An automatic performance evaluation management system for employees

Abstract

The application discloses an employee performance automatic examination and evaluation management system, and relates to the technical field of intelligent performance evaluation.The system comprises a structured data collection module, a behavior data modeling module, an initial performance modeling module, a dynamic weight optimization module, a performance report generation module and a blockchain storage module.The structured data collection module is used to connect the human resource information system, the enterprise resource planning system, the customer relationship management system and the project management platform in the enterprise through a standardized interface, and to collect the structured performance-related data generated by the employees in the above systems in real time based on a business event triggering mechanism.The multi-source structured and unstructured data are fused to construct a collaborative relationship modeling mechanism based on a graph neural network and a dynamic weight optimization mechanism based on reinforcement learning, so that a comprehensive, objective and personalized evaluation of the employee performance is realized.Meanwhile, the blockchain storage technology is introduced to ensure that the whole examination process is traceable and tamper-proof.

Description

Technical Field

[0001] This invention relates to the field of intelligent performance evaluation technology, and in particular to an automatic employee performance evaluation and management system. Background Technology

[0002] Intelligent performance evaluation technology refers to a technological system that integrates artificial intelligence, big data analytics, and automated systems to collect, semantically understand, model relationships, and quantitatively analyze structured task data and unstructured behavioral data generated by employees across multiple business and collaborative platforms in real time. This results in the construction of dynamic, interpretable, and personalized performance evaluation models. This technology overcomes the limitations of traditional subjective scoring and static indicators, objectively reflecting employees' task completion quality, alignment with strategic goals, collaborative contributions, and innovative behavior. Furthermore, it continuously optimizes evaluation criteria through machine learning, achieving fairer, more accurate, and development-oriented performance management.

[0003] Traditional performance appraisals often rely on periodic self-assessments or superiors' scores, lacking real-time and objective records of employees' actual behavior and output in the business system. Furthermore, the general weighting configuration makes it difficult to take into account strategic orientation, job characteristics, and individual growth paths, thus weakening the accuracy and guiding role of performance management. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an automatic employee performance appraisal and management system that solves the problems of traditional performance appraisal relying on periodic self-evaluation or superior scoring, lacking real-time and objective recording of employees' actual behavior and output in the business system, and the difficulty in taking into account strategic orientation, job characteristics and individual growth paths when configuring general weights, thus weakening the accuracy and guiding role of performance management.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an automatic employee performance appraisal and management system, comprising: The system includes a structured data acquisition module, a behavioral data modeling module, an initial performance modeling module, a dynamic weight optimization module, a performance report generation module, and a blockchain-based evidence storage module. The structured data acquisition module is used to connect to the enterprise's human resources information system, enterprise resource planning system, customer relationship management system and project management platform through standardized interfaces, and to collect structured performance-related data generated by employees in the above systems in real time based on a business event triggering mechanism. The behavioral data modeling module is used to seamlessly collect employees' unstructured digital behavioral data from collaborative communication platforms, meeting recording systems, enterprise knowledge bases, and code hosting platforms, and to perform semantic analysis and collaborative relationship modeling on the unstructured digital behavioral data to generate peripheral performance quantitative indicators. The initial performance modeling module is used to construct an initial comprehensive performance calculation model by combining job value benchmark scores, target alignment, collected structured performance-related data, and generated peripheral performance quantitative indicators. The dynamic weight optimization module is used to introduce reinforcement learning agents, with organizational goal achievement rate, performance dispute occurrence rate and employee engagement as optimization targets, to dynamically adjust the dimension weights in the initial comprehensive performance calculation model, and generate a personalized performance evaluation model that is adapted to the employee's job type and development stage. The performance report generation module is used to apply the personalized performance evaluation model to structured performance-related data and surrounding quantitative performance indicators to calculate the employee's comprehensive performance score and generate a performance evaluation report that includes a description of the contribution of each dimension. The blockchain evidence storage module is used to hash key information such as performance target confirmation records, data collection snapshots, weight adjustment logs, and comprehensive performance scores and then write them into the blockchain to form an immutable evidence storage chain for the entire performance evaluation process.

[0007] As a preferred embodiment of the employee performance automatic assessment and management system of the present invention, the following steps are included: seamlessly collecting unstructured digital behavior data of employees from collaborative communication platforms, meeting recording systems, enterprise knowledge bases, and code hosting platforms; performing semantic analysis and collaborative relationship modeling on the unstructured digital behavior data; and generating peripheral performance quantitative indicators, including: Receive message logs from collaborative communication platforms, transcribed texts from meeting recording systems, document editing records from enterprise knowledge bases, and commit history from code hosting platforms; For each message log, extract the sender identifier, receiver identifier, message timestamp, and text content; for each meeting transcript, extract the list of participants, speaking paragraphs, and their corresponding time intervals; for each knowledge base document, extract the author, reviser, citation count, and tag set; for each code commit, extract the developer identifier, associated task number, modified file path, and comment content. Based on the above extraction results, an employee-behavioral event matrix is ​​constructed. The matrix rows correspond to employees, and the columns correspond to behavioral event types. The matrix elements represent the participation intensity of an employee in a certain type of behavioral event. The participation intensity is calculated by weighting the event frequency, duration, or content length after normalization. Using employees as nodes and co-occurrence relationships of behaviors as edges, a directed weighted graph is constructed. If any two employees in the graph co-occur in at least one type of behavioral event, a directed edge is established. The weight of the edge is determined by the weighted sum of the joint occurrence frequencies of the two employees in all types of behavioral events. The constructed directed weighted graph is input into a graph neural network, and neighborhood information is aggregated through two layers of graph convolution operations to obtain the embedding vector of each employee. The embedded vectors are connected to three fully connected layers respectively, and the outputs of collaboration activity, knowledge contribution index and innovation behavior score constitute the peripheral performance quantitative indicator vector.

[0008] As a preferred embodiment of the employee performance automatic assessment and evaluation management system of the present invention, the weights of the edges in the directed weighted graph are calculated as follows: For any two employees and If the two have joint behavior in message interaction, meeting co-occurrence, document collaboration or code review, then the co-occurrence scores of the four types of behavior are calculated separately, and then weighted and summed according to a preset ratio. The message interaction co-occurrence score is equal to the geometric mean of the total number of messages exchanged between the two parties divided by the total number of messages exchanged by each party. The meeting co-occurrence score is equal to the number of meetings attended by both parties divided by the arithmetic mean of the total number of meetings each party attended. Document collaboration score equals the number of times two people jointly edit or reference the same document; Code review score equals employee The submitted code was by the employee The number of times the review is approved is related to the number of employees. The submitted code was by the employee The sum of the number of times the review was approved; The four categories of scores are multiplied by their respective weighting coefficients. The weights are then added together to obtain the final edge weights. ,in Furthermore, each weighting coefficient is pre-configured based on the characteristics of the company's collaborative culture.

[0009] As a preferred embodiment of the employee performance automatic assessment and management system of the present invention, the step of introducing a reinforcement learning agent, using organizational goal achievement rate, performance dispute occurrence rate, and employee engagement as optimization objectives, dynamically adjusts the dimension weights in the initial comprehensive performance calculation model to generate a personalized performance evaluation model adapted to the employee's job type and development stage, including: Define the state space as the structured performance-related data of all employees in the current assessment period, the surrounding quantitative performance indicators, and the current dimension weight vector; The action space is defined as a small perturbation of the dimension weight vector, after which the weights of each dimension remain non-negative and their sum is 1; The reward function is defined as a linear combination of organizational goal achievement rate, performance dispute rate, and employee engagement. The organizational goal achievement rate reflects the completion of strategic goals at the grassroots level. The performance dispute rate is the percentage of employees who file formal appeals against performance results. Employee engagement is obtained through anonymous pulse surveys. The near-end policy optimization algorithm is used to iteratively update the policy network parameters to maximize the expected cumulative reward; The policy network outputs the dimension weight vector for the next cycle, which is used to construct the personalized performance evaluation model.

[0010] As a preferred embodiment of the employee performance automatic assessment and evaluation management system of the present invention, the calculation process of the organizational goal achievement rate includes: Obtain the set of key outcome items for decomposing the company's annual strategic goals into each department; For each key outcome item, identify the employee group directly associated with it; Calculate the average performance score of this employee group during the current assessment period; The organizational goal achievement rate is obtained by weighting the average performance scores corresponding to each key outcome item according to their strategic importance. Strategic importance weights are set by the senior management team at the beginning of the quarter, reflecting the contribution of the key outcome to the overall strategy; if no weight is set for a key outcome, the default weight is 1. In the weighted average process, only employees who have completed data collection are included; employees with missing data are not included in the calculation.

[0011] As a preferred embodiment of the employee performance automatic assessment and management system of the present invention, the step of applying a personalized performance evaluation model to structured performance-related data and surrounding quantitative performance indicators to calculate the employee's comprehensive performance score and generate a performance evaluation report including dimensional contribution descriptions includes: Structured performance-related data is organized into a vector that includes dimensions such as sales achievement rate, work order processing timeliness, and customer satisfaction score. The surrounding performance quantitative indicators are organized into a vector that includes collaboration activity, knowledge contribution index, and innovation behavior score; Concatenate the two vectors into a single input vector; The weight vector of the dimensions output by the personalized performance evaluation model is used to sum the weighted values ​​of each dimension of the comprehensive input vector to obtain the comprehensive performance score. For each dimension, calculate its weighted value as a percentage of the overall performance score, which is the contribution of that dimension; write the name, original value, weight, weighted value, and contribution of each dimension into the performance evaluation report for employees and managers to review. The overall performance score is not subject to mandatory distribution or ranking; it is only used as a reference for development feedback and incentive allocation. If a certain dimension's original value is missing during the report generation process, the weight of that dimension will be automatically reset to zero, and the weights of the other dimensions will be renormalized proportionally.

[0012] As a preferred embodiment of the employee performance automatic assessment and management system of the present invention, the step of hashing key information such as performance target confirmation records, data collection snapshots, weight adjustment logs, and comprehensive performance scores and then writing them into the blockchain to form an immutable performance assessment process evidence chain includes: Perform a SHA-256 hash operation on the performance target confirmation record to generate the first hash value; the performance target confirmation record includes the employee's personal work plan, the approval opinion of the immediate supervisor, and the electronic signature timestamp of both parties; Perform a SHA-256 hash operation on the data collection snapshot to generate a second hash value; the data collection snapshot contains a complete copy of structured performance-related data and unstructured digital behavioral data at the assessment deadline; Perform a SHA-256 hash operation on the weight adjustment log to generate a third hash value; the weight adjustment log contains the dimension weight vector output by the reinforcement learning agent, the input state description, and the policy network version number; Perform a SHA-256 hash operation on the overall performance score to generate a fourth hash value; the overall performance score includes the final value and all input data identifiers used in the calculation. The four hash values ​​are concatenated into a single string in sequence, and the SHA-256 hash operation is performed on the string again to generate the final evidence hash. The final notarization hash, assessment period identifier, system version number, and timestamp will be written into the smart contract address of the consortium blockchain.

[0013] As a preferred embodiment of the employee performance automatic assessment and management system of the present invention, the standardized interface connects to the enterprise's human resources information system, enterprise resource planning system, customer relationship management system, and project management platform, and collects structured performance-related data generated by employees in the above systems in real time based on a business event triggering mechanism, including: Configure an event listener in the human resources information system to listen for events such as the resolution of employee attendance exceptions, completion of training courses, and effective job transfers. When an event occurs, the system's RESTful API is invoked to retrieve the employee ID associated with the event, the event type code, the time of occurrence, and the result status code. Configure event listeners in the enterprise resource planning system to listen for events such as production work order closure, procurement cost savings confirmation, and inventory count difference processing completion; after the event is triggered, extract the work order number, responsible person, and the amount of savings or difference rate quantification fields. Configure event listeners in the customer relationship management system to listen for events such as customer satisfaction survey submission, sales opportunity status change to "completed", and service request response timeout. Event data includes customer ID, employee ID, rating value, or status change time. Configure event listeners in the project management platform to listen for events such as task status updates to "completed", project milestones achieved, and delay risks resolved. Event information includes the task's unique identifier, the person in charge, and the difference between the actual completion time and the planned completion time. All event data is converted to a unified format and stored in a message queue, which is then consumed by the data acquisition service and written into the structured performance database.

[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the employee performance automatic assessment and evaluation management system as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the employee performance automatic assessment and evaluation management system as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by integrating multi-source structured and unstructured data, a collaborative relationship modeling mechanism based on graph neural networks and a dynamic weight optimization mechanism based on reinforcement learning are constructed to achieve a comprehensive, objective, and personalized evaluation of employee performance. At the same time, blockchain notarization technology is introduced to ensure that the entire assessment process is traceable and tamper-proof, thereby improving the fairness, transparency, and organizational adaptability of performance management, effectively reducing the risk of human intervention and disputes, and enhancing employee trust and dedication. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a schematic diagram of the automatic employee performance appraisal and evaluation management system in Example 1. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0021] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0022] Example 1, referring to Figure 1 As one embodiment of the present invention, this embodiment provides an automatic employee performance appraisal and management system, comprising: The system includes a structured data acquisition module, a behavioral data modeling module, an initial performance modeling module, a dynamic weight optimization module, a performance report generation module, and a blockchain-based evidence storage module.

[0023] By connecting to the enterprise's human resources information system, enterprise resource planning system, customer relationship management system and project management platform through standardized interfaces, the system collects structured performance-related data generated by employees in the above systems in real time based on a business event triggering mechanism.

[0024] Furthermore, configure event listeners in the human resources information system to monitor events such as the resolution of employee attendance exceptions, completion of training courses, and effective job transfers. When an event occurs, call the RESTful API provided by the system to obtain the employee ID, event type code, occurrence time, and result status code associated with the event. Configure event listeners in the enterprise resource planning system to monitor events such as production work order closure, confirmation of procurement cost savings, and completion of inventory count discrepancy processing. After an event is triggered, extract the work order number, responsible person, and the amount of savings or the discrepancy rate quantification field.

[0025] Configure event listeners in the customer relationship management system to monitor events such as customer satisfaction survey submissions, sales opportunity status changes to "completed," and service request response timeouts. Event data includes customer ID, employee ID, rating value, or status change time. Configure event listeners in the project management platform to monitor events such as task status updates to "completed," project milestone achievement, and delay risk mitigation. Event information includes a unique task identifier, responsible person, and the difference between the actual completion time and the planned completion time. All event data is converted to a unified format and stored in a message queue, consumed by the data acquisition service, and written to the structured performance database.

[0026] It should be noted that by deploying event listeners in various core business systems and capturing key performance events in real time based on standardized interfaces, automated, seamless, and timely collection of employee work outputs has been achieved, avoiding the lag and subjective bias caused by traditional manual reporting. At the same time, the unified data format conversion and message queue mechanism ensure the orderly aggregation and consistent processing of multi-source heterogeneous data, providing a high-quality, traceable, and structured input foundation for subsequent performance modeling.

[0027] Unstructured digital behavior data of employees is collected seamlessly from collaborative communication platforms, meeting recording systems, enterprise knowledge bases, and code hosting platforms. Semantic analysis and collaborative relationship modeling are performed on the unstructured digital behavior data to generate quantitative indicators of peripheral performance.

[0028] Furthermore, it receives message logs from collaborative communication platforms, transcribed texts from meeting recording systems, document editing records from enterprise knowledge bases, and submission history from code hosting platforms; it extracts the sender identifier, receiver identifier, message timestamp, and text content from each message log; it extracts the list of participants, speaking paragraphs, and their corresponding time intervals from each meeting transcript; it extracts the author, reviser, citation count, and tag set from each knowledge base document; and it extracts the developer identifier, associated task number, modified file path, and comment content from each code submission.

[0029] Based on the above extraction results, an employee-behavioral event matrix is ​​constructed. The rows of the matrix correspond to employees, and the columns correspond to behavioral event types. The matrix elements represent the participation intensity of an employee in a certain type of behavioral event. This participation intensity is calculated by weighting the normalized frequency, duration, or content length of the event. Using employees as nodes and behavioral co-occurrence relationships as edges, a directed weighted graph is constructed. If any two employees in the graph co-occur in at least one type of behavioral event, a directed edge is established. The weight of the edge is determined by the weighted sum of the joint occurrence frequencies of the two employees in various types of behavioral events. The constructed directed weighted graph is input into a graph neural network, and neighborhood information is aggregated through two layers of graph convolution operations to obtain the embedding vector of each employee. The embedding vectors are then connected to three fully connected layers to output collaborative activity, knowledge contribution index, and innovative behavior score, forming a vector of peripheral performance quantitative indicators.

[0030] The weights of edges in a directed weighted graph are calculated by considering any two employees. and If the two parties engage in joint behavior in message interaction, meeting co-occurrence, document collaboration, or code review, then the co-occurrence scores for each of the four types of behavior are calculated separately, and then weighted and summed according to a preset ratio. The message interaction co-occurrence score is equal to the geometric mean of the total number of messages exchanged between the two parties divided by the total number of messages exchanged by each party. The meeting co-occurrence score is equal to the arithmetic mean of the number of meetings attended by both parties jointly divided by the total number of meetings attended by each party. The document collaboration score is equal to the number of times the two parties jointly edit or reference the same document. The code review score is equal to the employee's... The submitted code was by the employee The number of times the review is approved is related to the number of employees. The submitted code was by the employee The sum of the number of times the review was approved; the scores of the four categories are multiplied by their respective weighting coefficients. The weights are then added together to obtain the final edge weights. ,in Furthermore, each weighting coefficient is pre-configured based on the characteristics of the company's collaborative culture.

[0031] Among them, the first in the graph neural network The layer node embedding update uses the following formula: ; In the formula, Indicates the first Employees in Embedded vectors after layer graph convolution; Indicates employees The set of first-order neighbors in a directed weighted graph; Indicates employees The number of neighbors, i.e. ; For the first The layer-learnable weight matrix has the same dimensions as the embedding vector; It is the ReLU activation function; For the first The embedding vector of the employee in the previous layer.

[0032] It should be noted that by seamlessly collecting and deeply semantically analyzing digital footprints such as collaborative communication, meeting minutes, knowledge bases, and code platforms, and combining graph neural networks to model collaborative relationships among employees, the method effectively transforms soft behaviors that are originally difficult to quantify into calculable peripheral performance indicators. The method not only captures employees' knowledge sharing, collaborative support, and innovative contributions outside of formal tasks, but also preserves the real interactive network characteristics within the organization through graph structures, thereby improving the comprehensiveness and behavioral orientation of performance evaluation.

[0033] An initial comprehensive performance calculation model is constructed by combining job value benchmark scores, target alignment, collected structured performance-related data, and generated peripheral performance quantitative indicators.

[0034] Furthermore, the system obtains the benchmark score for each position within the company's job system. This score is pre-assessed by the job evaluation committee based on the complexity of job responsibilities, scope of impact, and required competency levels, and is stored numerically in the human resources information system. It also obtains the set of individual performance goals jointly confirmed by each employee and their immediate supervisor at the beginning of the current performance cycle, as well as the set of key outcome items for the decomposition of the company's strategic goals to departments. For each individual performance goal, the system calculates its semantic similarity to the corresponding key outcome item, using a cosine similarity method based on word vector averaging to obtain a single-item goal alignment score. Finally, it takes the arithmetic mean of all individual performance goal alignment scores as the employee's goal alignment degree. The job value benchmark score, goal alignment, structured performance-related data vector, and peripheral performance quantitative indicator vector are normalized to ensure that the values ​​of each dimension are on the same scale. The normalization method uses min-max scaling to map the original values ​​to the range of 0 to 1. The four types of inputs after normalization are then weighted and merged according to a preset initial weight allocation scheme to form an initial comprehensive performance score. The initial weight allocation scheme is set by the human resources expert group based on the organization's development stage and management priorities to ensure that job value reflects the basic positioning, goal alignment reflects the consistency of direction, structured data measures the quality of task completion, and peripheral indicators capture collaborative and innovative behaviors.

[0035] Among them, the initial comprehensive performance score The calculation uses the following formula: ; In the formula, This indicates the employee's initial overall performance score; This represents the job value benchmark score for the employee's position, which ranges from 0 to 1 after min-max normalization. This indicates the employee's target alignment, with a value ranging from 0 to 1; Indicates the first The normalized values ​​of the structured performance-related data are as follows: item; Indicates the first The normalized values ​​of the surrounding performance quantitative indicators totaled item; , , , These are the initial weight coefficients for the corresponding dimensions, satisfying... , , , And the sum of all weights equals 1.

[0036] It should be noted that by organically integrating the four dimensions of inherent job value, alignment with goals and strategies, task execution results, and collaborative innovation behavior, an initial evaluation framework is constructed that takes into account job benchmarks, directional consistency, and individual behavioral performance. This design respects the differences in basic contributions caused by job differences, emphasizes the synergy between individual goals and organizational strategies, and incorporates objective task results and implicit collaborative value, laying a structural foundation for achieving fair, multidimensional, and interpretable performance evaluation.

[0037] By introducing reinforcement learning agents, and using organizational goal achievement rate, performance dispute rate, and employee engagement as optimization targets, the dimensional weights in the initial comprehensive performance calculation model are dynamically adjusted to generate personalized performance evaluation models that are adapted to employee job types and development stages.

[0038] Furthermore, the state space is defined as the structured performance-related data of all employees in the current assessment period, the surrounding quantitative performance indicators, and the current dimension weight vector; the action space is defined as a small perturbation of the dimension weight vector, after which the weights of each dimension remain non-negative and their sum is 1; the reward function is defined as a linear combination of organizational goal achievement rate, performance dispute occurrence rate, and employee engagement, where the organizational goal achievement rate reflects the completion of strategic goals decomposed to the grassroots level, the performance dispute occurrence rate is the proportion of employees who formally appeal performance results, and employee engagement is obtained through anonymous pulse surveys; the proximal strategy optimization algorithm is used to iteratively update the strategy network parameters to maximize the expected cumulative reward; the strategy network outputs the dimension weight vector for the next period, which is used to construct the personalized performance evaluation model.

[0039] The calculation process for organizational goal achievement rate includes: obtaining a set of key outcome items for each department, breaking down the company's annual strategic goals; identifying the directly related employee group for each key outcome item; calculating the average performance score of that employee group within the current assessment period; and weighting the average performance scores corresponding to each key outcome item according to their strategic importance weights to obtain the organizational goal achievement rate. The strategic importance weights are set by the senior management team at the beginning of the quarter, reflecting the contribution of the key outcome item to the overall strategy. If a key outcome item does not have a weight set, the default weight is 1. In the weighted averaging process, only employees who have completed data collection are included; employees with missing data are not included in the calculation.

[0040] The reward function expression is as follows: ; In the formula, This represents the total reward value for the current period; This indicates the organization's goal achievement rate, with a value ranging from 0 to 1. A higher value indicates a better degree of strategic goal achievement. This represents the performance dispute rate, with a value ranging from 0 to 1. The lower the value, the higher the acceptance of the performance results. This indicates employee engagement, with a value ranging from 0 to 1. A higher value indicates a stronger willingness to engage. These are positive real coefficients used to adjust the relative importance of the three indicators in the optimization process. The sum of the three is not required to be 1, but it must satisfy... .

[0041] It should be noted that the introduction of a reinforcement learning mechanism with organizational goal achievement, dispute control, and employee engagement as optimization orientations enables the performance evaluation model to continuously self-calibrate based on feedback from the organization's actual operations. This mechanism breaks through the rigidity of the static weight system, allowing evaluation standards to be dynamically adapted to job characteristics, development stages, and management priorities, thereby ensuring the implementation of strategies while enhancing employees' acceptance of the assessment results and the strength of their psychological contract.

[0042] The personalized performance evaluation model is applied to structured performance-related data and surrounding quantitative performance indicators to calculate the employee's comprehensive performance score and generate a performance evaluation report that includes dimensional contribution descriptions.

[0043] Furthermore, the structured performance-related data is organized into a vector containing dimensions such as sales achievement rate, work order processing timeliness, and customer satisfaction score; the peripheral performance quantitative indicators are organized into a vector containing collaboration activity, knowledge contribution index, and innovation behavior score; the two vectors are concatenated into a comprehensive input vector; the dimension weight vector output by the personalized performance evaluation model is applied to the comprehensive input vector to perform a weighted summation of each dimension, resulting in a comprehensive performance score; for each dimension, the proportion of its weighted value to the comprehensive performance score is calculated as the contribution of that dimension; the name, original value, weight, weighted value, and contribution of each dimension are written into the performance evaluation report for employees and managers to review; the comprehensive performance score is not forcibly distributed or ranked, but only serves as a reference for development feedback and incentive allocation; during the report generation process, if the original value of a certain dimension is missing, the weight of that dimension is automatically reset to zero, and the weights of the remaining dimensions are renormalized proportionally.

[0044] It should be noted that while outputting the overall performance score, a transparent explanation of the original values, weighting, and contribution ratio of each dimension is provided, enabling employees to clearly understand their strengths and areas for improvement. The report does not use mandatory ranking or distribution restrictions, but focuses on developmental feedback and incentive basis, which helps to create a growth-oriented performance culture, reduce internal competition and friction, and promote continuous improvement and capability enhancement.

[0045] The key information, including performance target confirmation records, data collection snapshots, weight adjustment logs, and comprehensive performance scores, is hashed and written into the blockchain to form an immutable evidence chain for the entire performance evaluation process.

[0046] Furthermore, a SHA-256 hash operation is performed on the performance target confirmation record to generate the first hash value; the performance target confirmation record includes the employee's personal work plan, the direct supervisor's approval opinion, and the electronic signature timestamp of both parties; a SHA-256 hash operation is performed on the data collection snapshot to generate the second hash value; the data collection snapshot contains a complete copy of the structured performance-related data and unstructured digital behavioral data at the assessment deadline; a SHA-256 hash operation is performed on the weight adjustment log to generate the third hash value; the weight adjustment log includes the dimension weight vector output by the reinforcement learning agent, the input state description, and the policy network version number; a SHA-256 hash operation is performed on the comprehensive performance score to generate the fourth hash value; the comprehensive performance score includes the final value and all input data identifiers used in the calculation; the four hash values ​​are concatenated into a single string in sequence, and a SHA-256 hash operation is performed on this string again to generate the final evidence hash; the final evidence hash, assessment period identifier, system version number, and timestamp are written to the smart contract address of the consortium blockchain.

[0047] It should be noted that by writing key information such as performance targets, data snapshots, weight adjustment logs, and final scores into the consortium blockchain through multiple hash operations, an end-to-end immutable chain of evidence is constructed. This mechanism not only enhances the audit compliance and legal validity of performance management, but also increases employees' trust in the fairness of the assessment process, effectively preventing and resolving labor disputes caused by data controversies.

[0048] This embodiment also provides a computer device suitable for an automatic employee performance appraisal and evaluation management system, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the automatic employee performance appraisal and evaluation management system proposed in the above embodiment.

[0049] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0050] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the employee performance automatic assessment and evaluation management system as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0051] In summary, this invention integrates multi-source structured and unstructured data to construct a collaborative relationship modeling mechanism based on graph neural networks and a dynamic weight optimization mechanism based on reinforcement learning. This enables a comprehensive, objective, and personalized evaluation of employee performance. Furthermore, the introduction of blockchain notarization technology ensures that the entire assessment process is traceable and tamper-proof, enhancing the fairness, transparency, and organizational adaptability of performance management. It effectively reduces the risk of human intervention and disputes, and strengthens employee trust and engagement.

[0052] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An automatic employee performance appraisal and management system, characterized in that: include: The system includes a structured data acquisition module, a behavioral data modeling module, an initial performance modeling module, a dynamic weight optimization module, a performance report generation module, and a blockchain-based evidence storage module. The structured data acquisition module is used to connect to the enterprise's human resources information system, enterprise resource planning system, customer relationship management system and project management platform through standardized interfaces, and to collect structured performance-related data generated by employees in the above systems in real time based on a business event triggering mechanism. The behavioral data modeling module is used to seamlessly collect employees' unstructured digital behavioral data from collaborative communication platforms, meeting recording systems, enterprise knowledge bases, and code hosting platforms, and to perform semantic analysis and collaborative relationship modeling on the unstructured digital behavioral data to generate peripheral performance quantitative indicators. The initial performance modeling module is used to construct an initial comprehensive performance calculation model by combining job value benchmark scores, target alignment, collected structured performance-related data, and generated peripheral performance quantitative indicators. The dynamic weight optimization module is used to introduce reinforcement learning agents, with organizational goal achievement rate, performance dispute occurrence rate and employee engagement as optimization targets, to dynamically adjust the dimension weights in the initial comprehensive performance calculation model, and generate a personalized performance evaluation model that is adapted to the employee's job type and development stage. The performance report generation module is used to apply the personalized performance evaluation model to structured performance-related data and surrounding quantitative performance indicators to calculate the employee's comprehensive performance score and generate a performance evaluation report that includes a description of the contribution of each dimension. The blockchain evidence storage module is used to hash key information such as performance target confirmation records, data collection snapshots, weight adjustment logs, and comprehensive performance scores and then write them into the blockchain to form an immutable evidence storage chain for the entire performance evaluation process.

2. The employee performance automatic appraisal and evaluation management system as described in claim 1, characterized in that: The process involves seamlessly collecting unstructured digital behavior data from collaborative communication platforms, meeting recording systems, enterprise knowledge bases, and code hosting platforms, performing semantic analysis and collaborative relationship modeling on this unstructured digital behavior data, and generating peripheral performance quantification indicators, including: Receive message logs from collaborative communication platforms, transcribed texts from meeting recording systems, document editing records from enterprise knowledge bases, and commit history from code hosting platforms; For each message log, extract the sender identifier, receiver identifier, message timestamp, and text content; for each meeting transcript, extract the list of participants, speaking paragraphs, and their corresponding time intervals; for each knowledge base document, extract the author, reviser, citation count, and tag set; for each code commit, extract the developer identifier, associated task number, modified file path, and comment content. Based on the above extraction results, an employee-behavioral event matrix is ​​constructed. The matrix rows correspond to employees, and the columns correspond to behavioral event types. The matrix elements represent the participation intensity of an employee in a certain type of behavioral event. The participation intensity is calculated by weighting the event frequency, duration, or content length after normalization. Using employees as nodes and co-occurrence relationships of behaviors as edges, a directed weighted graph is constructed. If any two employees in the graph co-occur in at least one type of behavioral event, a directed edge is established. The weight of the edge is determined by the weighted sum of the joint occurrence frequencies of the two employees in all types of behavioral events. The constructed directed weighted graph is input into a graph neural network, and neighborhood information is aggregated through two layers of graph convolution operations to obtain the embedding vector of each employee. The embedded vectors are connected to three fully connected layers respectively, and the outputs of collaboration activity, knowledge contribution index and innovation behavior score constitute a vector of peripheral performance quantitative indicators. Among them, the first in the graph neural network The layer node embedding update uses the following formula: ; In the formula, Indicates the first Employees in Embedded vectors after layer graph convolution; Indicates employees The set of first-order neighbors in a directed weighted graph; Indicates employees The number of neighbors, i.e. ; For the first The layer-learnable weight matrix has the same dimensions as the embedding vector; It is the ReLU activation function; For the first The embedding vector of the employee in the previous layer.

3. The employee performance automatic appraisal and evaluation management system as described in claim 2, characterized in that: The weights of the edges in the directed weighted graph are calculated as follows: For any two employees and If the two have joint behavior in message interaction, meeting co-occurrence, document collaboration or code review, then the co-occurrence scores of the four types of behavior are calculated separately, and then weighted and summed according to a preset ratio. The message interaction co-occurrence score is equal to the geometric mean of the total number of messages exchanged between the two parties divided by the total number of messages exchanged by each party. The meeting co-occurrence score is equal to the number of meetings attended by both parties divided by the arithmetic mean of the total number of meetings each party attended. Document collaboration score equals the number of times two people jointly edit or reference the same document; Code review score equals employee The submitted code was by the employee The number of times the review is approved is related to the number of employees. The submitted code was by the employee The sum of the number of times the review was approved; The four categories of scores are multiplied by their respective weighting coefficients. The weights are then added together to obtain the final edge weights. ,in Furthermore, each weighting coefficient is pre-configured based on the characteristics of the company's collaborative culture.

4. The employee performance automatic appraisal and evaluation management system as described in claim 3, characterized in that: The introduction of reinforcement learning agents, using organizational goal achievement rate, performance dispute rate, and employee engagement as optimization objectives, dynamically adjusts the dimension weights in the initial comprehensive performance calculation model to generate a personalized performance evaluation model adapted to employee job types and development stages, including: Define the state space as the structured performance-related data of all employees in the current assessment period, the surrounding quantitative performance indicators, and the current dimension weight vector; The action space is defined as a small perturbation of the dimension weight vector, after which the weights of each dimension remain non-negative and their sum is 1; The reward function is defined as a linear combination of organizational goal achievement rate, performance dispute rate, and employee engagement. The organizational goal achievement rate reflects the completion of strategic goals at the grassroots level. The performance dispute rate is the percentage of employees who file formal appeals against performance results. Employee engagement is obtained through anonymous pulse surveys. The near-end policy optimization algorithm is used to iteratively update the policy network parameters to maximize the expected cumulative reward; The policy network outputs the dimension weight vector for the next cycle, which is used to construct the personalized performance evaluation model. The reward function expression is as follows: ; In the formula, This represents the total reward value for the current period; This represents the organization's goal achievement rate, with a value ranging from 0 to 1. A higher value indicates a better degree of strategic goal achievement. This represents the performance dispute rate, with a value ranging from 0 to 1. The lower the value, the higher the acceptance of the performance results. This indicates employee engagement, with a value ranging from 0 to 1. A higher value indicates a stronger willingness to engage. These are positive real coefficients used to adjust the relative importance of the three indicators in the optimization process. The sum of the three is not required to be 1, but it must satisfy... .

5. The employee performance automatic appraisal and evaluation management system as described in claim 4, characterized in that: The calculation process for the organization's target achievement rate includes: Obtain the set of key outcome items for decomposing the company's annual strategic goals into each department; For each key outcome item, identify the employee group directly associated with it; Calculate the average performance score of this employee group during the current assessment period; The organizational goal achievement rate is obtained by weighting the average performance scores corresponding to each key outcome item according to their strategic importance. Strategic importance weights are set by the senior management team at the beginning of the quarter, reflecting the contribution of the key outcome to the overall strategy; if no weight is set for a key outcome, the default weight is 1. In the weighted average process, only employees who have completed data collection are included; employees with missing data are not included in the calculation.

6. The employee performance automatic appraisal and evaluation management system as described in claim 5, characterized in that: The process involves applying a personalized performance evaluation model to structured performance-related data and surrounding quantitative performance indicators to calculate the employee's comprehensive performance score and generate a performance evaluation report that includes explanations of dimensional contributions. Structured performance-related data is organized into a vector that includes dimensions such as sales achievement rate, work order processing timeliness, and customer satisfaction score. The surrounding performance quantitative indicators are organized into a vector that includes collaboration activity, knowledge contribution index, and innovation behavior score; Concatenate the two vectors into a single input vector; The weight vector of the dimensions output by the personalized performance evaluation model is used to sum the weighted values ​​of each dimension of the comprehensive input vector to obtain the comprehensive performance score. For each dimension, calculate its weighted value as a proportion of the overall performance score, which is taken as the contribution of that dimension. The names, raw values, weights, weighted values, and contributions of each dimension are written into the performance evaluation report for employees and managers to review. The overall performance score is not subject to mandatory distribution or ranking; it is only used as a reference for development feedback and incentive allocation. If a certain dimension's original value is missing during the report generation process, the weight of that dimension will be automatically reset to zero, and the weights of the other dimensions will be renormalized proportionally.

7. The employee performance automatic appraisal and evaluation management system as described in claim 6, characterized in that: The process of hashing key information such as performance target confirmation records, data collection snapshots, weight adjustment logs, and comprehensive performance scores before writing them into the blockchain to form an immutable, end-to-end evidence storage chain for performance evaluation includes: Perform a SHA-256 hash operation on the performance target confirmation record to generate the first hash value; the performance target confirmation record includes the employee's personal work plan, the approval opinion of the immediate supervisor, and the electronic signature timestamp of both parties; Perform a SHA-256 hash operation on the data collection snapshot to generate a second hash value; the data collection snapshot contains a complete copy of structured performance-related data and unstructured digital behavioral data at the assessment deadline; Perform a SHA-256 hash operation on the weight adjustment log to generate a third hash value; the weight adjustment log contains the dimension weight vector output by the reinforcement learning agent, the input state description, and the policy network version number; Perform a SHA-256 hash operation on the overall performance score to generate a fourth hash value; the overall performance score includes the final value and all input data identifiers used in the calculation. The four hash values ​​are concatenated into a single string in sequence, and the SHA-256 hash operation is performed on the string again to generate the final evidence hash. The final notarization hash, assessment period identifier, system version number, and timestamp will be written into the smart contract address of the consortium blockchain.

8. The employee performance automatic appraisal and evaluation management system as described in claim 7, characterized in that: The standardized interface connects to the enterprise's human resources information system, enterprise resource planning system, customer relationship management system, and project management platform. Based on a business event triggering mechanism, it collects structured performance-related data generated by employees in these systems in real time, including: Configure an event listener in the human resources information system to listen for events such as the resolution of employee attendance exceptions, completion of training courses, and effective job transfers. When an event occurs, the system's RESTful API is invoked to retrieve the employee ID associated with the event, the event type code, the time of occurrence, and the result status code. Configure event listeners in the enterprise resource planning system to listen for events such as production work order closure, procurement cost savings confirmation, and inventory count difference processing completion; after the event is triggered, extract the work order number, responsible person, and the amount of savings or difference rate quantification fields. Configure event listeners in the customer relationship management system to listen for events such as customer satisfaction survey submission, sales opportunity status change to "completed", and service request response timeout. Event data includes customer ID, employee ID, rating value, or status change time. Configure event listeners in the project management platform to listen for events such as task status updates to "completed", project milestones achieved, and delay risks resolved. Event information includes the task's unique identifier, the person in charge, and the difference between the actual completion time and the planned completion time. All event data is converted to a unified format and stored in a message queue, which is then consumed by the data acquisition service and written into the structured performance database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the employee performance automatic assessment and evaluation management system according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the employee performance automatic assessment and evaluation management system as described in any one of claims 1 to 7.

Images