A method for constructing a human resource management system

Abstract

The application provides a construction method of a human resource management system, belongs to the technical field of intelligent management, and determines a person's explicit and implicit capability vector by calling third-party and self-assessment files; real-time monitoring of attendance information and operation behavior sets extracts a comprehensive fuzzy capability vector and a cross-domain coverage subset; based on the attendance authenticity correction capability vector, the fuzzy elements are corrected by clear coefficient grading; clustering analysis of the final vectors in each field and cross-domain integration realize the ordered management of human resources and the timed updating of capability. The method solves the problems of unbalanced capability evaluation, dynamic updating lag, insufficient cross-domain adaptation and other problems in traditional systems, improves the refinement of human resource management and operational efficiency, and guarantees the quality of medical services.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology, and in particular to a method for constructing a human resource management system. Background Technology

[0002] In the field of intelligent management technology within the healthcare industry, hospital human resource management systems, as core support tools for optimizing the allocation of medical and related personnel and ensuring the quality and efficiency of medical services, directly impact hospital operational efficiency and the level of medical services. Currently, there is an imbalance in the dimensions of competency assessment for medical and nursing staff in hospital human resource management. Traditional methods often focus on explicit quantitative indicators such as diagnostic and treatment procedures and nursing execution, while neglecting implicit key competencies such as doctor-patient communication skills, emergency decision-making abilities, and multidisciplinary collaboration logic. This leads to a disconnect between competency assessments and actual clinical needs. Furthermore, the competencies of medical and nursing staff change rapidly, and the impact of work behaviors in different areas such as clinical diagnosis and treatment, scientific research and teaching, and administrative logistics on competency is real-time. However, existing systems struggle to capture these dynamic changes and update competency assessment results in real time, resulting in lagging competency data. Moreover, various institutional areas within hospitals (such as clinical medicine, nursing, and medical technology) also contribute to this imbalance. There is insufficient potential exploration among (scientific research and teaching, administration and logistics). For example, the case analysis skills of clinicians can be transferred to the scientific research field, and the communication skills of nurses can be adapted to the outpatient service field. The existing system cannot realize cross-domain ability adaptation analysis, which is not conducive to the flexible allocation of human resources. Moreover, there are many vague elements in the assessment of medical staff's ability, such as qualitative descriptions of good collaboration ability and proficiency in emergency response. There is a lack of systematic and clear processing methods, which affects the objectivity and reliability of the assessment results. In summary, the existence of these problems makes it difficult for the existing hospital human resource management system to meet the medical industry's needs for refined, dynamic and professional human resource management.

[0003] Therefore, the present invention provides a method for constructing a human resource management system. Summary of the Invention

[0004] This invention provides a method for constructing a human resource management system to solve the aforementioned technical problems.

[0005] This invention provides a method for constructing a human resource management system, comprising: Step 1: Retrieve each person's third-party assessment documents and self-assessment documents in different institutional areas from the stored database and conduct a comprehensive analysis to determine the explicit and implicit capability vectors of the person in the corresponding institutional areas; Step 2: Monitor the attendance information of each person in different system areas in real time and capture the set of operational behaviors of each person in different system areas for each implementation project. The attendance information includes remote check-in information, and the set of operational behaviors includes the actual behavior of each assessment standard set under the corresponding implementation project. Step 3: Analyze the corresponding explicit capability vector and implicit capability vector based on the set of operational behaviors of the personnel in the same institutional domain, extract the comprehensive fuzzy capability vector, and determine the cross-domain coverage subset of each fuzzy element in the comprehensive fuzzy capability vector; Step 4: Based on the authenticity of attendance information within the same system domain, perform a first correction on the first relevant element in the explicit ability vector and the second relevant element in the implicit ability vector of the personnel. Then, based on the cross-domain coverage subset of the same personnel within the corresponding system domain and the first-corrected ability vector, determine the clarity coefficient of the corresponding fuzzy element. If the clarity coefficient is greater than the preset coefficient, the corresponding fuzzy element is clarified according to the clarity coefficient and a second correction is performed. Otherwise, based on the difference between the preset coefficient and the clarity coefficient, determine the fuzzy probability of the corresponding fuzzy element, divide the range from 0 to 1 according to the fuzzy probability, and randomly sample according to rand(0,1) to determine the current state, and perform a third correction on the corresponding fuzzy element. Step 5: Based on the correction results of the fuzzy elements, obtain the current final vector of the personnel in the corresponding system domain, and perform cluster analysis on all current final vectors in different system domains to obtain the first sub-cluster map, and perform cluster analysis on all current final vectors in the same system domain to obtain the second sub-cluster map. Then, integrate all sub-cluster maps across domains for orderly human resource management, and update the personnel capabilities of the human resource management system regularly.

[0006] Preferably, it includes: dividing the range from 0 to 1, including: The range from 0 to the fuzzy probability is considered the fuzzy range, and the range from the fuzzy probability to 1 is considered the clear range; When a value is randomly selected based on rand(0,1), if the value falls within the fuzzy range, the current state is determined to be a fuzzy state; otherwise, the current state is determined to be a clear state.

[0007] Preferably, the explicit ability vector and implicit ability vector of the corresponding personnel are determined, including: Extract the first initial evaluation vector of the personnel in the same institutional field based on the third-party assessment and evaluation document and the second initial evaluation vector based on their own assessment and evaluation document; Keyword searches were conducted on each assessment document to identify the conventional and novel keywords for each assessment indicator, and the overall impact factor of the generation logic of each novel keyword on all conventional keywords of the corresponding assessment indicator was analyzed. When the comprehensive impact factor is greater than the preset impact factor, the corresponding novel keyword is regarded as the first keyword; otherwise, the corresponding novel keyword is removed. Determine the mapping relationship between the first keyword and each regular associated keyword, determine the derivation process according to the mapping relationship, and merge all derivation processes according to the derivation similarity of each derivation process to obtain the derivation class; The explicit and implicit evaluation elements of the first and second initial evaluation vectors are extracted respectively. The extracted explicit and implicit evaluation elements are then optimized according to the derived class to obtain a three-dimensional array for each evaluation element, which includes ability type, ability confidence and preference intensity. The explicit and implicit ability vectors of the corresponding personnel are then obtained.

[0008] Preferably, after real-time online monitoring of each person's attendance information in different policy areas, the process includes: analyzing the authenticity of the attendance information, specifically including: Collect multi-source data corresponding to the attendance information, including real-time operational data of the location associated area of ​​the attendance device, historical abnormal records of the device, and historical attendance behavior sequence of the personnel. The real-time operational data includes the frequency of service complaints and the density of people in the corresponding area, and the historical attendance behavior sequence of the personnel includes the distribution characteristics of attendance time and the behavior pattern of cross-device check-in. The initial trust weight is determined based on the stability of the personnel's historical attendance behavior sequence. The trustworthiness of the device is then adjusted by combining the service complaint frequency and traffic density data of the area associated with the location of the attendance device, and the dynamic verification threshold for this attendance is obtained. The collected attendance information is matched with the personnel's historical attendance behavior sequence. If the deviation between the check-in time, device location and historical behavior pattern exceeds the dynamic verification threshold, it is marked as a suspected anomaly. At this time, the matching degree between the real-time operation data of the area where the attendance device is located and the attendance is verified. If the matching degree between the population density and the device check-in frequency in the area during the attendance period is lower than the preset value, the attendance information is back-analyzed in combination with the corresponding department's operation demand threshold. If it is determined to be false, the degree of impact on the department's operation value is determined. If the degree of impact exceeds the preset risk threshold, the manual review process is triggered. Otherwise, the authenticity is directly determined based on the dynamic verification threshold.

[0009] Preferably, the extracted comprehensive fuzzy capability vector includes: Each actual behavior of the personnel in the same system domain is compared and analyzed with the standard behavior of the assessment standard corresponding to the ability element in the ability vector according to the implementation project. The difference behavior of the corresponding ability element is obtained, and the negative and positive promoting factors of the difference behavior are analyzed. The negative and positive promoting factors are fused to obtain the main promoting direction, and the main promoting direction is then used as the basis for further analysis. The standard behaviors defined by the assessment criteria of the corresponding ability elements in the explicit and implicit ability vectors are matched in a scenario-based manner. Based on the matching results, the deviation between the actual behavior and the standard behavior is extracted as the difference behavior. Among them, the difference behavior corresponding to the explicit ability elements focuses on the execution deviation that can be quantified, while the difference behavior corresponding to the implicit ability elements focuses on the decision deviation of the implicit behavior of collaboration logic and risk prediction. The negative and positive contributing factors for each differential behavior are identified separately. Negative contributing factors include triggers that lead to deterioration of performance due to operational errors, process delays, or collaboration conflicts. Positive contributing factors include triggers that drive performance iteration through innovation optimization, efficiency improvement, or risk avoidance. At the same time, the impact depth of each contributing factor is quantified by combining the complexity coefficient of the implemented project. Calculate the entropy value of the influence of negative promoting factors. Entropy value of the influence of positive promoting factors The magnitude of the entropy value is determined by the depth of influence of the factor, the frequency of occurrence, and the project complexity coefficient. when When the main promoting direction is determined to be the positive iteration direction, then, The preset impact difference threshold; when When the main promoting direction is determined to be the negative deterioration direction; Otherwise, the main promoting direction is determined to be a neutral and stable direction; Based on the main facilitating direction, the explicit and implicit capability vectors are fuzzified and adjusted, including: If the main promoting direction is the positive iteration direction, the fuzzy membership degree of the corresponding capability element is improved, and the iteration gain coefficient is introduced to expand the fuzzy coverage of the positive influence. If the main promoting direction is the negative deterioration direction, the fuzzy membership degree of the corresponding capability element is reduced, and a deterioration attenuation coefficient is introduced to reduce the fuzzy coverage of the negative influence. If the main promoting direction is a neutral and stable direction, maintain the fuzzy membership degree of the corresponding capability element, and introduce a stability correction coefficient to maintain the dynamic balance of the fuzzy coverage range. By integrating the fuzzification adjustment results of explicit and implicit capability vectors, a comprehensive fuzzy capability vector of the personnel in the corresponding institutional domain is obtained.

[0010] Preferably, determining the cross-domain coverage subset for each fuzzy element in the integrated fuzzy capability vector includes: Extract the behavior transfer feature vector of each fuzzy element in the comprehensive fuzzy capability vector and quantize it into a multi-dimensional numerical sequence. The behavior transfer feature vector includes the main promoting direction label of the corresponding fuzzy element, the proportion of negative promoting factors affecting entropy value, the proportion of positive promoting factors affecting entropy value, and the fuzzy membership degree. By integrating capability elements from all institutional domains, a cross-domain capability adaptability map is constructed. Each node represents a capability element of the target domain, and the node attributes include the adaptability coefficients of the corresponding institutional domain to the three main promoting directions: positive iteration, negative degradation, and neutral stability. The behavior transfer feature vector of the fuzzy element is matched with the attribute vector of the target domain capability element in the cross-domain capability adaptability map using cosine similarity. Combined with the adaptation coefficient of the main facilitation direction, the cross-domain coverage confidence is calculated. ,in, Let i be the cross-domain coverage confidence of fuzzy element i to target domain capability element j; For behavioral transfer feature vectors With the target ability element attribute vector Cosine similarity; target area The main promoting direction of fuzzy element i The adaptation coefficient; Filter all confidence levels The target domain capability elements are used to form a cross-domain coverage subset of the corresponding fuzzy elements, and each element in the subset is labeled with a facilitating direction adaptation tag. To cover the confidence threshold.

[0011] Preferably, determining the sharpness coefficient of the corresponding blurred element includes: ,in, This represents the sharpness coefficient for the corresponding blurred element; The average coverage confidence of the cross-domain coverage subset; The corrected stability of the first modified capability vector; The adaptation coefficient for the main promotion direction; , , These are the dimension weight coefficients, and ; The interval between the generation time and the current time for cross-domain coverage subsets; This is the decay rate coefficient; Weighting of contribution to the domain.

[0012] Preferably, the final vector of the person in the corresponding institutional domain is obtained based on the correction result of the fuzzy elements, including: The second revised sharpening fuzzy element and the third revised stateful fuzzy element are aligned with the first revised explicit capability vector and implicit capability vector according to the capability element dimension, and the fusion weight is assigned in combination with the main promoting direction corresponding to each fuzzy element. The average coverage confidence of the cross-domain coverage subset of fuzzy elements is used as a calibration factor to calibrate the fused vector. The calibrated vector is then normalized to generate the current final vector containing capability strength, clarity, and cross-domain adaptation labels.

[0013] Compared with the prior art, the beneficial effects of this application are as follows: By collecting multi-dimensional data in hospital scenarios, accurately constructing explicit and implicit capability vectors, dynamically correcting attendance based on authenticity, clarifying fuzzy elements, and integrating cross-domain clustering, the hospital has achieved comprehensiveness, dynamism, and accuracy in human resource capability assessment. This effectively addresses pain points such as complex hospital attendance scenarios, imbalanced capability assessment dimensions, and difficulties in cross-domain human resource allocation. It can accurately match human resource needs across various systems, optimize job matching and training plan development, improve the refinement and operational efficiency of hospital human resource management, and ensure the quality of medical services and patient safety.

[0014] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating a method for constructing a human resource management system according to an embodiment of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] This invention provides a method for constructing a human resource management system, such as... Figure 1 As shown, it includes: Step 1: Retrieve each person's third-party assessment documents and self-assessment documents in different institutional areas from the stored database and conduct a comprehensive analysis to determine the explicit and implicit capability vectors of the person in the corresponding institutional areas; Step 2: Monitor the attendance information of each person in different system areas in real time and capture the set of operational behaviors of each person in different system areas for each implementation project. The attendance information includes remote check-in information, and the set of operational behaviors includes the actual behavior of each assessment standard set under the corresponding implementation project. Step 3: Analyze the corresponding explicit capability vector and implicit capability vector based on the set of operational behaviors of the personnel in the same institutional domain, extract the comprehensive fuzzy capability vector, and determine the cross-domain coverage subset of each fuzzy element in the comprehensive fuzzy capability vector; Step 4: Based on the authenticity of attendance information within the same system domain, perform a first correction on the first relevant element in the explicit ability vector and the second relevant element in the implicit ability vector of the personnel. Then, based on the cross-domain coverage subset of the same personnel within the corresponding system domain and the first-corrected ability vector, determine the clarity coefficient of the corresponding fuzzy element. If the clarity coefficient is greater than the preset coefficient, the corresponding fuzzy element is clarified according to the clarity coefficient and a second correction is performed. Otherwise, based on the difference between the preset coefficient and the clarity coefficient, determine the fuzzy probability of the corresponding fuzzy element, divide the range from 0 to 1 according to the fuzzy probability, and randomly sample according to rand(0,1) to determine the current state, and perform a third correction on the corresponding fuzzy element. Step 5: Based on the correction results of the fuzzy elements, obtain the current final vector of the personnel in the corresponding system domain, and perform cluster analysis on all current final vectors in different system domains to obtain the first sub-cluster map, and perform cluster analysis on all current final vectors in the same system domain to obtain the second sub-cluster map. Then, integrate all sub-cluster maps across domains for orderly human resource management, and update the personnel capabilities of the human resource management system regularly.

[0019] Preferably, it includes: dividing the range from 0 to 1, including: The range from 0 to the fuzzy probability is considered the fuzzy range, and the range from the fuzzy probability to 1 is considered the clear range; When a value is randomly selected based on rand(0,1), if the value falls within the fuzzy range, the current state is determined to be a fuzzy state; otherwise, the current state is determined to be a clear state.

[0020] In this embodiment, the storage database refers to a database system that interfaces with the hospital information system, laboratory information system, and electronic medical record system to centrally store relevant information such as assessment documents, attendance records, clinical operation data, and research project data of all types of personnel, including medical staff, administrative and logistical staff, and researchers.

[0021] Third-party assessment documents refer to documents issued by independent external institutions, regulatory departments, or service recipients to evaluate the performance, professional competence, and service quality of relevant hospital personnel in corresponding policy areas. For example, third-party assessment documents for doctors in the clinical medical field may include medical quality control inspection reports containing indicators such as the implementation rate of treatment guidelines and the pass rate of surgical safety checks, as well as patient satisfaction questionnaires conducted by third-party institutions, which include evaluations of medical experience, communication effectiveness, and treatment outcomes.

[0022] Self-assessment documents refer to documents prepared by hospital staff based on the assessment standards of their respective departments and positions. These documents assess their work performance, professional competence, work difficulties, and areas for improvement within the corresponding system. For example, a clinician's quarterly self-assessment report may include information such as the number of patients seen, the handling of difficult cases, the number of surgeries performed, continuing education activities, and the progress of research projects.

[0023] The institutional field refers to the professional fields that hospitals define based on their business functions and work nature, with clearly defined work scopes, assessment standards, and industry norms, covering core areas such as medical services, nursing services, medical technology support, scientific research and teaching, and administrative logistics.

[0024] In this embodiment, the explicit capability vector refers to a vector composed of capability elements that hospital personnel can directly observe and quantify in the corresponding institutional domain. Each capability element is presented in the form of a three-dimensional array containing capability type, capability confidence, and preference intensity.

[0025] Implicit capability vectors refer to vectors composed of capability elements that are difficult to quantify directly in the corresponding institutional areas of hospital personnel, but have a significant impact on work quality, efficiency, and teamwork. Each capability element is also presented in the form of a three-dimensional array containing capability type, capability confidence, and preference intensity.

[0026] Real-time online monitoring refers to the real-time collection, transmission, and recording of personnel attendance and work operation behaviors through the hospital's internal network transmission system and smart terminal devices, ensuring the timeliness and authenticity of the data. For example, smart check-in terminals, mobile nursing PDAs, and doctor workstations deployed in various departments such as outpatient clinics, nurse stations, operating rooms, and laboratories can collect medical staff's check-in data, diagnosis and treatment operation data, and nursing execution data in real time; and the operation behaviors of researchers such as project application, data entry, and result submission can be monitored in real time through a scientific research project management platform.

[0027] Attendance information refers to the attendance records of hospital staff in the corresponding system areas, covering core data such as check-in time, check-in location, attendance type, and check-in device ID, and is adapted to the diverse attendance scenarios of hospitals.

[0028] Remote attendance information refers to the off-site attendance records made by hospital staff through mobile terminals when they are unable to clock in at the hospital's attendance equipment due to official business such as consultations, academic exchanges, community clinics, or procurement of supplies. This information includes location information and business-related information.

[0029] Implementation projects refer to specific work tasks or projects in various institutional areas of the hospital that have clear objectives, execution processes, and assessment standards. For example, implementation projects in the medical service field include the acute myocardial infarction emergency PCI diagnosis and treatment project and the standardized management project for type 2 diabetes.

[0030] Operational behavior set refers to a collection of specific work behaviors performed by personnel in the process of participating in the implementation of projects in a corresponding system area, in accordance with the project assessment standards. Each behavior is directly related to the project assessment standards. For example, the operational behavior set of the acute myocardial infarction emergency PCI diagnosis and treatment project includes the doctor's patient reception and assessment behavior (taking medical history and physical examination), examination request behavior (ordering electrocardiogram and myocardial enzyme spectrum test), diagnosis behavior (confirming the diagnosis of acute myocardial infarction), treatment decision-making behavior (determining the PCI treatment plan), surgical operation behavior (PCI surgery), and postoperative monitoring behavior (postoperative vital sign monitoring and complication observation), etc.

[0031] The assessment criteria refer to specific standards developed for the implementation of various institutional areas in hospitals, in conjunction with industry norms, medical quality requirements, and hospital management regulations, to evaluate the effectiveness, standardization, and results of work behaviors. These standards include both quantitative and qualitative standards.

[0032] Actual behavior refers to the specific work behaviors that personnel actually engage in during the execution of a project, and are the real operational performance that corresponds to the assessment standards.

[0033] The comprehensive fuzzy capability vector refers to a vector formed by combining the set of operational behaviors of personnel in the same institutional field, after fuzzifying the explicit capability vector and the implicit capability vector, and integrating them. Each capability element still retains the core information of capability type, capability confidence, and preference intensity, but reflects the uncertainty of capability performance through fuzzy membership.

[0034] By analyzing the number of influencing factors of capability elements and the degree of uncertainty of each factor, capability elements with a fuzzy membership degree < 0.9 in the comprehensive fuzzy capability vector are selected and identified as fuzzy elements.

[0035] Cross-domain coverage subsets refer to the set of relevant capability elements in other institutional domains that can be adapted to each fuzzy element in the comprehensive fuzzy capability vector based on its behavioral transfer characteristics.

[0036] The authenticity of attendance information refers to the verification of the hospital staff's attendance records through multi-dimensional data validation, including both genuine attendance and false attendance.

[0037] By establishing a mapping table between explicit competency elements and attendance behavior, we can identify which explicit competency elements are affected by attendance. These elements are marked as the first relevant elements and are used as the objects for correcting the correlation between attendance authenticity and attendance. Through expert review and departmental surveys, we can sort out the indirect correlation between implicit competency elements and attendance behavior. Elements such as integrity, responsibility, and self-discipline are all affected by attendance authenticity and are marked as the second relevant elements and included in the correction scope.

[0038] The first correction refers to the process of adjusting the confidence level of the first relevant element in the explicit ability vector and the second relevant element in the implicit ability vector based on the results of the authenticity determination of attendance information. This is to ensure that the ability vector can truly reflect the work status and professional quality of personnel. For example, if a doctor's initial confidence level for meeting the attendance rate standard is 0.92, and it is found that three of his in-hospital attendance records were false attendance records, according to the correction rule, 0.05 confidence level is deducted for each false attendance record. The corrected confidence level is 0.92 - 3 × 0.05 = 0.77.

[0039] The cross-domain coverage subset refers to the set of relevant capability elements that can be covered by fuzzy elements under the corresponding institutional domain, after comprehensively considering multiple dimensions such as capability type, main promotion direction, and influencing factors. It is a supplement and refinement of the dimensions of the cross-domain coverage subset. The clarity coefficient is a quantitative indicator used to measure the degree of uncertainty of fuzzy elements in the comprehensive fuzzy capability vector. Its value ranges from 0 to 1, and the preset coefficients are different for different institutional domains. Clarification processing refers to the process of reducing the uncertainty of fuzzy elements with clarity coefficients greater than the preset coefficients by strengthening quantitative indicators and supplementing verification data, so as to make the capability performance clearer.

[0040] The second correction refers to further adjusting the capability confidence and fuzzy membership degree of the fuzzy elements after the sharpening process, based on the sharpening coefficient, to ensure that the corrected elements maintain consistency with the capability vector after the first correction. For example, if a doctor's diagnostic capability for difficult cases is sharpened to a sharpening coefficient of 0.86, according to the second correction rule (the capability confidence increases by 0.005 for every 0.01 the sharpening coefficient exceeds the preset coefficient), with a preset coefficient of 0.8 and an excess of 0.06, the capability confidence increases from 0.84 to 0.84 + 0.06 × 0.005 = 0.843, and the fuzzy membership degree increases from 0.81 to 0.93.

[0041] In this embodiment, the fuzziness probability = |preset coefficient - clarity coefficient| / preset coefficient.

[0042] The third correction refers to the process of adjusting the capability confidence and fuzzy membership degree of a fuzzy element based on its current state. If the state is clear, the fuzzy membership degree is appropriately increased; if the state is fuzzy, the fuzzy membership degree is maintained or fine-tuned to ensure that the correction result conforms to the characteristics of the current state. For example, if a nurse's current state of predicting nursing risks for elderly patients is clear, the fuzzy membership degree is increased from 0.72 to 0.85, and the capability confidence degree is fine-tuned from 0.81 to 0.83; if a doctor's current state of multidisciplinary collaborative communication ability is fuzzy, the fuzzy membership degree remains unchanged at 0.76, and the capability confidence degree is fine-tuned (±0.01) to adapt to dynamic changes.

[0043] The current final vector refers to the vector that comprehensively and accurately reflects a person's current capability status in the corresponding institutional domain after integrating the clarified fuzzy elements after the second correction, the state-based fuzzy elements after the third correction, and the explicit capability vector and implicit capability vector after the first correction, through dimensional alignment, fusion weighting, and calibration normalization. It includes core information such as capability strength, clarity, and cross-domain adaptation labels. For example, the current final vector of an internist in the medical service field includes elements such as disease diagnosis accuracy (capability strength 0.91, clarity 0.94, cross-domain adaptation label: scientific research and teaching - case analysis), doctor-patient communication ability (capability strength 0.88, clarity 0.87, cross-domain adaptation label: community health - health education), and emergency response ability (capability strength 0.85, clarity 0.82, cross-domain adaptation label: administrative logistics - medical emergency coordination).

[0044] In this embodiment, the K-means clustering algorithm is used. The number of clusters K is set according to the hospital’s staff size and management needs (e.g., K=3 for clustering by ability level and K=4 for clustering by ability type). The ability strength, clarity, and cross-domain adaptation label of the current final vector are used as feature variables. The cluster center is determined through iterative calculation. The personnel whose feature variables are closest to the cluster center are grouped into one class to complete the cluster analysis.

[0045] The first sub-cluster map refers to a visual map formed by clustering all current final vectors under different institutional fields, with institutional fields as the horizontal dimension and cluster categories as the vertical dimension. It reflects the cross-domain distribution characteristics of personnel capabilities in different fields. For example, the first sub-cluster map includes five institutional fields horizontally: medical services, nursing services, medical technology, scientific research and teaching, and administrative logistics. Vertically, it includes three cluster categories: clinical core, scientific research transformation, and management support. The map clearly marks the number of personnel and core capability characteristics of each cluster category in each field. For example, in the medical services field, there are 35 clinical core personnel whose core capabilities are disease diagnosis and treatment and surgical operation; in the scientific research and teaching field, there are 12 scientific research transformation personnel whose core capabilities are scientific research project application and achievement transformation.

[0046] The second sub-cluster map refers to a visual map formed after clustering analysis of all current final vectors under the same institutional domain. It is based on cluster categories and supplemented by detailed competency characteristics, reflecting the competency levels and differentiated characteristics of personnel within the same domain. For example, the second sub-cluster map of the nursing service domain includes three cluster categories: specialist nursing elite, basic nursing backbone, and nursing potential growth type. The specialist nursing elite (15 people) has core competencies in specialist disease nursing and solving complex nursing problems, with a clarity score above 0.85; the basic nursing backbone (40 people) has core competencies in basic nursing operations and patient care, with a clarity score of 0.75-0.85; and the nursing potential growth type (25 people) has core competencies in learning basic nursing operations and implementing nursing standards, with a clarity score of 0.65-0.75.

[0047] Cross-domain integration refers to merging the clustering results of the first and second sub-cluster maps to form a unified human resource capability clustering system for the entire hospital. For example, after cross-domain integration, cross-domain clustering categories such as clinical-research composite (covering clinical core personnel in the medical service field + research transformation personnel in the research and teaching field), nursing-community service composite (covering specialized nursing elite personnel in the nursing service field + health education personnel in the community health field), and medical technology-administration coordination type (covering medical technology precision testing personnel + administrative and logistical management support personnel) are formed, clarifying the cross-domain capability adaptation direction for each category of personnel.

[0048] Regular updates to personnel capabilities refer to the periodic collection of new assessment documents, attendance information, and operational behavior sets based on the hospital's work cycle, such as monthly or quarterly. By repeating steps 1-4, the current final vector and cluster map of personnel are updated to ensure the timeliness of capability data.

[0049] The beneficial effects of the above technical solution are as follows: through multi-dimensional data collection in hospital scenarios, accurate construction of explicit and implicit capability vectors, dynamic correction based on attendance authenticity, clarification of fuzzy elements, and cross-domain clustering integration, the comprehensiveness, dynamism, and accuracy of hospital human resource capability assessment are achieved. This effectively solves pain points such as complex hospital attendance scenarios, imbalanced capability assessment dimensions, and difficulties in cross-domain human resource allocation. It can accurately match the human resource needs of various system areas, optimize job matching and training plan formulation, improve the refinement level and operational efficiency of hospital human resource management, and ensure the quality of medical services and patient safety.

[0050] This invention provides a method for constructing a human resource management system, which determines the explicit and implicit ability vectors of corresponding personnel, including: Extract the first initial evaluation vector of the personnel in the same institutional field based on the third-party assessment and evaluation document and the second initial evaluation vector based on their own assessment and evaluation document; Keyword searches were conducted on each assessment document to identify the conventional and novel keywords for each assessment indicator, and the overall impact factor of the generation logic of each novel keyword on all conventional keywords of the corresponding assessment indicator was analyzed. When the comprehensive impact factor is greater than the preset impact factor, the corresponding novel keyword is regarded as the first keyword; otherwise, the corresponding novel keyword is removed. Determine the mapping relationship between the first keyword and each regular associated keyword, determine the derivation process according to the mapping relationship, and merge all derivation processes according to the derivation similarity of each derivation process to obtain the derivation class; The explicit and implicit evaluation elements of the first and second initial evaluation vectors are extracted respectively. The extracted explicit and implicit evaluation elements are then optimized according to the derived class to obtain a three-dimensional array for each evaluation element, which includes ability type, ability confidence and preference intensity. The explicit and implicit ability vectors of the corresponding personnel are then obtained.

[0051] Natural language processing (NLP) technology is used to parse the text of third-party assessment documents, extract the assessment indicators and corresponding scores / evaluations, and transform them into capability elements with initial confidence levels. These elements are then integrated to form the first initial assessment vector. For example, a doctor's first initial assessment vector based on the Health Commission's quarterly quality control inspection report (a third-party assessment document) includes elements such as the rate of implementation of treatment guidelines (explicit element, initial confidence level 0.89), compliance with surgical safety checks (explicit element, initial confidence level 0.92), and awareness of continuous improvement in medical quality (implicit element, initial confidence level 0.85).

[0052] By extracting self-assessment indicators and scores from online self-assessment forms, these are transformed into initial competency elements and integrated to form a second initial assessment vector. For example, a nurse's second initial assessment vector based on the department's quarterly self-assessment form includes elements such as intravenous puncture success rate (explicit element, self-assessment confidence level 0.90), standardization of nursing documentation (explicit element, self-assessment confidence level 0.88), and confidence in emergency response (implicit element, self-assessment confidence level 0.82).

[0053] In this embodiment, the TF-IDF algorithm is used to retrieve keywords from the assessment documents and compare them with a regular keyword database. Successful matches are considered regular related keywords, while unmatched matches are considered novel keywords.

[0054] Performance indicators refer to specific metrics used to evaluate an individual's performance and capabilities within a corresponding system or area. They are the core content of performance evaluation documents and include both quantitative and qualitative indicators. For example, performance indicators in the medical service field include outpatient diagnostic accuracy (quantitative), surgical complication rate (quantitative), doctor-patient communication satisfaction (quantitative), and compliance with treatment guidelines (qualitative); performance indicators in the research and teaching field include the number of research projects approved (quantitative), the number of academic papers published (quantitative), and student evaluations (qualitative).

[0055] Conventional related keywords refer to keywords that have a fixed and universal relationship with the assessment indicators. They are the routine expressions of the assessment content of the indicator. For example, the conventional related keywords for the nursing operation standardization assessment indicator include that the operation process conforms to the aseptic technique execution standard; the conventional related keywords for the scientific research and innovation capability assessment indicator include core papers and technological innovation.

[0056] Novel keywords refer to keywords that appear in the assessment documents and are related to the assessment indicators but are not included in the regular keyword database. They reflect new dimensions or personalized performance in the assessment content. For example, the regular keywords for the assessment indicators of emergency treatment capabilities are response speed, standardized treatment, and success rate. However, a nurse's assessment document may contain novel keywords such as multi-device collaborative operation and patient emotional reassurance.

[0057] The generation logic refers to the reasons why novel keywords appear in the assessment documents and the corresponding assessment dimensions, that is, the specific connotation of the personnel's work performance or ability reflected by the keyword. It uses semantic analysis algorithms in NLP technology to parse the context text of the novel keyword, extract semantic information, clarify its corresponding assessment dimensions and generation reasons, and form a generation logic description.

[0058] In this embodiment, an impact factor evaluation model is established, which scores the novel keyword from 0 to 1 point based on three dimensions: relevance (semantic similarity between the novel keyword and the conventional keyword), importance (weight of the evaluation dimension corresponding to the novel keyword in the assessment indicator), and innovativeness (whether the performance reflected by the keyword is a new requirement advocated by the industry or hospital). The comprehensive impact factor is calculated as: relevance × 0.4 + importance × 0.3 + innovativeness × 0.3.

[0059] In this embodiment, based on the industry's assessment accuracy requirements, the medical service field needs to balance standardization and personalization, the scientific research field focuses on preserving the innovation dimension, and the administrative field focuses on process standardization. Specifically, the medical service field is 0.7, the scientific research and teaching field is 0.65, and the administrative and logistical field is 0.6.

[0060] The semantic similarity algorithm is used to calculate the semantic similarity between the first keyword and each regular related keyword. Combined with the logical analysis, the mapping relationship is determined and a mapping relationship table is formed. The mapping relationship includes: extension relationship, complementary relationship or reinforcement relationship.

[0061] The derivation process refers to the logical process by which the first keyword is derived from the conventional related keywords based on the mapping relationship between the first keyword and the conventional related keywords. This reflects the expansion logic of the assessment dimensions. For example, the derivation process of the first keyword "personalized care" and the conventional keyword "communicative patience" is from generalized communication patience to personalized communication and care based on individual patient differences.

[0062] In this embodiment, the cosine similarity algorithm is used to convert the text description of the derivation process into vectors, calculate the similarity between vectors, and obtain the derivation similarity. For example, the derivation similarity between the first keyword personalized care and patient communication (general communication → personalized communication) and the derivation similarity between the first keyword timely follow-up after discharge and convenient medical treatment (basic medical treatment convenience → post-discharge continuous convenient service) is 0.62.

[0063] Merging refers to grouping derivative processes with a similarity greater than a preset similarity threshold into one category (derivative category). For example, if the preset similarity threshold is 0.6, the derivative processes of personalized care - patient communication (similarity 0.62) and timely discharge follow-up - convenient medical treatment (similarity 0.62) are merged into the basic service → personalized / continuing service derivative category; cross-unit data sharing - progress achievement rate (similarity 0.35) is treated as a single promotion → cross-unit collaborative promotion derivative category.

[0064] A derivative category refers to a collection of derivative processes that share similar derivation logic after being merged. Each derivative category contains multiple derivative processes with a unified core derivation logic. For example, the derivative category of basic services → personalized / continuous services includes two derivative processes: personalized care - patient communication and timely discharge follow-up - convenient medical treatment. The core derivation logic is to expand the dimensions of personalized and continuous services on the basis of regular basic services.

[0065] Explicit assessment elements refer to the ability-related elements in the first and second initial assessment vectors that can be directly quantified and clearly observed, corresponding to the core content of explicit abilities. For example, outpatient diagnostic accuracy rate of 89% and surgical success rate of 92% in the first initial assessment vector, and intravenous puncture success rate of 95% and nursing documentation standardization rate of 93% in the second initial assessment vector are all explicit assessment elements.

[0066] Implicit assessment elements refer to the ability-related elements in the first and second initial assessment vectors that are difficult to quantify directly and need to be inferred through indirect manifestations. These elements correspond to the core content of implicit abilities. For example, strong awareness of continuous improvement in medical quality and clear emergency decision-making in the first initial assessment vector, and high teamwork and cooperation and active scientific research and innovation in the second initial assessment vector are both implicit assessment elements.

[0067] Optimization refers to the process of supplementing and adjusting the extracted explicit and implicit assessment elements based on the core derivation logic of the derived categories, thereby improving the three-dimensional array of capability elements (capability type, capability confidence, and preference intensity). For example, the explicit assessment element of outpatient diagnostic accuracy (initial confidence 0.89) is supplemented with the dimension of personalized diagnostic plan development based on the derivative category of basic services → personalized / continuous services, and the capability confidence is adjusted to 0.91. Combined with the doctor's self-assessment of preference intensity (willingness to actively conduct personalized diagnoses 0.88), a three-dimensional array is formed (outpatient diagnostic accuracy + personalized plan development, 0.91, 0.88). The implicit assessment element of teamwork and cooperation (initial confidence 0.85) is supplemented with the dimension of cross-departmental collaboration and communication based on the derivative category of single-push → collaborative / optimized push, and the capability confidence is adjusted to 0.87, with a preference intensity of 0.90, forming a three-dimensional array (teamwork + cross-departmental communication, 0.87, 0.90).

[0068] The beneficial effects of the above technical solution are as follows: through keyword retrieval, derivative analysis and evaluation element optimization, the explicit and implicit ability vectors of hospital personnel are refined. It retains the core information of conventional assessment dimensions and incorporates novel assessment dimensions, ensuring that the ability vectors can comprehensively reflect the actual performance and potential ability of personnel. This provides high-quality basic data for subsequent ability correction and cluster analysis, and improves the depth and accuracy of hospital human resource ability assessment.

[0069] This invention provides a method for constructing a human resource management system. After real-time online monitoring of each employee's attendance information in different policy areas, the method includes: performing authenticity analysis on the attendance information, specifically including: Collect multi-source data corresponding to the attendance information, including real-time operational data of the location associated area of ​​the attendance device, historical abnormal records of the device, and historical attendance behavior sequence of the personnel. The real-time operational data includes the frequency of service complaints and the density of people in the corresponding area, and the historical attendance behavior sequence of the personnel includes the distribution characteristics of attendance time and the behavior pattern of cross-device check-in. The initial trust weight is determined based on the stability of the personnel's historical attendance behavior sequence. The trustworthiness of the device is then adjusted by combining the service complaint frequency and traffic density data of the area associated with the location of the attendance device, and the dynamic verification threshold for this attendance is obtained. The collected attendance information is matched with the personnel's historical attendance behavior sequence. If the deviation between the check-in time, device location and historical behavior pattern exceeds the dynamic verification threshold, it is marked as a suspected anomaly. At this time, the matching degree between the real-time operation data of the area where the attendance device is located and the attendance is verified. If the matching degree between the population density and the device check-in frequency in the area during the attendance period is lower than the preset value, the attendance information is back-analyzed in combination with the corresponding department's operation demand threshold. If it is determined to be false, the degree of impact on the department's operation value is determined. If the degree of impact exceeds the preset risk threshold, the manual review process is triggered. Otherwise, the authenticity is directly determined based on the dynamic verification threshold.

[0070] Location-related areas refer to hospital functional areas that are directly related to the deployment location of attendance devices or the location of card-taking.

[0071] Real-time operational data refers to the real-time operational status data of the area associated with the location of the attendance equipment during the attendance period, including data such as the number of patients, the number of patients waiting for treatment, and the frequency of complaints collected in the area associated with the location.

[0072] Service complaint frequency refers to the number of complaints received from patients or related personnel in the area associated with the location of the attendance device within a specific time period (such as 1 hour before or after the attendance period).

[0073] People density data refers to the density of people in the area associated with the location of the attendance equipment during the attendance period. It is calculated by combining data such as the number of patients and staff in the area and reflects the workload of the area.

[0074] Historical anomaly records refer to records of attendance data anomalies and equipment malfunctions that occurred within a certain period, such as the past 30 days. These records include information such as the type of anomaly, the time of occurrence, and the handling results.

[0075] A person's historical attendance behavior sequence refers to a sequence of all attendance records for that person within a certain period, such as the past three months, arranged chronologically. This sequence includes patterns such as attendance time, clock-in location, and attendance type. Specifically, it involves extracting the person's attendance records for the past three months from the attendance management system and sorting them by timestamp to form a historical attendance behavior sequence.

[0076] The distribution characteristics of attendance time refer to the regularity of the clock-in time, such as the concentration range and fluctuation range, in the historical attendance behavior sequence of personnel. That is, statistical analysis of the clock-in time of personnel's historical attendance is carried out to calculate the average value, standard deviation, and concentration range of clock-in time, such as 8:20-8:30, to form a description of the distribution characteristics of attendance time.

[0077] In this embodiment, the dynamic verification threshold is calculated as follows: ,in, For dynamic verification thresholds; The baseline threshold; The standard deviation of historical attendance time deviation; Real-time pedestrian density in the area associated with the device during the attendance period; This represents the highest population density in the area's history. The frequency of service complaints in the area associated with the device over the past 7 days; This represents the highest number of service complaints in the region's history. This refers to the number of abnormal events that occurred with the attendance equipment in the past 30 days. This represents the highest number of abnormal events in the device's history. The average deviation rate of attendance time for members of the same attendance group; This represents the maximum deviation of the historical attendance time for personnel.

[0078] benchmark threshold Value: Default is 0.8, adjusted according to hospital size (0.85 for tertiary hospitals, 0.75 for secondary hospitals, and 0.7 for community hospitals). Average deviation rate of attendance time for members of the same attendance group Calculation method: =Average of absolute deviations in attendance time among group members / Standard attendance duration, where the standard attendance duration is set at 480 minutes based on an 8-hour workday.

[0079] Highest historical population density Highest frequency of service complaints in history The highest number of abnormal events in the history of the equipment This is based on statistics from nearly 12 months of historical data stored in the database, and is updated monthly.

[0080] The beneficial effects of the above technical solution are: to analyze and match multi-source data from different dimensions to determine the deviation and dynamic verification threshold, and then to determine the authenticity of attendance by matching and comparing attendance time periods, thus providing an effective basis for subsequent analysis.

[0081] This invention provides a method for constructing a human resource management system, which extracts a comprehensive fuzzy capability vector, including: Each actual behavior of the personnel in the same system domain is compared and analyzed with the standard behavior of the assessment standard corresponding to the ability element in the ability vector according to the implementation project. The difference behavior of the corresponding ability element is obtained, and the negative and positive promoting factors of the difference behavior are analyzed. The negative and positive promoting factors are fused to obtain the main promoting direction, and the main promoting direction is then used as the basis for further analysis. The standard behaviors defined by the assessment criteria of the corresponding ability elements in the explicit and implicit ability vectors are matched in a scenario-based manner. Based on the matching results, the deviation between the actual behavior and the standard behavior is extracted as the difference behavior. Among them, the difference behavior corresponding to the explicit ability elements focuses on the execution deviation that can be quantified, while the difference behavior corresponding to the implicit ability elements focuses on the decision deviation of the implicit behavior of collaboration logic and risk prediction. The negative and positive contributing factors for each differential behavior are identified separately. Negative contributing factors include triggers that lead to deterioration of performance due to operational errors, process delays, or collaboration conflicts. Positive contributing factors include triggers that drive performance iteration through innovation optimization, efficiency improvement, or risk avoidance. At the same time, the impact depth of each contributing factor is quantified by combining the complexity coefficient of the implemented project. Calculate the entropy value of the influence of negative promoting factors. Entropy value of the influence of positive promoting factors The magnitude of the entropy value is determined by the depth of influence of the factor, the frequency of occurrence, and the project complexity coefficient. when When the main promoting direction is determined to be the positive iteration direction, then, The preset impact difference threshold; when When the main promoting direction is determined to be the negative deterioration direction; Otherwise, the main promoting direction is determined to be a neutral and stable direction; Based on the main facilitating direction, the explicit and implicit capability vectors are fuzzified and adjusted, including: If the main promoting direction is the positive iteration direction, the fuzzy membership degree of the corresponding capability element is improved, and the iteration gain coefficient is introduced to expand the fuzzy coverage of the positive influence. If the main promoting direction is the negative deterioration direction, the fuzzy membership degree of the corresponding capability element is reduced, and a deterioration attenuation coefficient is introduced to reduce the fuzzy coverage of the negative influence. If the main promoting direction is a neutral and stable direction, maintain the fuzzy membership degree of the corresponding capability element, and introduce a stability correction coefficient to maintain the dynamic balance of the fuzzy coverage range. By integrating the fuzzification adjustment results of explicit and implicit capability vectors, a comprehensive fuzzy capability vector of the personnel in the corresponding institutional domain is obtained.

[0082] In this embodiment, the standard behavior refers to the standardized work behavior that is clearly defined by the assessment criteria for each ability element in the explicit ability vector and the implicit ability vector, and that conforms to industry norms, hospital management requirements and project execution goals. It is the benchmark for measuring actual behavior, that is, it is pre-set and can be used directly.

[0083] In this embodiment, each implementation project is labeled with a scenario type, such as emergency, outpatient, research, or administrative, and a scenario-standard behavior mapping relationship is established. During matching, the project scenario is first identified, and then the standard behaviors under the corresponding scenario are compared with the actual behaviors to ensure the relevance of the analysis. For example, in the emergency treatment scenario for acute stroke, the doctor's actual treatment behaviors (reception time, speed of examination request, treatment plan formulation process) are matched with the standard behaviors under the emergency scenario (reception to thrombolysis treatment time ≤ 60 minutes, priority to order head CT examination, and formulation of individualized treatment plans according to guidelines).

[0084] In this embodiment, by quantitative comparison (explicit capability elements) or semantic analysis (implicit capability elements), the deviation between actual behavior and standard behavior is extracted, and the deviation type (numerical deviation, process deviation, logical deviation) and deviation degree (absolute deviation, relative deviation) are identified to form a description of the difference in behavior. For example, the standard behavior for the explicit capability element of intravenous puncture operation is a puncture success rate ≥95% and a first-time puncture success rate ≥90%. A nurse's actual behavior is a puncture success rate of 92% and a first-time puncture success rate of 85%. The difference in behavior is that the puncture success rate is 3 percentage points lower than the standard and the first-time puncture success rate is 5 percentage points lower than the standard. The standard behavior for the implicit capability element of risk prediction ability is to identify ≥90% of potential surgical risks before surgery and formulate a response plan. A doctor's actual behavior is to identify 75% of potential surgical risks before surgery, but no response plan was formulated for 2 key risks. The difference in behavior is that the risk identification rate is 15 percentage points lower than the standard and the response plan for key risks is missing.

[0085] Negative contributing factors refer to triggering factors that lead to differential behavior and deterioration of performance, including specific situations such as operational errors, process delays, and collaboration conflicts. The depth of influence needs to be quantified in conjunction with the complexity coefficient of the implemented project. The depth of influence of negative contributing factors = expert score × project complexity coefficient, where the expert score ranges from 0 to 1.

[0086] Expert scoring rules: A 5-level scoring system is adopted (1 point = minimal impact, 2 points = minor impact, 3 points = moderate impact, 4 points = significant impact, 5 points = extremely significant impact). The scores are normalized to the [0,1] interval (normalization formula: normalized score = (expert score - 1) / 4). Project complexity coefficient calculation formula: Project complexity coefficient = (Score for number of participants × 0.2 + Score for technical difficulty × 0.4 + Score for execution cycle × 0.2 + Score for scope of impact × 0.2) Scoring rules for each dimension: All dimensions use a 1-5 point scale, and the scores are normalized to the [0,1] interval, where: Number of participants: 1 person = 1 point, 2-5 people = 2 points, 6-10 people = 3 points, 11-20 people = 4 points, more than 20 people = 5 points; Technical difficulty dimension: Routine operation = 1 point, basic professional skills required = 2 points, advanced skills required = 3 points, expert-level skills required = 4 points, cutting-edge interdisciplinary skills required = 5 points; Execution cycle dimension: within 1 week = 1 point, 1-4 weeks = 2 points, 1-3 months = 3 points, 3-6 months = 4 points, more than 6 months = 5 points; Dimensions of impact: Individual = 1 point, Department = 2 points, Hospital = 3 points, Regional healthcare system = 4 points, National / Industry = 5 points.

[0087] Positive contributing factors refer to the triggering factors that cause differential behavior (positive deviation) and drive iterative optimization of performance, including specific situations such as innovation optimization, efficiency improvement, and risk avoidance. The depth of influence is also quantified by combining the project complexity coefficient, and its quantification method is consistent with that of negative contributing factors.

[0088] In this embodiment, the project complexity coefficient = (normalized coefficient of score for number of participants × 0.2 + normalized coefficient of technical difficulty × 0.4 + normalized coefficient of execution cycle × 0.2 + normalized coefficient of impact scope × 0.2).

[0089] Entropy of influence: E = -k × Σ (Pi1 × lnPi1), where k is the Boltzmann constant, and Pi1 is the normalized probability of the i1th influence, which is obtained by weighting the influence depth, frequency of occurrence, and project complexity coefficient. Relevant data is automatically collected and substituted into the calculation to output the entropy value of the positive influence. With negative influence on entropy value .

[0090] Pi1 is the normalized probability of the i1th contributing factor, calculated as: Pi = (the influence depth of this factor × frequency of occurrence) / Σ (the influence depth of all factors × frequency of occurrence).

[0091] Frequency statistics rules: The statistical interval is the project implementation cycle. The occurrence of positive / negative promoting factors is counted according to the actual records. If no occurrence occurs, it is recorded as 0.

[0092] Preset impact difference threshold This refers to the critical value used to determine the main direction of promotion. It is set by the hospital according to the accuracy requirements of capability assessment in each institutional area. Different thresholds can be set for different areas, namely the medical service area (which is directly related to patient safety and has high sensitivity to changes in capability). =0.15; δ=0.2 in the administrative and logistical field (where the impact of capability changes is relatively mild); δ=0.2 in the scientific research and teaching field. =0.18.

[0093] The main promoting direction refers to the overall trend of change in the performance of a capability element determined by comparing the entropy values ​​of the combined positive and negative promoting factors. This includes three categories: positive iterative direction, negative deterioration direction, and neutral stable direction. For example, the interventional treatment technical capability of a doctor... =0.78、 =0.52, =0.26> =0.15, the main promoting direction is the positive iteration direction.

[0094] Fuzzy membership degree refers to a quantitative index used to describe the fuzzy characteristics of each capability element in the comprehensive fuzzy capability vector. The value range is 0-1. For example, the basic fuzzy membership degree is set based on the main promoting direction (0.75-0.9 for positive iteration direction, 0.5-0.65 for negative degradation direction, and 0.65-0.75 for neutral and stable direction). Fine-tuning is then performed based on the degree of fluctuation affecting the entropy value (the larger the entropy value, the lower the membership degree by 0.03-0.05; the smaller the entropy value, the higher the membership degree by 0.03-0.05). Finally, the fuzzy membership degree of each capability element is determined.

[0095] The iterative gain coefficient is an adjustment coefficient used to expand the fuzzy coverage of the positive influence when the main promoting direction is the positive iterative direction. The value range is 1.1-1.3, which is determined according to the difference between the entropy value of the positive influence and the entropy value of the negative influence. That is, the correspondence between the iterative gain coefficient and the entropy value difference is set (the coefficient is 1.1 for a difference of 0.1-0.2, 1.2 for a difference of 0.2-0.3, and 1.3 for a difference of >0.3). The coefficient is automatically matched according to the calculated entropy value difference and used for the enhancement of fuzzy membership degree.

[0096] The degradation attenuation coefficient is an adjustment coefficient used to reduce the fuzzy coverage of the negative influence when the main promoting direction is the negative degradation direction. The value range is 0.7-0.9, which is determined according to the difference between the entropy value of the negative influence and the entropy value of the positive influence. That is, a correspondence is established between the degradation attenuation coefficient and the entropy value difference (0.1-0.2 corresponds to coefficient 0.9, 0.2-0.3 corresponds to coefficient 0.8, and >0.3 corresponds to coefficient 0.7). The system automatically matches the coefficient for the reduction of fuzzy membership degree.

[0097] The stability correction coefficient is an adjustment coefficient used to maintain the dynamic balance of the fuzzy coverage range when the main promotion direction is a neutral and stable direction. The value is fixed at 1.0 to ensure that the fuzzy membership degree remains stable. That is, the system presets the stability correction coefficient to 1.0. When the main promotion direction is determined to be a neutral and stable direction, the coefficient is automatically called without increasing or decreasing the fuzzy membership degree, and only the original value is maintained.

[0098] Fuzzy adjustment refers to the process of adjusting the fuzzy membership degree of each capability element in the explicit capability vector and the implicit capability vector based on the main promoting direction and the corresponding adjustment coefficients (iterative gain coefficient, degradation attenuation coefficient, and stability correction coefficient).

[0099] In this embodiment, for the clear state (random value ∈ [P,1]): the adjusted fuzzy membership degree = the original fuzzy membership degree × 1.15 (upper limit 0.95) and the adjusted capability confidence degree = the original capability confidence degree × (1 + 0.02 × (Cl - Cl0) / Cl0); Fuzzy state (random value ∈ [0, P]): Adjusted fuzzy membership degree = original fuzzy membership degree × 0.95 (lower limit 0.6) Adjusted capability confidence degree = original capability confidence degree × (1 - 0.01 × P) (lower limit 0.6).

[0100] The beneficial effects of the above technical solution are as follows: by extracting differential behaviors through scenario-based matching, quantitatively analyzing the impact of positive and negative promoting factors, clarifying the main promoting direction and making fuzzy adjustments, dynamic optimization of the hospital personnel's ability vector is achieved. This not only accurately captures the changing trend of ability performance, but also scientifically handles the fuzziness problem in ability assessment, providing high-quality intermediate data for subsequent fuzzy element clarification and final ability vector construction, thereby improving the dynamism and reliability of hospital personnel's ability assessment.

[0101] This invention provides a method for constructing a human resource management system, comprising determining the cross-domain coverage subset of each fuzzy element in the comprehensive fuzzy capability vector, including: Extract the behavior transfer feature vector of each fuzzy element in the comprehensive fuzzy capability vector and quantize it into a multi-dimensional numerical sequence. The behavior transfer feature vector includes the main promoting direction label of the corresponding fuzzy element, the proportion of negative promoting factors affecting entropy value, the proportion of positive promoting factors affecting entropy value, and the fuzzy membership degree. By integrating capability elements from all institutional domains, a cross-domain capability adaptability map is constructed. Each node represents a capability element of the target domain, and the node attributes include the adaptability coefficients of the corresponding institutional domain to the three main promoting directions: positive iteration, negative degradation, and neutral stability. The behavior transfer feature vector of the fuzzy element is matched with the attribute vector of the target domain capability element in the cross-domain capability adaptability map using cosine similarity. Combined with the adaptation coefficient of the main facilitation direction, the cross-domain coverage confidence is calculated. ,in, Let i be the cross-domain coverage confidence of fuzzy element i to target domain capability element j; For behavioral transfer feature vectors With the target ability element attribute vector Cosine similarity; target area The main promoting direction of fuzzy element i The adaptation coefficient; Filter all confidence levels The target domain capability elements are used to form a cross-domain coverage subset of the corresponding fuzzy elements, and each element in the subset is labeled with a facilitating direction adaptation tag. To cover the confidence threshold.

[0102] In this embodiment, the behavior transfer feature vector includes four core dimensions: the main promoting direction label, the proportion of entropy value influenced by negative promoting factors, the proportion of entropy value influenced by positive promoting factors, and fuzzy membership degree. For example, the behavior transfer feature vector of a clinician's case data analysis ability (fuzzy elements, positive iteration of the main promoting direction) is [1 (positive iteration label), 0.3 ( Percentage), 0.7 ( [Proportion), 0.85 (fuzzy membership degree)], where labels are assigned to the three main promoting directions (positive iteration = 1, negative degradation = 0, neutral stability = 2), and the proportion of negative influence entropy is calculated. The proportion of positive influence on entropy value Combining fuzzy membership degrees, a four-dimensional numerical sequence is formed as the behavioral transfer feature vector. For example, the behavioral transfer feature vector of a clinician's case data analysis ability (fuzzy elements, main promoting direction of positive iteration) is [1 (positive iteration label), 0.3 ( Percentage), 0.7 ( (Percentage), 0.85 (fuzzy membership degree)].

[0103] Multidimensional numerical sequences refer to the core dimensions (primary facilitator direction label, E) of behavior transfer feature vectors. - Percentage, E + The percentage and fuzzy membership degree are converted into a combination of numerical values ​​in a unified format (0-1 or specified integer labels), such as a multidimensional numerical sequence of [1, 0.3, 0.7, 0.85].

[0104] A graph database (such as Neo4j) is used to construct a graph, and capability elements from various institutional domains are entered as nodes. Node attributes (adaptation coefficients) are defined, and edges represent the potential adaptation relationships of capability elements from different domains. Nodes can be filtered by domain and main promotion direction, thereby constructing a cross-domain capability adaptability graph. This graph integrates capability elements from all institutional domains of the hospital, constructing a visual graph with capability elements as nodes and adaptation relationships as edges. Each node (target domain capability element) contains the adaptation coefficient attribute of the corresponding institutional domain for the three main promotion directions. For example, the cross-domain capability adaptability map contains more than 200 capability element nodes in five institutional areas: medical services, nursing services, medical technology, scientific research and teaching, and administrative logistics. Among them, the attributes of the clinical data statistical analysis capability node in the scientific research and teaching field are [positive iterative adaptation coefficient 0.9, negative deterioration adaptation coefficient 0.3, and neutral stable adaptation coefficient 0.6]; the attributes of the health science popularization capability node in the community health field are [positive iterative adaptation coefficient 0.85, negative deterioration adaptation coefficient 0.4, and neutral stable adaptation coefficient 0.7]. The weight of the associated edge is calculated as (product of the two element adaptation coefficients) × 0.5 + (average cross-domain coverage confidence score) × 0.5.

[0105] The target domain refers to other institutional domains that may have an adaptation relationship with the fuzzy element, excluding the institutional domain to which the fuzzy element belongs. It is the source domain of the cross-domain coverage subset.

[0106] The functional departments, in conjunction with experts from various fields, determine the adaptation coefficients based on the core needs of the target field (innovation demand, fault tolerance, and process stability requirements). Among them, the positive adaptation coefficient is determined by the innovation demand, the negative adaptation coefficient is determined by the fault tolerance, and the neutral adaptation coefficient is determined by the process stability requirements.

[0107] Cosine similarity matching refers to calculating the cosine similarity between the behavioral transfer feature vector (multidimensional numerical sequence) of fuzzy elements and the attribute vector (adaptation coefficient sequence) of the target domain capability elements.

[0108] The coverage confidence threshold is set by the hospital based on the accuracy requirements of cross-domain manpower allocation, with a default value of 0.7. It is adjusted according to the domain. For example, the coverage confidence threshold between the medical service domain and the scientific research and teaching domain is 0.75 (requiring higher adaptability); and the threshold between the nursing service domain and the community health domain is 0.65 (the adaptability range can be appropriately relaxed).

[0109] The "promotion direction adaptation label" refers to the adaptation type label assigned to each target domain capability element in the cross-domain coverage subset, corresponding to the main promotion direction of the fuzzy element. It includes three categories: positive adaptation, negative adaptation, and neutral adaptation. For example, in the cross-domain coverage subset of the fuzzy element case data analysis capability (positive iteration direction), the clinical data statistical analysis capability is labeled with a positive adaptation label; in the subset of the fuzzy element patient psychological counseling capability (neutral and stable direction), the health science popularization capability is labeled with a neutral adaptation label.

[0110] The beneficial effects of the above technical solution are as follows: by constructing a cross-domain capability adaptability map and combining cosine similarity and adaptability coefficient to calculate the coverage confidence, the cross-domain adaptability potential of fuzzy capability elements of hospital personnel is accurately mined, providing data support for cross-domain human resource allocation and job adaptation, breaking down capability barriers between institutional domains, and improving the flexibility and rationality of hospital human resource allocation.

[0111] This invention provides a method for constructing a human resource management system, comprising determining the sharpness coefficient of corresponding fuzzy elements, including: ,in, This represents the sharpness coefficient for the corresponding blurred element; The average coverage confidence of the cross-domain coverage subset; The corrected stability of the first modified capability vector; The adaptation coefficient for the main promotion direction; , , These are the dimension weight coefficients, and ; The interval between the generation time and the current time for cross-domain coverage subsets; This is the decay rate coefficient; Weighting of contribution to the domain.

[0112] In this embodiment, based on the core needs of each field, the research field focuses on cross-domain capability transfer. =0.5, in the administrative field, the focus is on the stability of capability data. =0.4, in this scheme =0.4、 , .

[0113] In this embodiment, The monthly update cycle is 0.1, and the quarterly update cycle is 0.05.

[0114] In this embodiment, the domain contribution weights are: 1.2 for medical services, 1.0 for scientific research and teaching, and 0.8 for administrative and logistical support.

[0115] In this embodiment, ,in, Assigning domain weights to each capability element, with explicit elements in the healthcare service domain. 0.5, a latent element The value is 0.5, representing explicit elements in the administrative field. 0.6, a latent element It is 0.4.

[0116] Where C0 is the confidence level of capability before correction, and C1 is the confidence level of capability after correction.

[0117] The beneficial effects of the above technical solution are: by integrating and quantifying multi-dimensional parameters, a scientific and quantifiable method is provided for determining the clarity coefficient of fuzzy elements. This method considers the reliability of cross-domain adaptation, the stability of capability correction, and also takes into account timeliness and domain importance, ensuring that the clarity coefficient can objectively reflect the degree of uncertainty of fuzzy elements. This provides an accurate basis for the subsequent classification and correction of fuzzy elements and further improves the scientific nature of hospital personnel capability assessment.

[0118] This invention provides a method for constructing a human resource management system, which obtains the current final vector of the personnel in the corresponding system domain based on the correction results of fuzzy elements, including: The second revised sharpening fuzzy element and the third revised stateful fuzzy element are aligned with the first revised explicit capability vector and implicit capability vector according to the capability element dimension, and the fusion weight is assigned in combination with the main promoting direction corresponding to each fuzzy element. The average coverage confidence of the cross-domain coverage subset of fuzzy elements is used as a calibration factor to calibrate the fused vector. The calibrated vector is then normalized to generate the current final vector containing capability strength, clarity, and cross-domain adaptation labels.

[0119] In this embodiment, the fusion weight refers to the integration weight assigned to the corrected fuzzy elements based on the main promoting direction corresponding to each fuzzy element. This weight is used to adjust the influence of elements with different trend capabilities in the final vector. For example, the fusion weight of elements with the main promoting direction being the positive iteration direction is 1.1-1.2 (e.g., 1.15); the fusion weight of the negative degradation direction is 0.8-0.9 (e.g., 0.85); and the fusion weight of the neutral and stable direction is 1.0. The calibrated vector = initial vector × calibration factor.

[0120] Normalization refers to mapping the calibrated vector values ​​to the standard range of 0-1, eliminating the influence of differences in values ​​across different dimensions, and making the elements of the final vector comparable.

[0121] Capability strength refers to the quantitative performance strength of each capability element in the corresponding dimension after normalization, with a value range of 0-1.

[0122] Clarity refers to the level of clarity of each capability element after correction, integration, and calibration, with a value range of 0-1.

[0123] Cross-domain adaptation tags refer to the target domain tags that each capability element can be labeled with based on the screening results of the cross-domain coverage subset. For example, the cross-domain adaptation tags for professional and technical capabilities - case data analysis are scientific research and teaching fields and administrative and logistical fields (medical quality monitoring); the cross-domain adaptation tags for collaborative communication capabilities - patient psychological counseling are community health fields and rehabilitation nursing fields.

[0124] The current final vector refers to the complete vector that, after dimensional alignment, fusion weighting, calibration, and normalization, includes the strength, clarity, and cross-domain adaptation labels of each dimension's capabilities. It is the final result reflecting the current capability status of an individual. For example, the current final vector of an internist in the field of medical services is: [Professional and technical capabilities (strength 1.00, clarity 0.91, adaptation label: scientific research and teaching - clinical data statistics), collaborative communication capabilities (strength 0.50, clarity 0.83, adaptation label: community health - health education), innovation and iteration capabilities (strength 0.00, clarity 0.76, adaptation label: none), and responsibility and accountability capabilities (strength 0.72, clarity 0.88, adaptation label: administrative logistics - medical emergency coordination)].

[0125] The beneficial effects of the above technical solution are as follows: through the systematic processing of dimensional alignment, fusion weighting, calibration and normalization, the multi-source corrected capability elements are integrated into a unified and standardized final vector, which not only comprehensively reflects the capability strength and clarity of personnel in each dimension, but also marks the cross-domain adaptation direction, providing accurate and intuitive decision-making basis for hospital human resource allocation, job matching and training plan formulation, and effectively improving the level of refinement and efficiency of human resource management.

[0126] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for constructing a human resource management system, characterized in that, include: Step 1: Retrieve third-party assessment documents and self-assessment documents for each person in different institutional areas from the stored database and conduct comprehensive analysis to determine the explicit and implicit capability vectors of the person in the corresponding institutional areas; Step 2: Monitor the attendance information of each person in different system areas in real time and capture the set of operational behaviors of each person in different system areas for each implementation project. The attendance information includes remote check-in information, and the set of operational behaviors includes the actual behavior of each assessment standard set under the corresponding implementation project. Step 3: Analyze the corresponding explicit capability vector and implicit capability vector based on the set of operational behaviors of the personnel in the same institutional domain, extract the comprehensive fuzzy capability vector, and determine the cross-domain coverage subset of each fuzzy element in the comprehensive fuzzy capability vector; Step 4: Based on the authenticity of attendance information within the same system domain, perform a first correction on the first relevant element in the explicit ability vector and the second relevant element in the implicit ability vector of the personnel. Then, based on the cross-domain coverage subset of the same personnel within the corresponding system domain and the first-corrected ability vector, determine the clarity coefficient of the corresponding fuzzy element. If the clarity coefficient is greater than the preset coefficient, the corresponding fuzzy element is clarified according to the clarity coefficient and a second correction is performed. Otherwise, based on the difference between the preset coefficient and the clarity coefficient, determine the fuzzy probability of the corresponding fuzzy element, divide the range from 0 to 1 according to the fuzzy probability, and randomly sample according to rand(0,1) to determine the current state, and perform a third correction on the corresponding fuzzy element. Step 5: Based on the correction results of the fuzzy elements, obtain the current final vector of the personnel in the corresponding system domain, and perform cluster analysis on all current final vectors in different system domains to obtain the first sub-cluster map, and perform cluster analysis on all current final vectors in the same system domain to obtain the second sub-cluster map. Then, integrate all sub-cluster maps across domains for orderly human resource management, and update the personnel capabilities of the human resource management system regularly.

2. The method for constructing a human resource management system according to claim 1, characterized in that, include: The range from 0 to 1 is divided into the following categories: The range from 0 to the fuzzy probability is considered the fuzzy range, and the range from the fuzzy probability to 1 is considered the clear range; When a value is randomly selected based on rand(0,1), if the value falls within the fuzzy range, the current state is determined to be a fuzzy state; otherwise, the current state is determined to be a clear state.

3. The method for constructing a human resource management system according to claim 1, characterized in that, Determine the explicit and implicit ability vectors of the corresponding personnel, including: Extract the first initial evaluation vector of the personnel in the same institutional field based on the third-party assessment and evaluation document and the second initial evaluation vector based on their own assessment and evaluation document; Keyword searches were conducted on each assessment document to identify the conventional and novel keywords for each assessment indicator, and the overall impact factor of the generation logic of each novel keyword on all conventional keywords of the corresponding assessment indicator was analyzed. When the comprehensive impact factor is greater than the preset impact factor, the corresponding novel keyword is regarded as the first keyword; otherwise, the corresponding novel keyword is removed. Determine the mapping relationship between the first keyword and each regular associated keyword, determine the derivation process according to the mapping relationship, and merge all derivation processes according to the derivation similarity of each derivation process to obtain the derivation class; The explicit and implicit evaluation elements of the first and second initial evaluation vectors are extracted respectively. The extracted explicit and implicit evaluation elements are then optimized according to the derived class to obtain a three-dimensional array for each evaluation element, which includes ability type, ability confidence and preference intensity. The explicit and implicit ability vectors of the corresponding personnel are then obtained.

4. The method for constructing a human resource management system according to claim 1, characterized in that, After real-time online monitoring of each person's attendance information in different policy areas, the process includes: analyzing the authenticity of the attendance information, specifically including: Collect multi-source data corresponding to the attendance information, including real-time operational data of the location associated area of ​​the attendance device, historical abnormal records of the device, and historical attendance behavior sequence of the personnel. The real-time operational data includes the frequency of service complaints and the density of people in the corresponding area, and the historical attendance behavior sequence of the personnel includes the distribution characteristics of attendance time and the behavior pattern of cross-device check-in. The initial trust weight is determined based on the stability of the personnel's historical attendance behavior sequence. The trustworthiness of the device is then adjusted by combining the service complaint frequency and traffic density data of the area associated with the location of the attendance device, and the dynamic verification threshold for this attendance is obtained. The collected attendance information is matched with the personnel's historical attendance behavior sequence. If the deviation between the check-in time, device location and historical behavior pattern exceeds the dynamic verification threshold, it is marked as a suspected anomaly. At this time, the matching degree between the real-time operation data of the area where the attendance device is located and the attendance is verified. If the matching degree between the population density and the device check-in frequency in the area during the attendance period is lower than the preset value, the attendance information is back-analyzed in combination with the corresponding department's operation demand threshold. If it is determined to be false, the degree of impact on the department's operation value is determined. If the degree of impact exceeds the preset risk threshold, the manual review process is triggered. Otherwise, the authenticity is directly determined based on the dynamic verification threshold.

5. The method for constructing a human resource management system according to claim 1, characterized in that, Extract the comprehensive fuzzy capability vector, including: Each actual behavior of the personnel in the same system domain is compared and analyzed with the standard behavior of the assessment standard corresponding to the ability element in the ability vector according to the implementation project. The difference behavior of the corresponding ability element is obtained, and the negative and positive promoting factors of the difference behavior are analyzed. The negative and positive promoting factors are fused to obtain the main promoting direction, and the main promoting direction is then used as the basis for further analysis. The standard behaviors defined by the assessment criteria of the corresponding ability elements in the explicit and implicit ability vectors are matched in a scenario-based manner. Based on the matching results, the deviation between the actual behavior and the standard behavior is extracted as the difference behavior. Among them, the difference behavior corresponding to the explicit ability elements focuses on the execution deviation that can be quantified, while the difference behavior corresponding to the implicit ability elements focuses on the decision deviation of the implicit behavior of collaboration logic and risk prediction. The negative and positive contributing factors for each differential behavior are identified separately. Negative contributing factors include triggers that lead to deterioration of performance due to operational errors, process delays, or collaboration conflicts. Positive contributing factors include triggers that drive performance iteration through innovation optimization, efficiency improvement, or risk avoidance. At the same time, the impact depth of each contributing factor is quantified by combining the complexity coefficient of the implemented project. Calculate the entropy value of the influence of negative promoting factors. Entropy value of the influence of positive promoting factors The magnitude of the entropy value is determined by the depth of influence of the factors, the frequency of occurrence, and the project complexity coefficient. when When the main promoting direction is determined to be the positive iteration direction, then, The preset impact difference threshold; when When the main promoting direction is determined to be the negative deterioration direction; Otherwise, the main promoting direction is determined to be a neutral and stable direction; Based on the main facilitating direction, the explicit and implicit capability vectors are fuzzified and adjusted, including: If the main promoting direction is the positive iteration direction, the fuzzy membership degree of the corresponding capability element is improved, and the iteration gain coefficient is introduced to expand the fuzzy coverage of the positive influence. If the main promoting direction is the negative deterioration direction, the fuzzy membership degree of the corresponding capability element is reduced, and a deterioration attenuation coefficient is introduced to reduce the fuzzy coverage of the negative influence. If the main promoting direction is a neutral and stable direction, maintain the fuzzy membership degree of the corresponding capability element, and introduce a stability correction coefficient to maintain the dynamic balance of the fuzzy coverage range. By integrating the fuzzification adjustment results of explicit and implicit capability vectors, a comprehensive fuzzy capability vector of the personnel in the corresponding institutional domain is obtained.

6. The method for constructing a human resource management system according to claim 5, characterized in that, Determining the cross-domain coverage subset of each fuzzy element in the comprehensive fuzzy capability vector includes: Extract the behavior transfer feature vector of each fuzzy element in the comprehensive fuzzy capability vector and quantize it into a multi-dimensional numerical sequence. The behavior transfer feature vector includes the main promoting direction label of the corresponding fuzzy element, the proportion of negative promoting factors affecting entropy value, the proportion of positive promoting factors affecting entropy value, and the fuzzy membership degree. By integrating capability elements from all institutional domains, a cross-domain capability adaptability map is constructed. Each node represents a capability element of the target domain, and the node attributes include the adaptability coefficients of the corresponding institutional domain to the three main promoting directions: positive iteration, negative degradation, and neutral stability. The behavior transfer feature vector of the fuzzy element is matched with the attribute vector of the target domain capability element in the cross-domain capability adaptability map using cosine similarity. Combined with the adaptation coefficient of the main facilitation direction, the cross-domain coverage confidence is calculated. ,in, Let i be the cross-domain coverage confidence of fuzzy element i to target domain capability element j; For behavioral transfer feature vectors With the target ability element attribute vector Cosine similarity; target area The main promoting direction of fuzzy element i The adaptation coefficient; Filter all confidence levels The target domain capability elements are used to form a cross-domain coverage subset of the corresponding fuzzy elements, and each element in the subset is labeled with a facilitating direction adaptation tag. To cover the confidence threshold.

7. The method for constructing a human resource management system according to claim 1, characterized in that, Determine the sharpness coefficient of the corresponding blurred element, including: ,in, This represents the sharpness coefficient for the corresponding blurred element; The average coverage confidence of the cross-domain coverage subset; The corrected stability of the first modified capability vector; The adaptation coefficient for the main promotion direction; , , These are the dimension weight coefficients, and ; The interval between the generation time and the current time for cross-domain coverage subsets; This is the decay rate coefficient; Weighting of contribution to the domain.

8. The method for constructing a human resource management system according to claim 1, characterized in that, Based on the correction results of the fuzzy elements, the current final vector of the person in the corresponding institutional domain is obtained, including: The second revised sharpening fuzzy element and the third revised stateful fuzzy element are aligned with the first revised explicit capability vector and implicit capability vector according to the capability element dimension, and the fusion weight is assigned in combination with the main promoting direction corresponding to each fuzzy element. The average coverage confidence of the cross-domain coverage subset of fuzzy elements is used as a calibration factor to calibrate the fused vector. The calibrated vector is then normalized to generate the current final vector containing capability strength, clarity, and cross-domain adaptation labels.

Images