.

[0094] If the value assessment vector of a key position exhibits a random or inconsistent distribution in the standard level and system, it means that the key position may be affected by unclear job definition or cross-functional positioning. Therefore, by analyzing the consistency of the value distribution of multiple key positions in the job cluster to be classified, the distribution characteristic indicators of the job cluster in the job value assessment dimension can be determined.

[0095] As an exemplary implementation, the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension are determined, including: The first step is to determine the value gap modulus, grade span angle, and code distance between each key position in each cluster of positions to be classified, based on the value assessment vector and position code of each key position in the cluster.

[0096] In one embodiment, for every two key positions in the job cluster to be classified, first determine the magnitude (Euclidean distance) between each value assessment vector, then determine the magnitude (Euclidean distance) between every two value assessment vectors, then determine the difference in magnitude between every two value assessment vectors, obtain the angle value (cosine similarity) between the directions of every two value assessment vectors, and calculate the coded distance (organizational hierarchy distance) between every two key positions based on the job code of every two key positions (such as the path of the job ID in the organizational tree).

[0097] The second step is to determine the cluster similarity between each pair of key positions based on the value gap modulus, the angle between the grade spans, and the coding distance. Then, the cluster similarity is used to cluster all the key positions in the position cluster to be classified, thus obtaining each value grade class.

[0098] In one embodiment, the value gap modulus, the grade span angle, and the coding distance are first standardized to unify the dimensions. Then, based on the standardized value gap modulus, the grade span angle, and the coding distance, the cluster similarity between each pair of key positions is calculated.

[0099] As an example, the formula for calculating the cluster similarity between any two key positions can be:

[0100] Always This represents the cluster similarity between two key positions. This represents the standardized value of the value gap modulus between two key positions. This represents the standardized value of the angle between the grade spans corresponding to two key positions. This represents the standardized value of the coded cluster corresponding to the two key positions. These are the weighting coefficients (each can be 1 / 3).

[0101] In one embodiment, based on the cluster similarity of every two key positions within the same job cluster to be classified, a hierarchical clustering algorithm is used to perform cluster analysis on all key positions within the job cluster to be classified, resulting in several value level classes. The implementation process of the hierarchical clustering algorithm is not limited here.

[0102] The third step is to determine the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension based on the angle between the grade spans of two key positions in each value grade category and the value gap modulus of each key position.

[0103] In a job value assessment system, the true value level usually has a consistent distribution characteristic, that is, the direction and magnitude of the value assessment vector of key positions within the same value level category are not significantly different. Based on this, it can be known that if the key positions within a certain value level category fluctuate greatly, it may be due to unclear job responsibilities or cross-functional positioning, resulting in poor value consistency.

[0104] As an exemplary implementation, the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension are determined, including: The first sub-step involves calculating the average value of the span angle of all levels in each value level class and the variance of the value gap modulus to obtain the value consistency index for each value level class. The level class with the value consistency index greater than the preset consistency threshold is used as the reference level class.

[0105] Here, the larger the average value of the span angle of all levels in the value level category, the more chaotic the value direction of key positions in the cluster is, the lower the consistency of value direction, and the smaller the value consistency index. The larger the variance of the value gap modulus of all key positions in the value level category, the greater the volatility of the value level of all key positions in the cluster, and the smaller the value consistency index. Therefore, the value consistency index is positively correlated with both the average value and the variance.

[0106] As an example, the formula for calculating the value consistency index for each value level category can be:

[0107] Always This represents the value consistency index for the Cth value level category. This represents an exponential function with the natural exponent e as the base. Used to perform normalization processing for negative correlation of data. This represents the average of all span angle values ​​in the Cth value level category. Let represent the variance of the modulus of all value gaps in the Cth value class.

[0108] In one embodiment, in order to filter out value grade classes with strong consistency in value direction and value level among the job clusters to be classified, so as to facilitate the subsequent calculation of distribution characteristic indicators, grade classes with value consistency indicators greater than a preset consistency threshold are used as reference grade classes. The preset consistency threshold can be an empirical value of 0.7.

[0109] The second sub-step involves determining the distribution characteristic indicators of the job value assessment dimension of the job clusters to be classified, based on the number of grade categories corresponding to the job clusters to be classified, the value consistency indicators of each reference grade category, and the number of key positions.

[0110] Here, the distribution characteristic index is negatively correlated with the number of grade categories, and positively correlated with the value consistency index and the number of key positions.

[0111] As an example, the formula for calculating the distribution characteristic index of the job cluster to be classified in the job value assessment dimension can be:

[0112] In the formula, This represents the distribution characteristic index of the m-th job cluster to be classified in the Z-th organizational unit on the dimension of job value assessment. This indicates the sequence number of the job cluster to be assigned to. Represents the normalization function. Let C represent the number of grade categories within the m-th job cluster to be classified, C represent the number of reference grade categories, and c represent the value grade category index. This represents the value consistency index of the c-th reference grade class. This represents the number of key positions in the c-th reference level category. This represents the total number of key positions in the m-th position cluster to be categorized.

[0113] In the formula for calculating the distribution characteristic index, The larger the value, the more inconsistencies there are in the key positions within the cluster of positions to be classified, and the more dispersed the value distribution among these key positions. This has a greater impact on the accuracy of position classification. The larger the value consistency index, the greater its credibility, and the more obvious the distribution characteristics of the job clusters to be classified in the job value assessment dimension.

[0114] As a specific implementation, the individual job clusters to be assigned to the lowest organizational level in the organizational structure may not participate in the subsequent affiliation analysis process.

[0115] It should be noted that, within the same organizational level, job clusters awaiting categorization may unexpectedly be classified using similar job codes or value assessment models, leading to misclassification as having consistent value. However, in multi-level hierarchical structures, positions with unclear responsibilities often exhibit unstable hierarchical relationships, resulting in ambiguous reporting relationships or dual management. Therefore, further analysis of the hierarchical relationships of job clusters awaiting categorization within the organizational hierarchy is necessary to assess their impact on the accuracy of job categorization within the enterprise, corresponding to subsequent steps S33 to S34.

[0116] S33, perform a graph analysis of the hierarchical relationships of each key position in the organizational hierarchy to obtain the organizational affiliation depth index for each key position in each cluster of positions to be classified.

[0117] Here, the higher the organizational affiliation depth index of a key position, the deeper the key position is in the organizational structure, the clearer the affiliation, and the smaller the impact on the accuracy of position classification.

[0118] In one embodiment, for each key position within each job cluster to be assigned to in the target organizational unit, the reporting paths between superiors and subordinates of each key position are recorded in the organizational hierarchy diagram, and the organizational affiliation chain of each key position is constructed. If a key position has multiple reporting relationships in a certain organizational level, the deepest path is taken, and the depth index of each key position in the organizational tree is calculated from the organizational affiliation chain of each key position to obtain the organizational affiliation depth index corresponding to each key position in each job cluster to be assigned to.

[0119] It should be noted that the organizational hierarchy analyzed in this embodiment refers to the complete reporting chain from the target organizational unit and its superior organizational units up to the highest level of the enterprise.

[0120] S34. Based on the organizational affiliation depth index and the proportion of positions in each job cluster to be assigned to in the organizational hierarchy, determine the affiliation association index of each job cluster to be assigned to in the organizational hierarchy.

[0121] As an exemplary implementation, based on the organizational affiliation depth index and the proportion of positions in each job cluster to be assigned to in the organizational hierarchy, the affiliation association index of each job cluster to be assigned to in the organizational hierarchy is determined, including: The first step is to obtain the number of positions in each job cluster to be assigned to in the organizational hierarchy, and use the second ratio of the number of positions in each job cluster to the total number of positions in the target organizational unit as the hierarchy weight.

[0122] The higher the hierarchical weight, the greater the proportion of the job cluster to be classified in its organizational unit, and the higher the credibility of the organizational affiliation of the job cluster. This facilitates subsequent weighted summation analysis of the organizational affiliation depth index based on the hierarchical weight.

[0123] The second step is to perform a weighted summation of the organizational affiliation depth index based on the hierarchical weight of the job clusters to be classified in each organizational level, so as to obtain the degree of influence of each organizational level on the accuracy of job classification.

[0124] As an example, the formula for calculating the impact of the Zth organizational unit on the accuracy of job classification can be:

[0125] In the formula, This indicates the degree of influence of the Zth organizational unit on the accuracy of job classification. Represents a linear normalization function. This represents the number of job clusters to be assigned in the Zth organizational unit. This indicates the sequence number of the job cluster to be assigned to. This represents the organizational affiliation depth index (normalized) of the m-th job cluster to be assigned within the Z-th organizational unit. This represents the percentage of positions in the m-th job cluster to be assigned within the Z-th organizational unit (the second ratio), i.e., the hierarchical weight.

[0126] By referring to the influence of the z-th organizational unit on the accuracy of job classification, we can obtain the influence of each organizational unit on the accuracy of job classification.

[0127] The third step involves determining the organizational affiliation index for each key position based on the degree of influence of each organizational level on the accuracy of position classification, the ratio of the organizational affiliation depth index to the total organizational levels, and so on.

[0128] Here, the organizational affiliation indicator is negatively correlated with the degree of influence, and positively correlated with the third ratio.

[0129] As an example, the formula for calculating the organizational affiliation index for each key position can be:

[0130] In the formula, This represents the organizational affiliation indicator for the k-th key position. Represents the normalization function. This represents the organizational affiliation depth index of the k-th key position. Indicates the total number of levels in the company's organizational structure. This represents the third ratio. This represents the cumulative value indicating the degree of influence of all organizational levels on the accuracy of job classification. This represents a non-zero constant.

[0131] In the formula for calculating the organizational affiliation index, the organizational affiliation depth index may differ for different key positions, and the organizational levels may also differ; therefore, the cumulative value of the influence will vary. The results may differ. The larger the cumulative value of the impact, the more numerous and larger the proportion of the organizational hierarchy of the key position, and the more complex the organizational affiliation. This may reduce the affiliation stability and reliability of the key position, and the lower the credibility of the third ratio.

[0132] The fourth step is to conduct a comprehensive analysis of the organizational affiliation indicators of all key positions within the same job cluster to be assigned to in the target organizational unit, and obtain the affiliation association indicators of each job cluster to be assigned to in the organizational hierarchy.

[0133] In one embodiment, the average organizational affiliation index of all key positions within the same job cluster to be assigned is calculated, and the average value of the organizational affiliation index of the job cluster to be assigned is used as the affiliation association index of the job cluster to be assigned in the organizational hierarchy.

[0134] As an exemplary implementation, by integrating distribution characteristic indicators and membership association indicators, the influence weights of each job cluster to be classified on the accuracy of enterprise job classification are obtained, including: Here, both the distribution characteristic index and the affiliation correlation index are dimensionless. The larger the distribution characteristic index and the affiliation correlation index, the more stable the affiliation relationship of the key positions within the job cluster to be classified in the organizational hierarchy, the more concentrated the value assessment dimension distribution, the clearer the job value characteristics, and the more likely they are to be classified accurately. The smaller the weight of the job cluster to be classified on the accuracy of the company's job classification, the smaller the degree of adjustment of the classification parameters of the job cluster to be classified.

[0135] In one embodiment, for each job cluster to be classified, the product of the distribution characteristic index and the affiliation correlation index of the job cluster to be classified is calculated. This can be done by using... The function performs negative correlation normalization on the product to obtain the influence weight of the job cluster to be classified on the accuracy of the enterprise job classification.

[0136] As an example, the formula for calculating the influence weight can be:

[0137] In the formula, This represents the weight of the influence of the m-th job cluster to be classified in the z-th organizational unit on the accuracy of the enterprise's job classification. Indicators representing distribution characteristics Indicates the subordinate related indicators.

[0138] Thus, this embodiment has obtained the influence weights of each job cluster to be classified in the target organizational unit on the accuracy of the enterprise's job classification.

[0139] S4 uses influence weights to adaptively adjust the grading parameters of the corresponding job clusters to be graded in the target organizational unit, and obtains the optimized job grading parameters. Then, it obtains the job grading results of each organizational unit after priority, so as to carry out intelligent matching and salary mapping of enterprise job value.

[0140] As an exemplary implementation, the ranking parameters of the corresponding job clusters to be ranked in the target organizational unit are adaptively adjusted using influence weights to obtain optimized job ranking parameters, including: The first step is to determine the classification correction coefficient for each job cluster to be classified in the target organizational unit based on the weight of the impact of each job cluster to be classified on the accuracy of the enterprise's job classification.

[0141] In one embodiment, the classification correction coefficient for each job within each job cluster to be classified in the target organizational unit is set to the influence weight of the corresponding job cluster. The classification correction coefficient for jobs outside the job clusters to be classified in the target organizational unit is set to 1 (i.e., no adjustment). This yields the classification correction coefficient for each job in the target organizational unit. Furthermore, the classification correction coefficient is the same for each job within the same job cluster to be classified.

[0142] The second step is to adaptively adjust the scores of the job value assessment elements based on the classification correction coefficient for each job, thereby obtaining the optimized job classification parameters.

[0143] As an example, the formula for calculating the optimized job classification parameters can be:

[0144] In the formula, This represents the optimized job classification parameters (final element score). This indicates the initial job value assessment element score. Indicates the influence weight. This is an adjustment factor (an empirical value of 0.2 is acceptable).

[0145] As an exemplary implementation, the optimized job hierarchy structure for each organizational unit is obtained to perform intelligent matching of enterprise job values ​​and salary mapping, including: The first step is to obtain the optimized job classification parameters for each organizational unit by referring to the process of obtaining the optimized job classification parameters for the target organizational unit.

[0146] The second step is to intelligently match the value of enterprise positions and map salaries based on the optimized job classification parameters.

[0147] As an exemplary implementation, intelligent matching and salary mapping of enterprise job values ​​are performed based on the optimized job classification parameters, including: The first sub-step involves calculating the total value score for each job based on the optimized job classification parameters (scores for each evaluation element), and determining the corresponding job grade and level for each job by referring to the company's standard job grade system.

[0148] Obtain the optimized comprehensive value score of the i-th position, denoted as . After matching the job grade with the company's standard job grade comparison table, the corresponding grade and level of the job are determined. The expression for the job classification result can be:

[0149]

[0150] In the formula, This represents the job grade of the i-th position. This represents the job level of the i-th position. and This represents the mapping function. This represents the influence coefficient of the department where the i-th position is located on the determination of the job level.

[0151] The second sub-step involves calculating the salary bandwidth and median for each position based on the job classification results and the company's salary system standards. The expression for this can be:

[0152]

[0153]

[0154] In the formula, This represents the minimum salary for the i-th position. This represents the maximum salary for the i-th position. This represents the median salary for the i-th position. Indicates the base salary. This represents the coefficient of influence of job grade on salary. This represents the salary bandwidth coefficient.

[0155] The third sub-step involves establishing a job-salary mapping matrix and dynamically adjusting the salary positioning of specific employees based on their performance levels and competency assessment results to determine the final salary level.

[0156] In a more preferred embodiment, the adaptive adjustment of the classification parameters is a closed-loop optimization process based on online learning feedback. The system initializes a baseline classification model (such as a regression model based on factor scores). After the system generates classification results for a group of positions, they are not directly applied to compensation, but instead enter a human-machine collaborative verification stage, where HR experts confirm or correct some classification results (especially high-impact weighted job clusters) in the system.

[0157] The system collects expert correction data in real time and correlates it with the system's initial classification parameters and calculated influence weights. This data constitutes a real-time feedback stream. The system includes an incremental learning module that periodically (e.g., weekly) uses the latest feedback data to fine-tune the parameters of the aforementioned interference coefficient evaluation network and the baseline classification model. Specifically, the direction and magnitude of expert corrections are transformed into a loss function for online model updates.

[0158] This mechanism enables the system to continuously adapt to dynamic changes in the company's organizational structure and responsibilities, as well as subtle adjustments in managers' perceptions of job value. It solves the problem of traditional evaluation systems becoming rigid once deployed and unable to evolve with the times, thus achieving the system's self-evolution capability. The entire process constitutes a complete technological closed loop of perception, decision-making, feedback, and optimization.

[0159] It should be noted that dynamically adjusting salary positioning can smooth and optimize salary data, eliminate unfairness caused by local errors in the job evaluation or grading process, and effectively extract the true value level of the job, rather than local evaluation biases, thereby improving the internal fairness and external competitiveness of the company's compensation. Thus, this embodiment completes the intelligent matching and salary mapping of corporate job values.

[0160] To further verify the effectiveness of this invention in large and complex organizational structures, this embodiment takes an ultra-high voltage substation maintenance center under a provincial power grid company as an example, and performs intelligent classification and salary matching for its key positions.

[0161] S1. Data Acquisition and Vectorization. The system acquires the organizational structure data of the ultra-high voltage substation maintenance center from the power grid company's Human Resources Management System (HRMS). This target organizational unit contains two main job subsets: Subset A (Primary Maintenance Group), mainly responsible for the physical maintenance of high-voltage equipment such as transformers and circuit breakers, including job A1 (Senior Primary Maintenance Technician); Subset B (Secondary Maintenance Group), mainly responsible for the debugging of relay protection and automation control systems, including job B1 (Relay Protection Technician). Based on the power grid job descriptions, Natural Language Processing (NLP) is used to parse the data and construct a 5-dimensional attribute feature vector V={Ssafe,Svolt,Senv,Sskill,Sresp} for the power industry. The meanings and quantitative scores of each dimension are as follows: Ssafe represents (safety risk level): 1-5 points, with 5 points representing extremely high risks such as live-line work on ultra-high voltage. Svolt (voltage level): 1-5 points, with 5 points representing 1000kV ultra-high voltage and 1 point representing 10kV distribution network. Senv (Working Environment): 1-5 points, with 5 points representing high-altitude, outdoor, and strong electromagnetic environments. Sskill (Skill Complexity): 1-10 points, representing the technical threshold. Sresp (Emergency Responsibility): 1-10 points, representing decision-making responsibility during power grid failures. The processed job vector data for position A1 (Senior Technician in Primary Maintenance) is {5, 5, 5, 8, 7}; its characteristics are high risk, ultra-high voltage, outdoor high-altitude work, and high skill requirements. The processed job vector data for position B1 (Relay Protection Engineer) is {3, 5, 2, 9, 9}; its characteristics are medium risk, ultra-high voltage, indoor cabinet work, extremely high logical skills, and significant emergency decision-making responsibility.

[0162] S2. Calculation of Duty Differences and Interference Coefficients. The system calculates the degree of duty differences between job subset A and job subset B to determine whether they belong to the same job sequence. The Euclidean distance (feature difference value) Dist(A1,B1) between job A1 and job B1 is approximately 4.54. Although both A1 and B1 belong to the same maintenance center and have the same voltage level (Svolt 5), the difference in Senv (working environment: outdoor high-altitude vs. indoor) and Ssafe (safety risk) results in a feature difference value of 4.24. Therefore, the system determines that subset A and subset B belong to different job sequences (primary sequence vs. secondary sequence) and should be classified separately to avoid confusing the evaluation of "high-altitude physical skills" with "logical and mental skills".

[0163] Suppose that within a subset A of job positions, there exists a job (position A2, with the vector {1,1,1,2,2}) where the attendance clerk at a maintenance center has the position code A2. Calculations show that the minimum difference between A2 and the original job A1 is extremely large, causing the responsibilities of subset A to become significantly more complex. The interference coefficient calculated by the system will exceed a preset threshold (e.g., 0.7), thus marking subset A as a "job cluster to be reclassified," indicating that management support positions need to be separated.

[0164] S3. Distribution Characteristics and Affiliation Relationship Calculation. For the separated maintenance group (subset A), the system analyzes its performance within the organizational hierarchy. Position A1 scores extremely high (both 5) in the safety risk and work environment dimensions, exhibiting a clear "high-risk, high-compensation" distribution characteristic in the job value assessment dimension. The system calculates its high distribution characteristic index, indicating that the value characteristics of this job cluster are clear, mainly reflecting the job value corresponding to the intensity and risk of power grid work. Position A1's job code is "CG-bdzx-check01," located at the 4th level of the organizational structure, with a clear reporting path (team member -> team leader -> center director -> company vice president). The system confirms through graph analysis that its organizational affiliation depth index is 4 (out of 5), and there are no cross-departmental dotted reporting relationships. The calculated affiliation relationship index is high, indicating that this position has a stable position within the organization.

[0165] S4. Adaptive Adjustment of Classification Parameters and Salary Mapping. Traditional general evaluation models often overemphasize Skill (skill complexity), leading to higher scores for office-based secondary engineers compared to primary technicians who climb power towers. However, this invention, by calculating influence weights, identifies subset A (primary maintenance) as having highly distributed indicators (emphasizing safety and environmental value) and high membership correlation indicators. The system then performs adaptive adjustment, automatically increasing the weight coefficients of the "safety risk" and "work environment" dimensions while decreasing the weight of purely theoretical skills. Next, the system performs classification results, significantly improving the overall score for position A1 (senior primary maintenance technician), correctly classifying it as a "P8 expert position" (the original general model might have classified it as P7). Finally, the system performs salary mapping, matching the P8 job level with the power grid company's salary broadband that "leans towards production lines," determining that its median salary is higher than that of administrative personnel at the same level.

[0166] This invention identifies various job clusters to be classified within an organizational unit, analyzes the influence weights of different job clusters on the accuracy of job classification, and adaptively adjusts job classification parameters accordingly. This improves the accuracy and objectivity of job value assessment, ensuring the internal fairness and external competitiveness of the company's compensation system.

[0167] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A system for intelligent import and matching data consistency processing of job classification under a variable enterprise architecture, characterized in that: include: The data acquisition module is used to acquire the enterprise's organizational structure data and job description information to be processed; The job cluster determination module is used to take each organizational unit in the enterprise organizational structure data as the target organizational unit, and determine each job cluster to be assigned to the target organizational unit based on the attribute feature vector of each job in the target organizational unit and the enterprise architecture meta-model. The weight calculation module is used to determine the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension and the affiliation and association indicators in the organizational hierarchy, and to merge the distribution characteristic indicators and the affiliation and association indicators to obtain the influence weight. The adaptive grading module uses the influence weights to adaptively adjust the grading parameters and obtain optimized job grading results.

2. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 1, characterized in that, The job cluster determination module determines each job cluster to be assigned to the target organizational unit, including: The target organizational unit is divided into several job subsets; based on the attribute feature vector of each job in each job subset, the degree of difference in responsibilities between each two job subsets is analyzed, and each job subset corresponding to each job sequence is determined. Determine each benchmark job and its matching degree in each job subset. The matching degree is the minimum value among all attribute feature differences between two job subsets of the same job sequence. The benchmark jobs are the two jobs corresponding to the minimum value. Based on the benchmark positions and their matching degree of each position subset, the responsibility complexity of each position subset is determined, and based on the score of each position in each position subset on the position value assessment elements, the average element score of each position subset is determined. Based on the responsibility complexity and the average score of each element in each job subset, the interference coefficient of each job subset on the accuracy of job classification is determined. The job subsets with interference coefficients not less than a preset interference threshold are taken as job clusters to be classified.

3. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 2, characterized in that, The step involves analyzing the degree of responsibility difference between any two job subsets based on the attribute feature vector of each job within each job subset, and determining the job subsets corresponding to each job sequence, including: Calculate the difference between the attribute feature vector of the target job in each job subset and the attribute feature vector of each job in the second job subset, and record it as the feature difference value. Select the minimum feature difference value. The average of the minimum feature difference values ​​corresponding to all positions in the first position subset is taken as the degree of difference in responsibilities between the first position subset and the second position subset. The first position subset and the second position subset are two different position subsets in the target organizational unit, and the target position is any position in the first position subset. Based on the degree of difference in responsibilities between each pair of job subsets, all job subsets in the target organizational unit are clustered to obtain each job cluster, and each job subset in the same job cluster is identified as the job subset corresponding to the same job sequence.

4. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 3, characterized in that, The step of determining the responsibility complexity of each job subset based on the baseline jobs and their matching degrees for each job subset includes: For each job subset, the benchmark jobs that appear repeatedly in the benchmark job set corresponding to the job subset are identified as key jobs. The number of benchmark jobs corresponding to each key job, as well as the number of benchmark jobs and key jobs within the job subset, are counted. The responsibility complexity of a job subset is determined based on the number of benchmark jobs and minimum matching degree corresponding to each key job, and the first ratio of the number of key jobs to the number of benchmark jobs in the job subset. The responsibility complexity is negatively correlated with the number of benchmark jobs and the first ratio, and positively correlated with the minimum matching degree.

5. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 4, characterized in that, The step of determining the interference coefficient of each job subset on the accuracy of job classification based on the responsibility complexity and the average score of each element for each job subset includes: For each job subset, obtain the degree of responsibility difference between every two job subsets corresponding to the job sequence to which the job subset belongs, and select the minimum degree of responsibility difference; A fusion analysis is performed on the responsibility complexity, the mean score of the elements, and the degree of difference in the minimum responsibility of the job subset to determine the interference coefficient of the job subset on the accuracy of job classification. The interference coefficient is positively correlated with the complexity of the duties, and negatively correlated with the mean score of the elements and the degree of difference in the minimum duties.

6. The intelligent import and matching data consistency processing system for job classification based on a variable enterprise architecture as described in claim 4 is characterized in that, The weight calculation module determines the distribution characteristic indicators of each job cluster to be classified in the job value assessment dimension and the affiliation and association indicators in the organizational hierarchy, including: Obtain each key position in each job cluster to be classified in the target organizational unit, and then determine the deviation vector of each key position relative to the enterprise standard job level in terms of job value assessment elements. Determine the value gap and grade span between each key position and its corresponding standard position level to form a value assessment vector. Based on the value assessment vector and position code of each key position in each position cluster to be classified, determine the distribution characteristic indicators of each position cluster to be classified in the position value assessment dimension. A graph analysis is performed on the hierarchical relationships of each key position in the organizational hierarchy to obtain the organizational affiliation depth index for each key position in each position cluster to be assigned to. Based on the organizational affiliation depth index and the proportion of positions in each position cluster to be assigned to in the organizational hierarchy, the affiliation relationship index of each position cluster to be assigned to in the organizational hierarchy is determined.

7. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 6, characterized in that, The step of determining the distribution characteristic indicators of each job cluster in terms of job value assessment dimension based on the value assessment vector and job code of each key job in each job cluster to be classified includes: For each job cluster to be classified, the value gap modulus, grade span angle, and code distance between each two key jobs are determined based on the value assessment vector and job code of each key job in the job cluster. The cluster similarity between each pair of key positions is determined based on the value gap modulus, the grade span angle, and the coding distance. The cluster similarity is then used to perform hierarchical clustering of all key positions in the position cluster to be classified, resulting in each value grade class. Based on the angle between the grade spans of every two key positions in each value grade category and the value gap modulus of each key position, the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension are determined.

8. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 7, characterized in that, The method of determining the distribution characteristic indicators of the job clusters to be classified in the job value assessment dimension based on the angle between the grade spans of every two key positions in each value grade category and the value gap modulus of each key position includes: Calculate the average value of the span angle of all levels in each value level class and the variance of the value gap modulus to obtain the value consistency index of each value level class. The level class with the value consistency index greater than the preset consistency threshold is used as the reference level class. The value consistency index is negatively correlated with the average value and the variance. Based on the number of grade categories corresponding to the job clusters to be classified, the value consistency index of each reference grade category, and the number of key positions, the distribution characteristic index of the job clusters to be classified in the job value assessment dimension is determined. The distribution characteristic index is negatively correlated with the number of grade categories and positively correlated with both the value consistency index and the number of key positions.

9. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 6, characterized in that: The process of determining the affiliation index of each job cluster to be assigned in the organizational hierarchy based on the organizational affiliation depth index and the proportion of jobs in each job cluster to be assigned in the organizational hierarchy includes: Obtain the number of positions in each job cluster to be assigned in the organizational hierarchy, and use the second ratio of the number of positions in the job cluster to the total number of positions in the target organizational unit as the hierarchy weight. The organizational affiliation depth index is weighted and summed based on the hierarchical weight of the job clusters to be classified in each organizational level to obtain the degree of influence of each organizational level on the accuracy of job classification. For each key position, the organizational affiliation index is determined based on the degree of influence of each organizational level on the accuracy of position classification and the third ratio of the organizational affiliation depth index to the total organizational levels. The organizational affiliation index is negatively correlated with the degree of influence and positively correlated with the third ratio. A comprehensive analysis of the organizational affiliation indicators of all key positions within the same job cluster to be assigned to in the target organizational unit is conducted to obtain the affiliation association indicators of each job cluster to be assigned to in the organizational hierarchy.

10. The intelligent import and matching data consistency processing system for job classification under a variable enterprise architecture as described in claim 1, characterized in that, The weight calculation module integrates the distribution characteristic index and the membership association index to obtain the influence weight of each job cluster to be classified on the accuracy of enterprise job classification, including: For each job cluster to be classified, calculate the product of the distribution characteristic index and the membership association index of the job cluster to be classified. The product is normalized by applying a decay function to obtain the influence weight of the job cluster to be classified on the accuracy of enterprise classification. Then, the classification parameters are adaptively adjusted by the adaptive classification module to obtain the optimized job classification result.

Images