A lifecycle management system and method for key and sensitive positions
The full lifecycle management system based on knowledge graphs and machine learning models solves the problems of subjectivity and lag in the management of key and sensitive positions. It enables real-time position identification, qualification review, job rotation planning, and risk warning, improving the scientific nature and real-time performance of management and ensuring the enforceability and traceability of the system.
Patent Information
- Application Number
- CN202611010078.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-08
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, the management of key and sensitive positions mainly relies on manual methods, which leads to strong subjectivity in position identification, lagging qualification review, blind job rotation planning, lack of risk warning and difficulty in execution supervision, long information transmission cycle and difficulty in effectively controlling risky positions.
The system adopts a full lifecycle management system based on knowledge graphs and machine learning models. Through job evaluation, multi-dimensional compliance profiles, quantitative job correlation matrices, and anomaly detection models, it can dynamically update the list of key and sensitive positions, conduct real-time qualification reviews, optimize job rotation plans, and provide tiered early warnings. It also supports electronic signatures and intelligent handover list generation.
It has improved the scientific, real-time and penetrating nature of key and sensitive position management, dynamically updated the list of key and sensitive positions, prevented unqualified personnel from taking up their posts in real time, avoided the risk of job rotation at the source, realized full-chain automated closed-loop management, and ensured the auditability and traceability of system implementation.
Smart Images

Figure CN123118046A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of artificial intelligence and human resource risk management technology, and in particular to a full lifecycle management system and method for key and sensitive positions. Background Technology
[0002] In modern enterprises, key and sensitive positions involve core business processes such as fund management, procurement and bidding, information security, and compliance and risk control. Effective identification, job review, job rotation management, and daily risk supervision of such positions are important aspects of building an enterprise's internal control system.
[0003] Currently, enterprises primarily employ a manual management model driven by institutional documents for the management of key and sensitive positions. Position identification is carried out by each department based on qualitative descriptions; personnel qualification review involves the responsible personnel reviewing files across systems to provide their opinions; job rotation destinations are determined by department heads through consultation based on experience; daily risk monitoring relies on periodic inspections and annual audits for retrospective review; and information is manually transmitted between each link.
[0004] However, as the scale of business operations expands and the frequency of organizational restructuring increases, the information transmission cycle in each link of the above model is further lengthened. The delay in reporting and updating job responsibilities after changes in job duties causes newly added risky positions to remain outside the control list for a long time. Cross-system item-by-item review results in qualification review taking several days while the appointment process has already progressed to subsequent stages. Due to the lack of global information on the business connections between positions, the destination of job rotations can lead to situations where personnel can still access their original resources after rotation. Abnormal patterns in daily behavior can only be identified by waiting for the annual audit, and several months have passed between the occurrence of the abnormality and its discovery.
[0005] Therefore, there is an urgent need for a technology that can solve the problems of subjective job identification, delayed qualification review, blind job rotation planning, lack of risk warning, and difficulty in execution supervision in the full life cycle management of key and sensitive positions. Summary of the Invention
[0006] The technical problem to be solved by the present invention is to provide a full life cycle management system and method for key and sensitive positions, aiming to solve at least one of the above-mentioned technical problems.
[0007] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: Firstly, this application provides a full lifecycle management system for key and sensitive positions, employing the following technical solution: A lifecycle management system for key and sensitive positions, comprising: The job management module is used to evaluate various jobs in the target company based on knowledge graphs and machine learning models, and dynamically update the list of key and sensitive jobs based on the evaluation results. The review module is used to acquire multi-source heterogeneous data, and based on the multi-source heterogeneous data, determine a multi-dimensional compliance profile of the target personnel; when the target personnel are to be appointed to a position in the list of key sensitive positions, the review result is determined based on the multi-dimensional compliance profile of the target personnel and the preset qualification review rules. The job rotation planning module is used to generate job rotation plans for employees who meet the job rotation conditions, based on a quantitative job correlation score matrix, preset job rotation constraints, and optimization algorithms. The job correlation score matrix is determined based on the business process relationship and responsibility similarity between jobs. The early warning and handling module is used to acquire the job behavior characteristics of any key sensitive personnel in real time; calculate an anomaly score based on the historical job behavior characteristic sequence of the key sensitive personnel and a preset anomaly detection model; and trigger a graded early warning based on the anomaly score and execute corresponding handling actions in conjunction with it.
[0008] Based on the above technical solution, the present invention can be further improved as follows.
[0009] Furthermore, it also includes: The job rotation execution supervision and handover module is used to generate a job rotation notification form based on the job rotation plan and push it to relevant personnel; extract target entities from the work documents of the original personnel based on a preset named entity recognition model, and generate a job handover list based on the association between the target entities and the job permission list; and backtrack and verify the service interval of the same person in the same position based on the job rotation history, and generate a warning message when the service interval does not meet the preset conditions.
[0010] Furthermore, it also includes: The system implementation evaluation and accountability auxiliary module is used to calculate and display the implementation evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period; based on the identification of the involved personnel and the time interval of the case, it performs multi-dimensional backtracking queries on the system implementation history of the involved personnel, and generates a responsibility analysis report based on the backtracking query results.
[0011] Furthermore, the job management module includes: The job risk assessment unit is used to score the risk of each job based on a pre-built job risk scoring model to obtain the risk probability value of each job; jobs with risk probability values exceeding a preset threshold are marked as key sensitive jobs, and a list of key sensitive jobs is generated, which includes job identifier, department, risk score and marking basis. The dynamic update unit is used to monitor organizational structure adjustment events of the target enterprise according to a preset period. In response to the monitored events, it incrementally extracts data on changed positions, recalculates the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions based on a pre-built position risk scoring model, and updates the list of key sensitive positions based on the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions.
[0012] Furthermore, the job management module also includes: The job identification unit is used to acquire multi-source raw data for each job position in the target company. The multi-source raw data includes organizational structure data, job descriptions, job responsibility and authority texts, fund approval node data, and procurement and bidding records. Based on natural language processing technology and the multi-source raw data, determine the job entity, permission entity, resource entity, and the relationship between each entity; A knowledge graph is constructed based on the job entity, the permission entity, the resource entity, and the relationships between each entity.
[0013] Furthermore, the job risk assessment unit, when constructing the job risk scoring model, is specifically used for: Obtain historical violation case data, and based on the historical violation case data, determine a sample dataset. The sample dataset includes multiple samples, which are either positive or negative samples. The positive samples represent the positions involved in the historical violation case data, and the negative samples represent the positions not involved in the historical violation case data. Based on the knowledge graph, the adjacency weight, resource density value and business process centrality of the corresponding job node for each sample are determined to form a feature vector. The machine learning model is trained based on the positive samples, the negative samples, and the feature vectors to obtain the job risk scoring model.
[0014] Furthermore, the review module includes: A multi-source data fusion unit is used to acquire multi-source heterogeneous data, including contract employee identification, qualifications, annual performance evaluation level, violation handling records, number of reports, economic responsibility audit conclusions, and training completion status; and to determine a personnel qualification wide table based on the multi-source heterogeneous data, wherein the personnel qualification wide table represents structured personnel data; The personnel compliance profiling unit is used to determine the multi-dimensional compliance profile of each person based on the personnel qualification wide table, the preset sentiment analysis model, the preset time series analysis model and the indicator mapping rules. The multi-dimensional compliance profile includes compliance record score, compliance evaluation index, business capability index and disciplinary impact period status. The real-time qualification verification unit is used to obtain the employee number and target position identifier of the target personnel; query whether the target position is located in the list of key sensitive positions; in response to the target position being located in the list of key sensitive positions, obtain the multi-dimensional compliance profile of the target personnel; perform verification based on the multi-dimensional compliance profile and preset qualification review rules, and determine the review result, which includes review passed or review blocked.
[0015] Furthermore, the personnel compliance profiling unit is specifically used for: The evaluation text is obtained from the personnel qualification wide table, and the evaluation text is analyzed based on the preset sentiment analysis model to obtain a positive sentiment score. Extract the assessment grades for several consecutive years from the personnel qualification wide table, and calculate the assessment trend slope based on the preset time series analysis model; Based on the mapping relationship between the number of reports and the type of punishment, a compliance record score is determined; Based on the positive emotion score and the completion rate of training status, a compliance evaluation index is determined; Based on the slope of the assessment trend and the number of qualifications, the business capability index is determined; The status of the period of influence of the punishment is determined by comparing the end time of the violation handling record with the current system time. Based on the compliance record score, the compliance evaluation index, the business capability index, and the status of the disciplinary impact period, a personnel compliance profile is generated.
[0016] Furthermore, the job rotation planning module includes: The job relevance calculation unit is used to obtain the co-occurrence frequency of business processes and the responsibility text among various jobs; calculate the process relevance score based on the business process co-occurrence frequency; calculate the responsibility similarity score based on the semantic vector of the responsibility text; and determine the relevance score among various jobs and generate a job relevance score matrix based on the process relevance score and the responsibility similarity score. The job rotation plan generation unit is used to obtain a list of personnel to be rotated, which includes personnel who have served continuously in their current key and sensitive positions for a preset number of years; determine a set of candidate target positions for each personnel to be rotated based on the job relevance score matrix; and determine the optimal target position from the set of candidate target positions based on preset job rotation constraints and an optimization algorithm to generate a job rotation plan; wherein the preset job rotation constraints include: the same person cannot rotate back to their current position within a set time interval, the number of people rotating in the same department at the same time does not exceed a preset upper limit of proportion, and the relevance score of the target position for rotation does not exceed a preset relevance threshold.
[0017] Furthermore, the optimization algorithm used in the job rotation scheme generation unit is a genetic algorithm, and the objective function of the genetic algorithm is to maximize the sum of the risk hedging index and the talent training value index of all personnel to be rotated.
[0018] Furthermore, the job rotation planning module also includes: The deferred job rotation identification unit is used to send a batch query request to the human resources system before generating the job rotation plan to obtain the expected retirement date and leave status of each employee in the list of employees to be rotated. Individuals whose expected retirement date differs from their current date by less than a preset number of days are marked as those nearing retirement. Remove the personnel who are about to retire and those who are on leave from the list of personnel awaiting rotation and add them to the list of personnel whose rotation is postponed. The list of personnel whose rotation is postponed includes personnel identification and postponement reason code. The list of personnel to be rotated after being removed is sent to the rotation plan generation unit.
[0019] Furthermore, the early warning and handling module includes: The behavioral feature acquisition unit is used to acquire approval action data and financial correlation data of personnel in key and sensitive positions in real time; extract approval action features based on the approval action data; extract financial correlation features based on the financial correlation data; and generate job behavioral features of personnel in key and sensitive positions based on the approval action features and the financial correlation features. An anomaly detection and graded early warning unit is used to arrange multiple job behavior characteristics of key sensitive personnel within a preset time window in chronological order to obtain a historical job behavior characteristic sequence; input the historical job behavior characteristic sequence into a preset anomaly detection model to obtain an anomaly score; wherein the preset anomaly detection model is a time-series anomaly detection model based on an autoencoder; and determine an early warning level based on the anomaly score and preset multi-level thresholds, wherein the early warning level includes a first level, a second level, and a third level; wherein, when the anomaly score is greater than the first threshold and less than or equal to the second threshold, it is determined to be at the first level; when the anomaly score is greater than the second threshold and less than or equal to the third threshold, it is determined to be at the second level; and when the anomaly score is greater than the third threshold, it is determined to be at the third level. The response and coordination unit is used to execute corresponding response actions based on the warning level.
[0020] Furthermore, the early warning and handling module also includes: The conversation assistance unit is used to monitor personnel recruitment events, job transfer events, and first-level warning events; in response to the monitored events, it extracts risk point tags from the multi-dimensional compliance profiles of the personnel in the key and sensitive positions; based on the preset mapping rules between risk point tags and reminder content, it determines the key reminder content; based on the key reminder content, it calls the template engine to generate a conversation prompt sheet; it monitors the upload status of the conversation records corresponding to the conversation prompt sheet, and if the conversation records are not uploaded within a preset time limit, it upgrades the first-level warning to the second-level warning; The job transfer trigger unit is used to acquire Level 3 warning events; in response to the acquisition of Level 3 warning events, it sends an account freeze command to the access control platform to freeze all job-related system accounts of the personnel in the key sensitive positions.
[0021] Furthermore, the rotation execution supervision and handover module includes: The job rotation task process engine is used to receive the job rotation plan output by the job rotation planning module, instantiate a process task for each job rotation action in the job rotation plan, call the electronic signature service, and generate a job rotation notification based on the personnel information, original position, new position and reporting deadline in the process task; and push the job rotation notification to the job rotation personnel and department heads through multiple set channels. The intelligent handover list generation unit is used to obtain work order content, email body, and project documents from the original personnel's to-do list; based on a named entity recognition model using natural language processing, it extracts target entities from the work order content, email body, and project documents. The target entities include supplier name, contract number, process instance identifier in approval, and document number; it associates the target entities with the original personnel's job permission list to generate a structured job handover list. The job handover list displays the hierarchical relationship of responsibilities, to-do items, key contacts, and confidential assets in a tree structure; it receives confirmation operations from the original personnel and the rotating personnel on the job handover list, and records the operator, timestamp, and client address corresponding to each confirmation operation; The job qualification retrospective verification unit is used to scan the job rotation history record table according to a preset cycle to check whether there are two job records for the same personnel ID and the same job ID, and whether the interval between the end time of the previous job and the start time of the next job is less than a preset return interval threshold. If such a situation exists, an early warning information is pushed to the management dashboards of the audit department and the organization and personnel department. The early warning information includes the personnel name, the job involved, the time of the violation return, and the information of the responsible person.
[0022] Furthermore, the system implementation evaluation and accountability support module includes: The execution indicator monitoring dashboard is used to aggregate and calculate the execution evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period. The execution evaluation indicators include job rotation rate, job rotation timeliness rate, conversation completion rate, and handover list completeness rate. The execution evaluation indicators and overall ranking of each department are displayed in the form of visual charts. The violation incident tracing unit is used to obtain the identification of the personnel involved and the time range of the case; to perform a retrospective query on the personnel involved based on the case time range and obtain the retrospective query results; and to generate a responsibility analysis report based on the retrospective query results. The retrospective query includes: obtaining all job rotation records of the personnel involved within the case time range and determining whether there are any cases of failure to rotate jobs within the time limit; obtaining the qualification review log of the personnel involved and determining whether there are any records of failure to block violations that should have been blocked; and obtaining the risk warning history of the personnel involved and determining whether there are any warning records that have been triggered but not handled.
[0023] Secondly, this application provides a method for full lifecycle management of key and sensitive positions, employing the following technical solution: A method for full lifecycle management of key and sensitive positions includes: Based on knowledge graphs and machine learning models, the various positions in the target company are evaluated, and the list of key and sensitive positions is dynamically updated based on the evaluation results. Acquire multi-source heterogeneous data, and based on the multi-source heterogeneous data, determine a multi-dimensional compliance profile of the target personnel; when the target personnel are to be appointed to a position in the list of key sensitive positions, determine the review result based on the multi-dimensional compliance profile of the target personnel and the preset qualification review rules; Based on a quantitative job relevance score matrix, preset job rotation constraints, and optimization algorithms, a job rotation plan is generated for employees who meet the job rotation conditions. The job relevance score matrix is determined based on the business process relationship and responsibility similarity between jobs. For any key and sensitive personnel, the job behavior characteristics of the key and sensitive personnel are acquired in real time; an anomaly score is calculated based on the historical job behavior characteristic sequence of the key and sensitive personnel and the preset anomaly detection model; and a graded early warning is triggered based on the anomaly score and corresponding handling actions are executed in conjunction.
[0024] Thirdly, this application provides an electronic device that adopts the following technical solution: An electronic device includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in the second aspect, a method for full lifecycle management of critical and sensitive positions.
[0025] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in the second aspect, a method for full lifecycle management of critical and sensitive positions.
[0026] In summary, the beneficial effects of this invention are: The job management module, based on knowledge graphs and machine learning models, objectively and quantitatively evaluates jobs and dynamically updates the list of key and sensitive positions, solving the problems of strong subjectivity and easy omissions and errors in traditional manual judgment. The review module integrates multi-source heterogeneous data to construct multi-dimensional compliance profiles and verifies them in real time during the appointment process, transforming personnel qualification review from a static and lagging mode to a dynamic and real-time mode, effectively preventing unqualified personnel from taking up their posts. The job rotation planning module generates rotation plans based on a quantified job relevance score matrix and optimization algorithms, transforming the fuzzy concept of closely related positions into a calculable relevance score, thus addressing the issue from the source. This invention avoids the secondary risk of increased risk after job rotation; it uses an early warning and response module based on real-time behavioral characteristic sequences and anomaly detection models to provide tiered early warnings and coordinated responses, shifting risk prevention from reactive post-event handling to proactive pre-event prevention; and it utilizes a job rotation execution supervision and handover module to achieve electronic signatures, automatic generation of intelligent handover lists, and backflow verification, ensuring that the job rotation system is implemented from paper regulations into an auditable and traceable execution process; and it uses a system execution evaluation and accountability support module to quantify system execution into visual indicators and support the tracing of responsibility for violations, providing objective and tamper-proof data support for accountability. This invention achieves fully automated closed-loop management of the entire chain, from job identification, qualification review, job rotation planning, execution supervision, risk early warning to accountability evaluation, significantly improving the scientific, real-time, and penetrating nature of key and sensitive job management. Attached Figure Description
[0027] Figure 1 This is a schematic diagram of the structure of a full lifecycle management system for key and sensitive positions provided in one embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of another key and sensitive position full lifecycle management system provided in one embodiment of the present invention; Figure 3 A flowchart illustrating a method for full lifecycle management of key and sensitive positions, provided as an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present invention. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0029] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.
[0030] like Figure 1 As shown, this application provides a full lifecycle management system for key and sensitive positions. The system is deployed in an enterprise intranet environment and interacts with human resources, supervision and management, auditing, office automation (OA) and financial systems through data interfaces. The system adopts a microservice architecture, with each module encapsulated as an independent microservice and unified routing and load balancing through an API gateway.
[0031] The system mainly includes: The job management module is used to evaluate various jobs in the target company based on knowledge graphs and machine learning models, and dynamically update the list of key and sensitive jobs based on the evaluation results. The review module is used to acquire multi-source heterogeneous data, and based on the multi-source heterogeneous data, determine a multi-dimensional compliance profile of the target personnel; when the target personnel are to be appointed to a position in the list of key sensitive positions, the review result is determined based on the multi-dimensional compliance profile of the target personnel and the preset qualification review rules. The job rotation planning module is used to generate job rotation plans for employees who meet the job rotation conditions, based on a quantitative job correlation score matrix, preset job rotation constraints, and optimization algorithms. The job correlation score matrix is determined based on the business process relationship and responsibility similarity between jobs. The early warning and handling module is used to acquire the job behavior characteristics of any key sensitive personnel in real time; calculate an anomaly score based on the historical job behavior characteristic sequence of the key sensitive personnel and a preset anomaly detection model; and trigger a graded early warning based on the anomaly score and execute corresponding handling actions in conjunction with it.
[0032] Optional, the job management module includes: The job identification unit is used to acquire multi-source raw data of each job in the target enterprise. The multi-source raw data includes organizational structure data, job descriptions, job responsibility and authority texts, fund approval node data, and procurement and bidding records. Based on natural language processing technology and the multi-source raw data, the unit determines job entities, authority entities, resource entities, and the relationships between each entity. Based on the job entities, authority entities, resource entities, and the relationships between each entity, a knowledge graph is constructed.
[0033] The job risk assessment unit is used to score the risk of each job based on a pre-built job risk scoring model to obtain the risk probability value of each job; jobs with risk probability values exceeding a preset threshold are marked as key sensitive jobs, and a list of key sensitive jobs is generated, which includes job identifier, department, risk score and marking basis. The dynamic update unit is used to monitor organizational structure adjustment events of the target enterprise according to a preset period. In response to the monitored events, it incrementally extracts data on changed positions, recalculates the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions based on a pre-built position risk scoring model, and updates the list of key sensitive positions based on the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions.
[0034] In this embodiment, the job identification unit first extracts organizational structure data from the OA system to obtain departmental hierarchy and job setting information; extracts job descriptions and responsibility and authority texts from the human resources system to obtain specific job descriptions and authority scopes for each job; extracts fund approval node data from the financial system to obtain the position and approval amount of each job in the fund approval process; and extracts procurement bidding participation records from the procurement system to obtain the role and participation frequency of each job in the procurement process.
[0035] Then, using dependency parsing techniques from natural language processing, we extract job entity, permission entity, and resource entity, as well as the binary relations between them, from the above text. For example, from "the purchasing position has the initial review authority for contracts worth over 500,000 yuan," we extract the triple (purchasing position, initial review authority, 500,000 yuan). Using a graph database as a carrier, we construct a knowledge graph containing job nodes, permission nodes, resource nodes, and edges representing upstream and downstream relationships in the business flow. The attribute records of the edges represent quantifiable values of resource-intensive metrics.
[0036] In this embodiment, the job risk assessment unit constructs a feature vector based on the adjacency weights, resource density values, and business process betweenness centrality of each job node in the knowledge graph. It then uses a pre-trained machine learning model (such as an XGBoost binary classifier) to score the risk of each job, outputting a risk probability value between 0 and 1. Jobs with risk probability values exceeding a preset threshold (e.g., 0.7) are automatically marked as key sensitive jobs, generating a list of key sensitive jobs.
[0037] Specifically, the job risk assessment unit, when used to construct a job risk scoring model, is used for: Obtain historical violation case data, and based on the historical violation case data, determine a sample dataset. The sample dataset includes multiple samples, which are either positive or negative samples. The positive samples represent the positions involved in the historical violation case data, and the negative samples represent the positions not involved in the historical violation case data. Based on the knowledge graph, the adjacency weight, resource density value and business process centrality of the corresponding job node for each sample are determined to form a feature vector. The machine learning model is trained based on the positive samples, the negative samples, and the feature vectors to obtain the job risk scoring model.
[0038] The job risk assessment unit extracts violation case data from the past five years from the supervision and management system and the audit system. Jobs involved in the cases are used as positive samples, and jobs without violations are used as negative samples, forming a sample dataset. For each sample, the adjacency weights, resource density values, and business process betweenness centrality of the corresponding job node are extracted from the knowledge graph to form a feature vector. The XGBoost algorithm is used to train the sample dataset, with a maximum tree depth of 6 and a learning rate of 0.3, for 100 training rounds to obtain the job risk scoring model.
[0039] In this embodiment of the application, the dynamic update unit listens for organizational structure adjustment events of the target enterprise according to a preset period (such as every 24 months), and at the same time listens for organizational structure change events (such as department mergers, new business system launches, etc.) on the enterprise service bus in real time.
[0040] The dynamic update unit is configured with a timed trigger to automatically initiate a full-scale job risk assessment process every 24 months, reassessing the risk of all jobs. Simultaneously, the dynamic update unit continuously monitors the message queue on the enterprise service bus, capturing organizational structure adjustment events. In response to detected events, the dynamic update unit incrementally extracts relevant data for changed jobs (including the job descriptions of newly added jobs, adjusted approval permissions, etc.), updates the information of corresponding nodes and edges in the knowledge graph, and then recalculates the risk probability value of the affected jobs based on the job risk assessment model. Because there are upstream and downstream business process relationships between jobs, a change in one job may affect the risk level of its related jobs; therefore, the dynamic update unit also recalculates the risk probability values of the upstream and downstream related jobs of the affected job. Based on the recalculation results, the list of key sensitive jobs is updated, adding newly added high-risk jobs to the list and removing jobs whose risk has been reduced to below the threshold. A pending confirmation request for job list update is pushed to the system administrator, awaiting manual confirmation before it officially takes effect.
[0041] Optionally, the review module includes: A multi-source data fusion unit is used to acquire multi-source heterogeneous data, including contract employee identification, qualifications, annual performance evaluation level, violation handling records, number of reports, economic responsibility audit conclusions, and training completion status; and to determine a personnel qualification wide table based on the multi-source heterogeneous data, wherein the personnel qualification wide table represents structured personnel data; The personnel compliance profiling unit is used to determine the multi-dimensional compliance profile of each person based on the personnel qualification wide table, the preset sentiment analysis model, the preset time series analysis model and the indicator mapping rules. The multi-dimensional compliance profile includes compliance record score, compliance evaluation index, business capability index and disciplinary impact period status. The real-time qualification verification unit is used to obtain the employee number and target position identifier of the target personnel; query whether the target position is located in the list of key sensitive positions; in response to the target position being located in the list of key sensitive positions, obtain the multi-dimensional compliance profile of the target personnel; perform verification based on the multi-dimensional compliance profile and preset qualification review rules, and determine the review result, which includes review passed or review blocked.
[0042] In this embodiment, the multi-source data fusion unit uses an ETL tool to pull incremental data from multiple source systems according to a preset synchronization cycle (e.g., 1 AM daily). It synchronizes contract employee identification, professional qualification certificates, and annual performance ratings (Excellent, Good, Competent, Needs Improvement) from the human resources system; records of violations (including disciplinary type, start and end times) and number of reports from the supervision and management system; economic responsibility audit conclusions (no abnormalities, warnings, accountability) from the audit system; and training completion status from the training system. Using employee ID as the primary key, the above data is written into a wide table of personnel qualifications in a distributed data warehouse, achieving unified storage and management of multi-source heterogeneous data.
[0043] In this embodiment, the sentiment analysis model refers to a model fine-tuned from a pre-trained language model (such as the BERT Chinese pre-trained model) for scoring the sentiment tendency of text. In this embodiment, it is used to calculate positive sentiment scores for evaluation text data. The time series analysis model refers to a model for trend analysis based on time series data. In this embodiment, a linear regression model is used to calculate the trend slope of assessment levels over several consecutive years as an indicator of business capability stability. The indicator mapping rule refers to the correspondence used when converting raw data into a profile index, such as the mapping relationship between the number of reports and disciplinary types to compliance record scores.
[0044] The personnel compliance profiling unit is specifically used for: The evaluation text is obtained from the personnel qualification wide table, and the evaluation text is analyzed based on the preset sentiment analysis model to obtain a positive sentiment score. The assessment grades for several consecutive years are extracted from the personnel qualification table, and the assessment trend slope is calculated based on the preset time series analysis model. A positive slope indicates an improvement trend in ability, and a negative slope indicates a decline trend in ability. Based on the mapping relationship between the number of reports and the type of punishment, a compliance record score is determined. For example, no reports and no punishment are mapped to 100 points. Each additional report deducts a certain number of points. If there is a record of violation, the score is further deducted according to the severity of the punishment. Based on the positive affect score and the completion rate of training status, a compliance evaluation index is determined, for example, the positive affect score accounts for 0.6 and the compliance training completion rate accounts for 0.4. Based on the slope of the assessment trend and the number of qualifications, a business capability index is determined. For example, the normalized slope of the assessment trend accounts for 0.7 and the normalized number of qualifications accounts for 0.3. Based on the comparison between the end time of the violation processing record and the current system time, the status of the disciplinary impact period is determined. If the current system time is still within the disciplinary impact period, the status is true (Boolean value True), otherwise it is false (Boolean value False). Based on the compliance record score, the compliance evaluation index, the business capability index, and the status of the disciplinary impact period, a personnel compliance profile is generated.
[0045] In this embodiment of the application, when the human resources department initiates a job appointment process in the system, the process engine calls the REST API interface of the real-time qualification verification unit, passes in the target employee's employee number and the target job identifier, and the real-time qualification verification unit performs the following verification actions: First, check if the target position is on the list of key sensitive positions; second, if the target position is on the list of key sensitive positions, obtain the multi-dimensional compliance profile of the target personnel; third, perform comparison and verification based on the preset qualification review rules; if the disciplinary impact period status is true (i.e., the personnel are still within the disciplinary impact period), or the business capability index is below 40 points, or the compliance evaluation index is below 60 points, then return the review blocking status and specific reasons, and generate a qualification review failure report to push to the initiator; if none of the above conditions are met, then return the review passed status.
[0046] Optionally, the job rotation planning module includes: The job relevance calculation unit is used to obtain the co-occurrence frequency of business processes and the responsibility text among various jobs; calculate the process relevance score based on the business process co-occurrence frequency; calculate the responsibility similarity score based on the semantic vector of the responsibility text; and determine the relevance score among various jobs and generate a job relevance score matrix based on the process relevance score and the responsibility similarity score. The job rotation plan generation unit is used to obtain a list of personnel to be rotated, which includes personnel who have served continuously in their current key and sensitive positions for a preset number of years; determine a set of candidate target positions for each personnel to be rotated based on the job relevance score matrix; and determine the optimal target position from the set of candidate target positions based on preset job rotation constraints and an optimization algorithm to generate a job rotation plan; wherein the preset job rotation constraints include: the same person cannot rotate back to a position already held within a set time interval, the number of people rotating in the same department at the same time does not exceed a preset upper limit, and the relevance score of the target position for rotation is not lower than a preset minimum threshold.
[0047] In this embodiment, the job relevance calculation unit extracts approval flow path data from the business process management system over the past year, counts the frequency of occurrence of any two jobs as upstream and downstream nodes, normalizes the frequency, and obtains a process relevance score. Simultaneously, TF-IDF feature extraction is performed on the job description text of each job, and the cosine similarity of the TF-IDF vectors of any two job description texts is calculated to obtain a job similarity score. The process relevance score and the job similarity score are weighted and summed according to preset weights to obtain the relevance score between each job, which is stored as an N×N two-dimensional array as a job relevance score matrix, where N is the total number of jobs. For example, if the process relevance score between the purchasing and acceptance jobs is 0.8 and the job similarity score is 0.6, then their relevance score is 0.8×0.6+0.6×0.4=0.72, indicating that these two jobs have a high relevance and are not suitable for direct rotation.
[0048] In this embodiment, the job rotation plan generation unit first obtains a list of personnel who have served continuously in their current key and sensitive positions for 5 years and 11 months from the human resources system as a list of personnel to be rotated (i.e., the set P of personnel to be rotated). Based on the job relevance score matrix, for each personnel to be rotated, positions with a relevance score higher than 0.7 are excluded, and a set of candidate target positions is determined.
[0049] The rotation system is constrained as follows: Constraint 1: The same employee cannot be rotated back to their original position within two years; Constraint 2: The number of employees rotating within the same department during the same period cannot exceed a preset maximum ratio; Constraint 3: The rotated position cannot be a position whose correlation score with the original position is higher than a preset correlation threshold of 0.7. A genetic algorithm is used to find the optimal rotation scheme: Each individual is encoded as a matching mapping from the set of employees to be rotated, P, to the set of candidate target positions. The initial population size is set to 200, the crossover probability is 0.8, and the mutation probability is 0.05. The objective function is to maximize the sum of the risk hedging index (i.e., the sum of the reciprocals of the correlation scores between the original and new positions) and the talent development value index (i.e., the increase in the job complexity of the new position) after all employees are rotated. A larger value indicates a better risk isolation effect after rotation. During the constraint check, individuals that do not meet any constraint are marked as infeasible solutions and assigned a penalty value. After 500 iterations, the optimal solution set is output. Each solution corresponds to a job rotation plan, including personnel identification, original position, target position, expected effective date, and risk hedging index. The genetic algorithm runs on the server CPU and generates feasible solutions for a problem involving 50 rotating personnel and 200 available positions in an average of 45 seconds.
[0050] The job rotation planning module also includes: The deferred job rotation identification unit is used to send a batch query request to the human resources system before generating the job rotation plan to obtain the expected retirement date and leave status of each employee in the list of employees to be rotated. Individuals whose expected retirement date differs from their current date by less than a preset number of days are marked as those nearing retirement. Remove the personnel who are about to retire and those who are on leave from the list of personnel awaiting rotation and add them to the list of personnel whose rotation is postponed. The list of personnel whose rotation is postponed includes personnel identification and postponement reason code. The list of personnel to be rotated after being removed is sent to the rotation plan generation unit.
[0051] In this embodiment, the deferred job rotation identification unit is invoked before the job rotation plan generation unit executes. This unit sends a batch query request to the human resources system to obtain the expected retirement date and current leave status of each person in the list of personnel awaiting job rotation. Personnel whose expected retirement date differs from their current date by less than 365 days are marked as nearing retirement; personnel currently pregnant, breastfeeding, or on maternity leave are marked as on leave. The deferred job rotation identification unit removes these personnel from the list of personnel awaiting job rotation, adds them to the deferred job rotation list, and sends the removed list to the job rotation plan generation unit for further processing. The deferred job rotation list is reassessed periodically (e.g., monthly), and when the reason for the deferred period is eliminated (e.g., the end of leave), the relevant personnel are automatically reinstated to the list of personnel awaiting job rotation.
[0052] Optionally, the early warning and handling module includes: The behavioral feature acquisition unit is used to acquire approval action data and financial correlation data of personnel in key and sensitive positions in real time; extract approval action features based on the approval action data; extract financial correlation features based on the financial correlation data; and generate job behavioral features of personnel in key and sensitive positions based on the approval action features and the financial correlation features. An anomaly detection and graded early warning unit is used to arrange multiple job behavior characteristics of key sensitive personnel within a preset time window in chronological order to obtain a historical job behavior characteristic sequence; input the historical job behavior characteristic sequence into a preset anomaly detection model to obtain an anomaly score; wherein the preset anomaly detection model is a time-series anomaly detection model based on an autoencoder; and determine an early warning level based on the anomaly score and preset multi-level thresholds, wherein the early warning level includes a first level, a second level, and a third level; wherein, when the anomaly score is greater than the first threshold and less than or equal to the second threshold, it is determined to be at the first level; when the anomaly score is greater than the second threshold and less than or equal to the third threshold, it is determined to be at the second level; and when the anomaly score is greater than the third threshold, it is determined to be at the third level. The response and coordination unit is used to execute corresponding response actions based on the warning level.
[0053] In this embodiment, the behavior feature collection unit sets up an interceptor on the approval flow interface of the OA system to capture every approval action of key sensitive personnel in real time. The behavior feature collection unit extracts the following approval action features: first, violation of approval path features, detecting whether the approval action bypasses the established approval process nodes; second, abnormal approval time features, calculating the difference between the current approval time and the historical average time for this type of approval, and marking it as abnormal if it is less than two standard deviations below the average time. Simultaneously, the behavior feature collection unit obtains supplier information from expense reports submitted or approved by key sensitive personnel from the financial system interface, performs fuzzy matching with the list of stakeholders maintained in the supervision and management system, and generates a financial association feature tag if a match is successful. The above approval action features and financial association features together constitute the behavior features of the key sensitive personnel. All behavior features are written to a Kafka message queue in the form of an event stream, consumed by the Flink stream processing engine, and stored in the time-series database InfluxDB, forming behavior feature data organized by time series.
[0054] In this embodiment, the anomaly detection and graded early warning unit reads all behavioral characteristic records of each key sensitive personnel from a time-series database for the past 90 days, arranging them chronologically to form a historical behavioral characteristic sequence. This sequence is input into a time-series anomaly detection model based on an autoencoder. The model encodes and decodes the input sequence to reconstruct it, calculating the reconstruction error as an anomaly score. The score ranges from 0 to 1, with higher values indicating greater deviation from the normal baseline. The anomaly detection and graded early warning unit graded the anomaly score based on preset multi-level thresholds: a score greater than 0.6 and less than or equal to 0.75 is classified as a first-level warning; a score greater than 0.75 and less than or equal to 0.9 is classified as a second-level warning; and a score greater than 0.9 is classified as a third-level warning. The warning event includes personnel identification, warning level, a description of the abnormal behavior segment, and an associated timestamp. The warning message is pushed to the front-end management dashboard in real time via WebSocket. The anomaly detection model has a hidden layer dimension of 64 in the autoencoder. During training, normal behavior data from the past 90 days is used as the training set, and offline retraining is performed every 7 days to adapt to natural changes in behavior patterns.
[0055] The early warning and handling module also includes: The conversation assistance unit is used to monitor personnel recruitment events, job transfer events, and first-level warning events; in response to the monitored events, it extracts risk point tags from the multi-dimensional compliance profiles of the personnel in the key and sensitive positions; based on the preset mapping rules between risk point tags and reminder content, it determines the key reminder content; based on the key reminder content, it calls the template engine to generate a conversation prompt sheet; it monitors the upload status of the conversation records corresponding to the conversation prompt sheet, and if the conversation records are not uploaded within a preset time limit, it upgrades the first-level warning to the second-level warning; The job transfer trigger unit is used to acquire Level 3 warning events; in response to the acquisition of Level 3 warning events, it sends an account freeze command to the access control platform to freeze all job-related system accounts of the personnel in the key sensitive positions.
[0056] As a further optional implementation of the embodiments of this application, such as Figure 2 As shown, a full lifecycle management system for key and sensitive positions also includes: The job rotation execution supervision and handover module is used to generate a job rotation notification form based on the job rotation plan and push it to relevant personnel; extract target entities from the work documents of the original personnel based on a preset named entity recognition model, and generate a job handover list based on the association between the target entities and the job permission list; and perform backtracking verification on the service interval of the same personnel in the same position based on the job rotation history, and generate a warning message when the service interval does not meet the preset conditions. Specifically, the job rotation execution supervision and handover module includes: The job rotation task process engine is used to receive the job rotation plan output by the job rotation planning module, instantiate a process task for each job rotation action in the job rotation plan, call the electronic signature service, and generate a job rotation notification based on the personnel information, original position, new position and reporting deadline in the process task; and push the job rotation notification to the job rotation personnel and department heads through multiple set channels. The intelligent handover list generation unit is used to obtain work order content, email body, and project documents from the original personnel's to-do list; based on a named entity recognition model using natural language processing, it extracts target entities from the work order content, email body, and project documents. The target entities include supplier name, contract number, process instance identifier in approval, and document number; it associates the target entities with the original personnel's job permission list to generate a structured job handover list. The job handover list displays the hierarchical relationship of responsibilities, to-do items, key contacts, and confidential assets in a tree structure; it receives confirmation operations from the original personnel and the rotating personnel on the job handover list, and records the operator, timestamp, and client address corresponding to each confirmation operation; The job qualification retrospective verification unit is used to scan the job rotation history record table according to a preset cycle to check whether there are two job records for the same personnel ID and the same job ID, and whether the interval between the end time of the previous job and the start time of the next job is less than a preset return interval threshold. If such a situation exists, an early warning information is pushed to the management dashboards of the audit department and the organization and personnel department. The early warning information includes the personnel name, the job involved, the time of the violation return, and the information of the responsible person.
[0057] As a further optional implementation of the embodiments of this application, such as Figure 2 As shown, a full lifecycle management system for key and sensitive positions also includes: The system implementation evaluation and accountability auxiliary module is used to calculate and display the implementation evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period; based on the identification of the involved personnel and the time interval of the case, it performs multi-dimensional backtracking queries on the system implementation history of the involved personnel, and generates a responsibility analysis report based on the backtracking query results.
[0058] The system implementation evaluation and accountability support module includes: The execution indicator monitoring dashboard is used to aggregate and calculate the execution evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period. The execution evaluation indicators include job rotation rate, job rotation timeliness rate, conversation completion rate, and handover list completeness rate. The execution evaluation indicators and overall ranking of each department are displayed in the form of visual charts. The violation incident tracing unit is used to obtain the identification of the personnel involved and the time range of the case; to perform a retrospective query on the personnel involved based on the case time range and obtain the retrospective query results; and to generate a responsibility analysis report based on the retrospective query results. The retrospective query includes: obtaining all job rotation records of the personnel involved within the case time range and determining whether there are any cases of failure to rotate jobs within the time limit; obtaining the qualification review log of the personnel involved and determining whether there are any records of failure to block violations that should have been blocked; and obtaining the risk warning history of the personnel involved and determining whether there are any warning records that have been triggered but not handled.
[0059] This invention provides a full lifecycle management system for critical and sensitive positions. Upon startup, the position management module first executes the position identification and assessment process. The position identification unit extracts multi-source raw data for each position from the OA system, human resources system, financial system, and procurement system, and uses natural language processing technology to construct a knowledge graph containing position nodes, permission nodes, and resource nodes. The position risk assessment unit, based on the feature vectors of each position node in the knowledge graph, uses the XGBoost model to score the risk of all positions, marking positions with a risk probability value exceeding 0.7 as critical and sensitive positions, generating a list of critical and sensitive positions. The dynamic update unit continuously monitors organizational structure adjustment events to ensure the timeliness of the list.
[0060] In daily personnel management, the review module is triggered when the organization's HR department initiates an appointment process for key and sensitive positions. The multi-source data fusion unit synchronizes personnel data from various source systems to the personnel qualification wide table via daily scheduled ETL tasks. Based on the data in the personnel qualification wide table, the personnel compliance profiling unit, combined with sentiment analysis and time series analysis models, has pre-calculated and stored multi-dimensional compliance profiles for each person. The real-time qualification verification unit receives the target employee's employee number and target position identifier from the appointment process engine, queries whether the target position is on the list of key and sensitive positions, and if so, obtains the target employee's multi-dimensional compliance profile for rule verification, returning a review pass or review block result, achieving second-level qualification review.
[0061] When the rotation cycle arrives, the rotation planning module is triggered. The deferred rotation identification unit first filters out employees nearing retirement and those on leave and adds them to the deferred rotation list. The job relevance calculation unit calculates the job relevance score matrix based on approval flow path data and job description text. The rotation plan generation unit determines the candidate target job set based on the job relevance score matrix and uses a genetic algorithm to generate the optimal rotation plan while satisfying the rotation constraints.
[0062] After the job rotation plan is generated, the job rotation execution supervision and handover module takes over the execution process. The job rotation task workflow engine instantiates workflow tasks for each job rotation action, generates a job rotation notification form through electronic signature service, and pushes it to the rotating personnel and department heads. The intelligent handover list generation unit extracts target entities from the original personnel's work documents based on the named entity recognition model, associates them with the job permission list, and generates a structured job handover list. Both parties confirm each item in the system, and the system records the operator, timestamp, and client IP address for each confirmation. The qualification backtracking verification unit scans the job rotation history table daily to detect any unauthorized return to the job and generates early warning information.
[0063] The early warning and handling module continues to operate after the rotating staff arrive at their posts and during daily supervision. The behavioral feature acquisition unit captures approval action data and financial correlation data of key and sensitive personnel in real time, extracts approval action features and financial correlation features, and writes them into the time series database. The anomaly detection and graded early warning unit performs anomaly detection every minute on the behavioral feature sequence of each key and sensitive personnel over the past 90 days, calculates anomaly scores, and issues early warning levels based on multi-level thresholds.
[0064] Throughout the entire process described above, the system implementation evaluation and accountability support module runs continuously in the background. The implementation indicator monitoring dashboard aggregates operation logs from each module every 24 hours, calculating implementation evaluation indicators such as rotation rate, timely rotation rate, completion rate of interviews, and completeness rate of handover lists, displaying departmental rankings in visual charts. When violations of discipline or law occur, the violation tracing unit conducts multi-dimensional backtracking queries based on the identification of involved personnel and the time frame of the case, automatically generating a "Responsibility Analysis Report" to identify the broken links in system implementation and the responsible persons, providing evidentiary support for accountability.
[0065] Figure 3 A flowchart illustrating a method for full lifecycle management of key and sensitive positions is shown.
[0066] like Figure 3 As shown, a method for full lifecycle management of key and sensitive positions includes: S1, based on knowledge graphs and machine learning models, evaluates various positions in the target company and dynamically updates the list of key and sensitive positions based on the evaluation results. S2, acquire multi-source heterogeneous data, and determine a multi-dimensional compliance profile of the target personnel based on the multi-source heterogeneous data; when the target personnel are to be appointed to a position in the list of key sensitive positions, determine the review result based on the multi-dimensional compliance profile of the target personnel and the preset qualification review rules; S3, based on the quantified job relevance score matrix, preset job rotation constraints and optimization algorithm, generates a job rotation plan for employees who meet the job rotation conditions. The job relevance score matrix is determined based on the business process relationship and responsibility similarity between jobs. S4. For any key sensitive personnel, acquire the job behavior characteristics of the key sensitive personnel in real time; calculate the anomaly score based on the historical job behavior characteristic sequence of the key sensitive personnel and the preset anomaly detection model; trigger a graded early warning based on the anomaly score and execute the corresponding handling actions in conjunction.
[0067] Figure 4 This is a structural block diagram of an electronic device 300 according to an embodiment of this application.
[0068] like Figure 4As shown, the electronic device 300 includes a processor 301 and a memory 302, and may further include one or more of an information input / output (I / O) interface 303, a communication component 304, and a communication bus 305.
[0069] The processor 301 controls the overall operation of the electronic device 300 to complete all or part of the steps in the aforementioned method for full lifecycle management of critical and sensitive positions. The memory 302 stores various types of data to support the operation of the electronic device 300. This data may include, for example, instructions for any application or method operating on the electronic device 300, as well as application-related data. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as one or more of Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0070] I / O interface 303 provides an interface between processor 301 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 304 is used to test wired or wireless communication between electronic device 300 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 304 may include a Wi-Fi component, a Bluetooth component, and an NFC component.
[0071] The communication bus 305 may include a path for transmitting information between the aforementioned components. The communication bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 305 may be divided into an address bus, a data bus, a control bus, etc.
[0072] The electronic device 300 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute a key sensitive position full lifecycle management method given in the above embodiments.
[0073] The following describes the computer-readable storage medium provided in the embodiments of this application. The computer-readable storage medium described below can be referred to in correspondence with the method for full life cycle management of key and sensitive positions described above.
[0074] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for full lifecycle management of key and sensitive positions.
[0075] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0076] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0077] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing application concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions claimed in this application.
Claims
1. A critical sensitive post life cycle management system, characterized in that, include: The job management module is used to evaluate various jobs in the target company based on knowledge graphs and machine learning models, and dynamically update the list of key and sensitive jobs based on the evaluation results. The review module is used to acquire multi-source heterogeneous data and, based on this data, to determine a multi-dimensional compliance profile of the target personnel. When the target personnel are to be appointed to a position in the list of key and sensitive positions, the review result is determined based on the target personnel's multi-dimensional compliance profile and the preset qualification review rules; The job rotation planning module is used to generate job rotation plans for employees who meet the job rotation conditions, based on a quantitative job correlation score matrix, preset job rotation constraints, and optimization algorithms. The job correlation score matrix is determined based on the business process relationship and responsibility similarity between jobs. The early warning and response module is used to obtain the job behavior characteristics of any key and sensitive personnel in real time. Anomaly scores are calculated based on the historical behavioral characteristic sequences of personnel in key and sensitive positions and a preset anomaly detection model. Based on the anomaly scores, tiered early warnings are triggered and corresponding action actions are executed in conjunction with the action.
2. The key sensitive post whole life cycle management system according to claim 1, characterized in that It also includes: The job rotation execution supervision and handover module is used to generate a job rotation notification form based on the job rotation plan and push it to relevant personnel; extract target entities from the work documents of the original personnel based on a preset named entity recognition model, and generate a job handover list based on the association between the target entities and the job permission list; and backtrack and verify the service interval of the same person in the same position based on the job rotation history, and generate a warning message when the service interval does not meet the preset conditions.
3. The key sensitive post whole life cycle management system according to claim 1, characterized in that It also includes: The system implementation evaluation and accountability auxiliary module is used to calculate and display the implementation evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period; based on the identification of the involved personnel and the time interval of the case, it performs multi-dimensional backtracking queries on the system implementation history of the involved personnel, and generates a responsibility analysis report based on the backtracking query results.
4. The key sensitive post whole life cycle management system according to claim 1, characterized in that, The job management module includes: The job risk assessment unit is used to score the risk of each job based on a pre-built job risk scoring model to obtain the risk probability value of each job; jobs with risk probability values exceeding a preset threshold are marked as key sensitive jobs, and a list of key sensitive jobs is generated, which includes job identifier, department, risk score and marking basis. The dynamic update unit is used to monitor organizational structure adjustment events of the target enterprise according to a preset period. In response to the monitored events, it incrementally extracts data on changed positions, recalculates the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions based on a pre-built position risk scoring model, and updates the list of key sensitive positions based on the risk probability value of the affected positions and the risk probability value of the upstream and downstream related positions of the affected positions.
5. The key sensitive post whole life cycle management system according to claim 4, characterized in that, The job management module also includes: The job identification unit is used to acquire multi-source raw data for each job position in the target company. The multi-source raw data includes organizational structure data, job descriptions, job responsibility and authority texts, fund approval node data, and procurement and bidding records. Based on natural language processing technology and the multi-source raw data, determine the job entity, permission entity, resource entity, and the relationship between each entity; A knowledge graph is constructed based on the job entity, the permission entity, the resource entity, and the relationships between each entity.
6. The key sensitive post whole life cycle management system according to claim 4, characterized in that, The job risk assessment unit, when constructing a job risk scoring model, is specifically used for: Obtain historical violation case data, and based on the historical violation case data, determine a sample dataset. The sample dataset includes multiple samples, which are either positive or negative samples. The positive samples represent the positions involved in the historical violation case data, and the negative samples represent the positions not involved in the historical violation case data. Based on the knowledge graph, the adjacency weight, resource density value and business process centrality of the corresponding job node for each sample are determined to form a feature vector. The machine learning model is trained based on the positive samples, the negative samples, and the feature vectors to obtain the job risk scoring model.
7. The key sensitive post whole life cycle management system according to claim 1, characterized in that, The review module includes: A multi-source data fusion unit is used to acquire multi-source heterogeneous data, including contract employee identification, qualifications, annual performance evaluation level, violation handling records, number of reports, economic responsibility audit conclusions, and training completion status; and to determine a personnel qualification wide table based on the multi-source heterogeneous data, wherein the personnel qualification wide table represents structured personnel data; The personnel compliance profiling unit is used to determine the multi-dimensional compliance profile of each person based on the personnel qualification wide table, the preset sentiment analysis model, the preset time series analysis model and the indicator mapping rules. The multi-dimensional compliance profile includes compliance record score, compliance evaluation index, business capability index and disciplinary impact period status. The real-time qualification verification unit is used to obtain the employee number and target position identifier of the target personnel; query whether the target position is located in the list of key sensitive positions; in response to the target position being located in the list of key sensitive positions, obtain the multi-dimensional compliance profile of the target personnel; perform verification based on the multi-dimensional compliance profile and preset qualification review rules, and determine the review result, which includes review passed or review blocked.
8. The key sensitive post whole life cycle management system according to claim 7, characterized in that, The personnel compliance profiling unit is specifically used for: The evaluation text is obtained from the personnel qualification wide table, and the evaluation text is analyzed based on the preset sentiment analysis model to obtain a positive sentiment score. Extract the assessment grades for several consecutive years from the personnel qualification wide table, and calculate the assessment trend slope based on the preset time series analysis model; Based on the mapping relationship between the number of reports and the type of punishment, a compliance record score is determined; Based on the positive emotion score and the completion rate of training status, a compliance evaluation index is determined; Based on the slope of the assessment trend and the number of qualifications, the business capability index is determined; The status of the period of influence of the punishment is determined by comparing the end time of the violation handling record with the current system time. Based on the compliance record score, the compliance evaluation index, the business capability index, and the status of the disciplinary impact period, a personnel compliance profile is generated.
9. The system according to claim 1, wherein, The job rotation planning module includes: The job relevance calculation unit is used to obtain the co-occurrence frequency of business processes and the responsibility text among various jobs; calculate the process relevance score based on the business process co-occurrence frequency; calculate the responsibility similarity score based on the semantic vector of the responsibility text; and determine the relevance score among various jobs and generate a job relevance score matrix based on the process relevance score and the responsibility similarity score. The job rotation plan generation unit is used to obtain a list of personnel to be rotated, which includes personnel who have served continuously in their current key and sensitive positions for a preset number of years; determine a set of candidate target positions for each personnel to be rotated based on the job relevance score matrix; and determine the optimal target position from the set of candidate target positions based on preset job rotation constraints and an optimization algorithm to generate a job rotation plan; wherein the preset job rotation constraints include: the same person cannot rotate back to their current position within a set time interval, the number of people rotating in the same department at the same time does not exceed a preset upper limit of proportion, and the relevance score of the target position for rotation does not exceed a preset relevance threshold.
10. The key sensitive post whole life cycle management system according to claim 9, characterized in that, The optimization algorithm used in the job rotation scheme generation unit is a genetic algorithm. The objective function of the genetic algorithm is to maximize the sum of the risk hedging index and the talent training value index of all personnel to be rotated.
11. The system according to claim 9, wherein, The job rotation planning module also includes: The deferred job rotation identification unit is used to send a batch query request to the human resources system before generating the job rotation plan to obtain the expected retirement date and leave status of each employee in the list of employees to be rotated. Individuals whose expected retirement date differs from their current date by less than a preset number of days are marked as those nearing retirement. Remove the personnel who are about to retire and those who are on leave from the list of personnel awaiting rotation and add them to the list of personnel whose rotation is postponed. The list of personnel whose rotation is postponed includes personnel identification and postponement reason code. The list of personnel to be rotated after being removed is sent to the rotation plan generation unit.
12. The system according to claim 1, wherein, The early warning and handling module, include: The behavioral feature acquisition unit is used to acquire approval action data and financial correlation data of personnel in key and sensitive positions in real time; extract approval action features based on the approval action data; extract financial correlation features based on the financial correlation data; and generate job behavioral features of personnel in key and sensitive positions based on the approval action features and the financial correlation features. An anomaly detection and graded early warning unit is used to arrange the behavioral characteristics of multiple positions of the key and sensitive personnel in a preset time window in chronological order to obtain a historical job behavior characteristic sequence; and input the historical job behavior characteristic sequence into a preset anomaly detection model to obtain an anomaly score; The preset anomaly detection model is a time-series anomaly detection model based on an autoencoder; based on the anomaly score and preset multi-level thresholds, a warning level is determined, which includes a first level, a second level, and a third level; wherein, when the anomaly score is greater than the first threshold and less than or equal to the second threshold, it is determined to be the first level; when the anomaly score is greater than the second threshold and less than or equal to the third threshold, it is determined to be the second level; and when the anomaly score is greater than the third threshold, it is determined to be the third level. The response and coordination unit is used to execute corresponding response actions based on the warning level.
13. The key sensitive post whole life cycle management system according to claim 12, characterized in that, The early warning and handling module also includes: The conversation assistance unit is used to monitor personnel recruitment events, job transfer events, and first-level warning events; in response to the monitored events, it extracts risk point tags from the multi-dimensional compliance profiles of the personnel in the key and sensitive positions; based on the preset mapping rules between risk point tags and reminder content, it determines the key reminder content; based on the key reminder content, it calls the template engine to generate a conversation prompt sheet; it monitors the upload status of the conversation records corresponding to the conversation prompt sheet, and if the conversation records are not uploaded within a preset time limit, it upgrades the first-level warning to the second-level warning; The job transfer trigger unit is used to acquire Level 3 warning events; in response to the acquisition of Level 3 warning events, it sends an account freeze command to the access control platform to freeze all job-related system accounts of the personnel in the key sensitive positions.
14. The system according to claim 2, wherein, The rotation execution supervision and handover module includes: The job rotation task process engine is used to receive the job rotation plan output by the job rotation planning module, instantiate a process task for each job rotation action in the job rotation plan, call the electronic signature service, and generate a job rotation notification based on the personnel information, original position, new position and reporting deadline in the process task; and push the job rotation notification to the job rotation personnel and department heads through multiple set channels. The intelligent handover list generation unit is used to obtain work order content, email body, and project documents from the original personnel's to-do list; based on a named entity recognition model using natural language processing, it extracts target entities from the work order content, email body, and project documents. The target entities include supplier name, contract number, process instance identifier in approval, and document number; it associates the target entities with the original personnel's job permission list to generate a structured job handover list. The job handover list displays the hierarchical relationship of responsibilities, to-do items, key contacts, and confidential assets in a tree structure; it receives confirmation operations from the original personnel and the rotating personnel on the job handover list, and records the operator, timestamp, and client address corresponding to each confirmation operation; The job qualification retrospective verification unit is used to scan the job rotation history record table according to a preset cycle to check whether there are two job records for the same personnel ID and the same job ID, and whether the interval between the end time of the previous job and the start time of the next job is less than a preset return interval threshold. If such a situation exists, an early warning information is pushed to the management dashboards of the audit department and the organization and personnel department. The early warning information includes the personnel name, the job involved, the time of the violation return, and the information of the responsible person.
15. The system according to claim 3, wherein, The system implementation evaluation and accountability support module includes: The execution indicator monitoring dashboard is used to aggregate and calculate the execution evaluation indicators of each department based on the operation logs of each module according to a preset aggregation period. The execution evaluation indicators include job rotation rate, job rotation timeliness rate, conversation completion rate, and handover list completeness rate. The execution evaluation indicators and overall ranking of each department are displayed in the form of visual charts. The violation incident tracing unit is used to obtain the identification of the personnel involved and the time range of the case; to perform a retrospective query on the personnel involved based on the case time range and obtain the retrospective query results; and to generate a responsibility analysis report based on the retrospective query results. The retrospective query includes: obtaining all job rotation records of the personnel involved within the case time range and determining whether there are any cases of failure to rotate jobs within the time limit; obtaining the qualification review log of the personnel involved and determining whether there are any records of failure to block violations that should have been blocked; and obtaining the risk warning history of the personnel involved and determining whether there are any warning records that have been triggered but not handled.
16. A method for critical sensitive position life cycle management, characterized in that, include: Based on knowledge graphs and machine learning models, the various positions in the target company are evaluated, and the list of key and sensitive positions is dynamically updated based on the evaluation results. Acquire multi-source heterogeneous data, and based on this data, determine a multi-dimensional compliance profile of the target personnel; When the target personnel are to be appointed to a position in the list of key and sensitive positions, the review result is determined based on the target personnel's multi-dimensional compliance profile and the preset qualification review rules; Based on a quantitative job relevance score matrix, preset job rotation constraints, and optimization algorithms, a job rotation plan is generated for employees who meet the job rotation conditions. The job relevance score matrix is determined based on the business process relationship and responsibility similarity between jobs. For any key sensitive personnel, the job behavior characteristics of the key sensitive personnel are obtained in real time; Anomaly scores are calculated based on the historical behavioral characteristic sequences of personnel in key and sensitive positions and a preset anomaly detection model. Based on the anomaly scores, tiered early warnings are triggered and corresponding action actions are executed in conjunction with the action.
17. An electronic device, comprising: Includes a processor, which is coupled to a memory; The processor is configured to execute a computer program stored in the memory, so that the electronic device performs the method as described in claim 16.
18. A computer-readable storage medium, characterized in that, Includes a computer program or instructions that, when run on a computer, cause the computer to perform the method as described in claim 16.
