Human resource management method and system based on artificial intelligence

Abstract

The invention discloses a human resource management method and system based on artificial intelligence, and relates to the technical field of human resource management, and the method comprises the following steps: guaranteeing personal data sovereignty through an employee side personal data insurance library and homomorphic encryption; the method comprises the following steps: constructing double agents which belong to employees and organizations and represent respective interests, and generating and signing an automatically executed initial bidirectional contract through anti-fact simulation and game theory negotiation; during the duration of the contract, the employee intelligent agent monitors the multivariate contribution in real time, and adjusts the contract terms according to a dynamic formula; dynamic value measurement and distribution of multi-element contribution are realized by combining causal reasoning interpretable analysis and a Shapley value algorithm through a federated learning cooperative training model integrating differential privacy and homomorphic encryption; by constructing a full-link privacy architecture, a bidirectional agent negotiation system and alliance chain cross-organization ecology, privacy and data utilization and organization and employee benefit balance are realized, talent ecology is constructed, and management intelligence and fairness are improved.

Description

Technical Field

[0001] This invention relates to the field of human resource management technology, and specifically to a human resource management method and system based on artificial intelligence. Background Technology

[0002] With the penetration of artificial intelligence technology into enterprise management, human resource management is gradually transforming towards digitalization and intelligence, giving rise to various management systems based on big data analysis and intelligent matching algorithms. However, existing technical solutions still remain at the level of process optimization or partial functional intelligence, failing to resolve the core contradictions in human resource management, such as privacy protection versus data utilization, balancing the interests of the organization and employees, and cross-organizational talent collaboration.

[0003] Existing human resource management systems suffer from inadequate privacy protection mechanisms and ambiguous data sovereignty. They generally employ centralized data storage architectures, requiring employee personal data (such as performance records, skill information, and payroll data) to be uploaded to corporate servers, which easily leads to data leaks and misuse risks. Furthermore, existing privacy protection measures are mostly passive (such as encrypted transmission), failing to achieve a precise balance between "usable but invisible" data. Employees lack autonomy over the authorization, tracking, and deletion of their personal data upon expiration, with data sovereignty entirely controlled by the company. This reduces employee trust in the system and limits the value extraction of fragmented data, creating a dilemma of data silos and privacy breaches.

[0004] Traditional human resource management techniques still adhere to a one-way management logic of "organization-led, employee-passive obedience." Even with the introduction of intelligent algorithms, they primarily focus on organizational needs such as task allocation and talent selection, lacking an autonomous decision-making body that represents employee interests. Contract terms are often fixed templates, making it difficult to adapt to dynamic scenarios such as changes in employee capabilities and adjustments to project requirements. Negotiation processes rely on manual communication, which is inefficient and susceptible to subjective biases, failing to achieve a precise balance between organizational goals and employee demands. Furthermore, contribution assessments often use single task indicators, leading to egalitarianism or subjective judgment in value distribution, which fails to motivate employees and carries a high risk of losing key talent.

[0005] Furthermore, existing human resource management systems are mostly confined to single-enterprise environments, failing to break down organizational boundaries and achieve efficient talent resource flow. In cross-organizational collaborations, employee skill certification relies on paper certificates or unilateral corporate endorsements, which lack credibility and are difficult to mutually recognize. The lack of a unified skill evidence storage and verification system leads to high costs and low efficiency in cross-organizational talent screening. At the same time, the system only focuses on internal human resource allocation and fails to build an industry-level talent skills ecosystem, limiting the maximization of talent value and failing to meet the needs of modern enterprise collaborative innovation for cross-organizational talent mobility. Summary of the Invention

[0006] The purpose of this invention is to provide an artificial intelligence-based human resource management method and system to address the problems of inadequate privacy protection, rigid decision-making paradigms, imbalance of interests, and lack of cross-organizational collaboration in existing technologies.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] According to the first aspect of this disclosure, an artificial intelligence-based human resource management method is proposed, comprising the following steps:

[0009] S1. By deploying a personal data vault on the employee side, homomorphic encryption algorithms are used to encrypt, store, and process personal data to ensure data sovereignty;

[0010] S2. Construct and run employee intelligent agents and organizational intelligent agents, wherein the employee intelligent agent runs on the employee side and represents the interests of the employee, and the organizational intelligent agent runs on the organization side and represents the interests of the organization;

[0011] S3. Through the interaction and negotiation between the employee intelligent agent and the organization intelligent agent, an initial two-way contract is generated and signed based on the counterfactual simulation and game theory model. The two-way contract is a smart contract that can be executed automatically.

[0012] S4. During the duration of the initial two-way contract, the employee's contribution data in multiple dimensions is monitored and calculated in real time by the employee's intelligent agent to form a multi-dimensional contribution evaluation result of the employee.

[0013] S5. Based on the multi-factor evaluation results, the terms of the two-way contract are automatically adjusted and updated according to the dynamic adjustment formula;

[0014] S6. By integrating a federated learning framework with differential privacy and homomorphic encryption, the decision-making models of the employee agent and the organization agent are collaboratively trained under the premise of privacy protection.

[0015] S7. Based on the causal reasoning engine, perform interpretable analysis on the two-way contract and the multi-factor contribution evaluation results, and based on the Shapley value algorithm, perform dynamic value measurement and allocation on the multi-factor contribution evaluation results.

[0016] Furthermore, in step S3, the process of generating and signing the initial bilateral contract includes the following steps:

[0017] S301. The employee agent generates a contract proposal based on local private information and uses zero-knowledge proof to verify the feasibility of the contract proposal to the organization agent.

[0018] S302. The organizational agent generates a counterfactual simulation based on the organizational goals to evaluate the impact of different contract schemes on the organizational objective function.

[0019] S303. Through multi-round Bayesian game negotiation, a Nash equilibrium solution that satisfies the constraints of both parties is reached between the employee agent and the organization agent.

[0020] S304. Automatically generate and sign the smart contract based on the Nash equilibrium solution.

[0021] Furthermore, in step S5, the dynamic adjustment formula for automatically adjusting and updating the terms of the bilateral contract is as follows:

[0022]

[0023] in, For employee intelligent agents With organizational intelligence Between in time The state vector of the established two-way contract includes work objectives, performance standards, and incentive coefficients; For employee intelligent agents In time The actual multi-faceted contribution assessment results; This is a preset threshold vector; This is an additional feedback signal vector from the external environment; Let be the gap function. and These are the corresponding weighting coefficients.

[0024] Furthermore, in step S6, the optimization objective of the federated learning framework is:

[0025]

[0026] in, Parameters for a shared knowledge representation model optimized for collaboration between employee and organizational agents; The total number of employee agents participating in collaborative learning; For the first The personal work dataset held by each employee agent; for The number of samples in the dataset; For dataset-based The calculated local loss function; For regularization hyperparameters; For regularization terms;

[0027] The execution process of the federated learning framework includes: each employee agent utilizing local resources... After training the shared knowledge representation model and obtaining its update, add the updated shared knowledge representation model to the updated model. Distributed noise is used to meet differential privacy requirements; and the federated learning coordinator uses a homomorphic encryption algorithm to securely aggregate the encrypted model updates from each employee agent to obtain the updated shared knowledge representation model parameters. .

[0028] Furthermore, in step S7, the multi-factor contribution evaluation results are dynamically valued and allocated based on the Shapley value algorithm, including the following steps:

[0029] S701. The fair value of each employee is calculated using the Shapley value algorithm, expressed as:

[0030]

[0031] in, A collection of all employee agents involved in value distribution; For a specific employee agent whose Shapley value is to be calculated, i.e., an employee whose fair value needs to be determined; For set It does not include employee intelligent agents. Any subset of represents a collaborative team; For subset The number of intelligent agents among employees; The total number of employee intelligent entities, i.e., the set Size; Let be the characteristic function, representing the subset of employee intelligent agents When a collaborative team is formed, the total value that the team can create is defined by the results of a multi-faceted contribution assessment. For employee intelligent agents Join the team Afterwards, new The total value that can be created; For employee intelligent agents For the cooperative team The marginal contribution, i.e., due to the employee's intelligent agent With the addition of the team The increase in total value; This is used to combine weights for all possible collaborating teams. The weighted average reflects the proportion of employee agents in all possible order of addition. The probability of adding at a specific position; For the calculated employee intelligent agent The Shapley value represents the fair share of value that the employee agent deserves throughout the organization's collaborations.

[0032] S702. Value allocation based on dynamic weights is expressed as follows:

[0033]

[0034] in, This can be considered a time step or allocation cycle, as value allocation is dynamic, with each cycle... After the period ends, the weights are recalculated based on the contributions made during that period. Represents employee intelligent agents In the cycle The contribution metric is obtained from the multivariate contribution evaluation results; It is an adjustable sensitivity parameter used to control the strength of the impact of contribution differences on the allocation weights; It is an exponential function; For the calculated employee intelligent agent In the next cycle The value allocation weight that should be obtained;

[0035] S703, Store value distribution records and contribution proofs on the blockchain.

[0036] According to the second aspect of this disclosure, an artificial intelligence-based human resource management system is proposed, employing the human resource management method of the first aspect, including:

[0037] The data sovereignty privacy computing unit includes a personal data vault deployed on the employee side for local storage and processing of personal data using homomorphic encryption algorithms; and a federated learning coordinator for coordinating a federated learning framework that integrates differential privacy and homomorphic encryption.

[0038] The intelligent agent decision negotiation unit includes multiple employee intelligent agents and at least one organizational intelligent agent. Each employee intelligent agent includes a contribution monitoring module for real-time monitoring and calculation of multivariate contribution evaluation results. The organizational intelligent agent includes an organizational objective function and a matching algorithm, as well as a two-way contract protocol engine for interactive negotiation and two-way contract generation between the employee intelligent agents and the organizational intelligent agent.

[0039] The cognitive value unit includes a causal reasoning engine for performing interpretable analysis of the bidirectional contract and the multi-factor contribution evaluation results based on a causal graph; and a value measurement and allocation module for performing dynamic value measurement and allocation based on the Shapley value algorithm.

[0040] Furthermore, the personal data insurance also includes:

[0041] The data usage authorization module is used to receive fine-grained authorization instructions from employees regarding data usage;

[0042] The data flow tracking module is used to record all data access and usage behaviors into the blockchain;

[0043] The data deletion execution module is used to automatically delete the corresponding data after the authorization expires.

[0044] Furthermore, the system also includes an ecological intelligent agent unit for managing cross-organizational talent mobility and skills ecosystem optimization, and interacting with the organizational intelligent agent;

[0045] The ecological intelligent agent unit includes a cross-organizational skills verification module, used for:

[0046] Receive cross-organizational skill certification requests submitted by the employee agent, wherein the cross-organizational skill certification requests include details of the skills to be certified and hash values ​​of the skill certification materials;

[0047] Invoke the preset cross-organization verification node manager to filter and authorize at least one verification node from the consortium blockchain according to the cross-organization skill certification request;

[0048] The verification results of the authorized authentication node on the hash value of the skill details and skill proof materials, the digital signature of the authentication node, and the hash value of the skill proof materials are stored together in the consortium blockchain to form an immutable skill evidence chain.

[0049] The authentication result, which includes the verification conclusion, is synchronously sent to the employee agent that initiated the request and the corresponding organizational agent.

[0050] Furthermore, the system also includes a transparent decision log unit and a dispute resolution coordination unit, wherein:

[0051] The transparent decision log unit is used to record the causal explanations of all key decisions to the blockchain;

[0052] The dispute resolution coordination unit is used to initiate a smart contract-based mediation process when negotiations fail.

[0053] Furthermore, the bidirectional contract protocol engine includes:

[0054] The contract generation module is used to generate and sign initial two-way contracts based on a game theory model.

[0055] The dynamic adjustment module is used to automatically adjust and update the terms of effective two-way contracts based on dynamic adjustment formulas.

[0056] Compared with existing technologies, the artificial intelligence-based human resource management method and system provided by this invention have the following beneficial effects:

[0057] 1. By constructing a full-chain privacy architecture, we achieve a unified approach to data privacy, security, sovereignty, and availability, avoiding the risk of leakage from centralized storage at the source and fully complying with data compliance requirements; employees have autonomous control over data authorization and deletion, significantly improving system trust; federated learning breaks down data silos, improving model decision-making accuracy while protecting privacy, and resolving the core contradiction between privacy and efficiency in traditional HR.

[0058] 2. By constructing an equal negotiation system between employee intelligent agents and organizational intelligent agents, and combining game theory and counterfactual simulation to achieve dynamic balance, the system automatically balances organizational goals and employee demands, avoiding subjective biases; it achieves autonomous management of the entire lifecycle of contracts, significantly reducing manual coordination costs; and it dynamically matches contributions with rewards, stimulating employee enthusiasm, reducing the turnover rate of core talents, and upgrading management from process-driven to intelligent collaboration-driven. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0060] Figure 1 A flowchart of an artificial intelligence-based human resource management method provided in an embodiment of the present invention;

[0061] Figure 2 A block diagram of an artificial intelligence-based human resource management system provided for an embodiment of the present invention. Detailed Implementation

[0062] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0063] Example 1:

[0064] This invention provides a human resource management method and system based on artificial intelligence, as shown in the appendix. Figure 1 As shown, it includes the following steps:

[0065] S1. By deploying a personal data vault on the employee side, homomorphic encryption algorithms are used to encrypt, store, and process personal data, ensuring data sovereignty. This achieves "local storage and encrypted processing" of personal data, giving employees full control over data usage. It mitigates the risk of data leakage at the source, addresses privacy violations caused by centralized data storage and enterprise-led use in traditional HR systems, and enhances employee trust in the system. Furthermore, homomorphic encryption supports data computation in encrypted mode, providing data support for subsequent contribution evaluation and model training without decryption, balancing privacy and data availability.

[0066] S2. Construct and run employee intelligent agents and organizational intelligent agents, where the employee intelligent agent operates on the employee side and represents employee interests, and the organizational intelligent agent operates on the organization side and represents organizational interests; by constructing autonomous decision-making entities representing the interests of both parties, two-way interaction with equal interests is achieved; the employee intelligent agent, as the spokesperson for employee interests, automatically safeguards employee demands (such as reasonable incentives and career development), avoiding damage to interests due to insufficient individual negotiation ability; the organizational intelligent agent makes rational decisions based on organizational goals, avoiding subjective biases in human management (such as preferential task allocation and unfair evaluation); the autonomous interaction between the two intelligent agents reduces the cost of manual coordination and improves the efficiency and objectivity of human resource management.

[0067] S3. Through interactive negotiation between employee intelligent agents and organizational intelligent agents, an initial two-way contract is generated and signed based on counterfactual simulation and game theory models. The two-way contract is a smart contract that can be executed automatically. The process of generating and signing the initial two-way contract includes the following steps:

[0068] S301. The employee intelligent agent generates a contract proposal based on local private information and uses zero-knowledge proof to verify the feasibility of the contract proposal with the organization intelligent agent; the zero-knowledge proof can verify the feasibility of the contract proposal without disclosing the employee's private information.

[0069] S302. The organizational agent generates counterfactual simulations based on organizational goals to evaluate the impact of different contractual schemes on the organizational objective function; the counterfactual simulations are used to assess the potential risks of the contractual schemes.

[0070] S303. Through multi-round Bayesian game negotiation, a Nash equilibrium solution that satisfies the constraints of both parties is reached between the employee agent and the organization agent; Bayesian game is used to conduct multi-round negotiation to balance the interests of both parties, thereby finding the Nash equilibrium solution;

[0071] S304. Based on the Nash equilibrium solution, smart contracts are automatically generated and signed to ensure that the contract terms are automatically executed.

[0072] By employing zero-knowledge proofs, counterfactual simulations, and Bayesian game theory, a fair contract acceptable to both parties is reached under the premise of privacy protection, which is then transformed into an automatically executable smart contract. The automatic execution feature of smart contracts reduces the cost of manual supervision and ensures that the agreement is executed immediately (such as automatically triggering incentives after achieving the target). The Bayesian game theory mechanism takes into account the constraints of both parties (such as the time cost of employees and the budget constraints of the organization), thereby improving the acceptance and execution rate of the contract.

[0073] S4. During the initial two-way contract period, employee intelligent agents monitor and calculate employee contribution data across multiple dimensions in real time, forming a multi-dimensional contribution evaluation result. This approach breaks through the traditional single indicator of task completion rate, collecting multi-dimensional employee contributions (such as task quality, innovative proposals, collaborative support, and knowledge transfer) in real time to form a comprehensive evaluation result that fully reflects employee value. Real-time monitoring ensures the timeliness of contribution data, providing real-time data support for subsequent dynamic adjustments to the contract and value allocation. Multi-dimensional evaluation guides employees to focus on improving comprehensive capabilities rather than just task completion, contributing to the overall improvement of organizational effectiveness.

[0074] S5. Based on the results of the multi-faceted contribution assessment, the terms of the bilateral contract are automatically adjusted and updated according to the dynamic adjustment formula; the dynamic adjustment formula for automatically adjusting and updating the terms of the bilateral contract is as follows:

[0075]

[0076] in, For employee intelligent agents With organizational intelligence Between in time The state vector of the established two-way contract includes work objectives, performance standards, and incentive coefficients; For employee intelligent agents In time The actual multi-faceted contribution assessment results; This is a preset threshold vector; This is an additional feedback signal vector from the external environment; Let be the gap function. and These are the corresponding weighting coefficients.

[0077] Based on multi-faceted contribution assessment results and dynamic adjustment formulas, the system automatically updates contract terms (work objectives, performance standards, incentive coefficients), allowing contracts to adapt to actual work changes. This solves the problem of rigidity and lag in traditional fixed contracts (such as incentives not being adjusted synchronously after employee capabilities improve), achieving dynamic matching of contribution and reward. Its formulaic adjustments ensure that the adjustment logic is transparent and quantifiable, avoiding the subjectivity of manual adjustments. Its additional feedback signals can adapt to changes in the external environment (such as adjustments to project urgency and updates to industry standards), improving the adaptability of the contract.

[0078] S6. By integrating a federated learning framework with differential privacy and homomorphic encryption, the decision-making models of employee agents and organizational agents are collaboratively trained under the premise of privacy protection; the optimization objective of the federated learning framework is:

[0079]

[0080] in, Parameters for a shared knowledge representation model optimized for collaboration between employee and organizational agents; The total number of employee agents participating in collaborative learning; For the first The personal work dataset held by each employee agent; for The number of samples in the dataset; For dataset-based The calculated local loss function; For regularization hyperparameters; For regularization terms;

[0081] The execution process of the federated learning framework includes: each employee agent utilizing local resources... After training the shared knowledge representation model and obtaining its update, add the updated shared knowledge representation model information. Distributed noise is used to meet differential privacy requirements; and the federated learning coordinator uses a homomorphic encryption algorithm to securely aggregate the encrypted model updates from each employee agent to obtain the updated shared knowledge representation model parameters. .

[0082] This allows for the aggregation of the value of dispersed data from all employees without disclosing their local raw data, enabling collaborative training of the intelligent agent decision-making model and effectively improving the model's generalization ability and decision-making accuracy. Its dual protection of differential privacy and homomorphic encryption ensures that data privacy is not leaked during the training process.

[0083] S7. Based on the causal reasoning engine, perform interpretable analysis on the results of the two-way contract and the multi-factor contribution evaluation, and based on the Shapley value algorithm, perform dynamic value measurement and allocation on the multi-factor contribution evaluation results, including the following steps:

[0084] S701. The fair value of each employee is calculated using the Shapley value algorithm, expressed as:

[0085]

[0086] in, A collection of all employee agents involved in value distribution; For a specific employee agent whose Shapley value is to be calculated, i.e., an employee whose fair value needs to be determined; For set It does not include employee intelligent agents. Any subset of represents a collaborative team; For subset The number of intelligent agents among employees; The total number of employee intelligent entities, i.e., the set Size; Let be the characteristic function, representing the subset of employee intelligent agents When a collaborative team is formed, the total value that the team can create is defined by the results of a multi-faceted contribution assessment. For employee intelligent agents Join the team Afterwards, new The total value that can be created; For employee intelligent agents For the cooperative team The marginal contribution, i.e., due to the employee's intelligent agent With the addition of the team The increase in total value; This is used to combine weights for all possible collaborating teams. The weighted average reflects the proportion of employee agents in all possible order of addition. The probability of adding at a specific position; For the calculated employee intelligent agent The Shapley value represents the fair share of value that the employee agent deserves throughout the organization's collaborations.

[0087] S702. Value allocation based on dynamic weights is expressed as follows:

[0088]

[0089] in, This can be considered a time step or allocation cycle, as value allocation is dynamic, with each cycle... After the period ends, the weights are recalculated based on the contributions made during that period. Represents employee intelligent agents In the cycle The contribution metric is derived from the multivariate contribution assessment results; It is an adjustable sensitivity parameter used to control the strength of the impact of contribution differences on the allocation weights; It is an exponential function; For the calculated employee intelligent agent In the next cycle The value allocation weight that should be obtained;

[0090] S703, Store value distribution records and contribution proofs on the blockchain.

[0091] Using causal reasoning improves decision-making transparency and reduces employee questioning of decisions; the Shapley value algorithm takes into account both individual contributions and team collaboration value, solving the unfairness problems of traditional egalitarianism or single-indicator dominance in distribution, while blockchain notarization ensures that distribution records are tamper-proof and traceable, enhancing the credibility of distribution results.

[0092] The overall technical solution transforms human resource management from "process-driven" to "ecosystem-driven," and from "standardized management" to "personalized symbiosis," truly achieving the co-evolution of the organization and its employees. Its core design philosophy is to enhance, rather than replace, human judgment, and to empower, rather than control, employee development, thereby maximizing the potential of each individual while improving organizational effectiveness.

[0093] Example 2:

[0094] This invention also provides an artificial intelligence-based human resource management system, applied in Embodiment 1, as shown in the appendix. Figure 2 As shown, it includes:

[0095] The data sovereignty and privacy computing unit includes a personal data vault deployed on the employee side for local storage and processing of personal data using homomorphic encryption algorithms; and a federated learning coordinator for coordinating a federated learning framework integrating differential privacy and homomorphic encryption to manage the distributed model training process. This federated learning coordinator improves the generalization ability and accuracy of the agent decision-making model by aggregating the value of the dispersed data from all employees. Its differential privacy and homomorphic encryption technologies provide dual protection to ensure that the original data is not leaked during training. This achieves unified coordination of the training rhythm of multi-employee agents, avoids model fragmentation, and ensures the consistency and stability of the global model.

[0096] The intelligent agent decision negotiation unit includes multiple employee intelligent agents and at least one organizational intelligent agent. Each employee intelligent agent includes a contribution monitoring module for real-time monitoring and calculation of multivariate contribution evaluation results. The organizational intelligent agent includes an organizational objective function and matching algorithm, as well as a two-way contract protocol engine for interactive negotiation and two-way contract generation between employee intelligent agents and organizational intelligent agents.

[0097] The employee intelligent agent represents the interests of employees, autonomously participates in contract negotiation, monitors contribution data in real time, and safeguards employee demands. It automatically protects employee interests, avoids damage to rights due to insufficient individual negotiation ability, reduces the cost of manual communication between employees and the organization, and improves management efficiency. The organizational intelligent agent represents the interests of the organization, makes rational decisions based on organizational goals (such as performance improvement, cost control, and talent development), participates in contract negotiation and talent matching, and is used to quantify the organization's core demands in the organizational objective function. It avoids the subjective bias of manual decision-making, achieves precise matching of talent, tasks, and teams, improves the efficiency of organizational resource allocation, and negotiates with the employee intelligent agent on an equal footing to balance organizational interests and employee demands, thereby reducing the risk of contract disputes.

[0098] The cognitive value unit includes a causal reasoning engine for interpretable analysis of two-way contracts and multi-faceted contribution assessment results based on causal graphs; and a value measurement and allocation module for performing dynamic value measurement and allocation based on the Shapley value algorithm. Through causal graphs (variables: contract terms, contribution dimensions, organizational goals, incentive results), it interpretably analyzes the rationality of two-way contracts and the influencing factors of multi-faceted contribution assessment results. This helps identify key influencing factors in management (such as the "true causal effect of training terms on contribution improvement"), providing data support for contract optimization and management strategy adjustments, thereby improving decision-making transparency, reducing employee skepticism, and enhancing trust.

[0099] This value measurement and allocation module performs dynamic value measurement and allocation based on the Shapley value algorithm, including marginal contribution calculation, dynamic weight adjustment, and blockchain notarization. Its Shapley value algorithm can accurately quantify the individual contribution of employees and the collaborative value of the team, solving the unfair problems of traditional egalitarianism and single indicator dominance. By dynamically adjusting the weights to adapt to the real-time changes in employee contributions, it realizes the principle of more pay for more work and better pay for better work.

[0100] Ecological intelligent agent unit, used to manage cross-organizational talent mobility and skills ecosystem optimization, and interact with organizational intelligent agents;

[0101] The ecological intelligent agent unit includes a cross-organizational skills verification module, used for:

[0102] Receive cross-organizational skill certification requests submitted by employee agents. The cross-organizational skill certification requests include details of the skills to be certified and hash values ​​of the skill certification materials.

[0103] Invoke the preset cross-organization verification node manager to filter and authorize at least one verification node from the consortium blockchain based on the cross-organization skill certification request;

[0104] The verification results of the authorized certification node on the hash value of the skill details and skill proof materials, the digital signature of the certification node and the hash value of the skill proof materials are stored together in the consortium blockchain to form an immutable skill evidence chain.

[0105] The authentication result, including the verification conclusion, is synchronized to the employee agent that initiated the request and the corresponding organizational agent.

[0106] This ecological intelligent agent unit breaks down organizational boundaries, builds a cross-enterprise talent and skills ecosystem, improves the utilization rate of human resources, reduces the cost of talent screening for cross-organizational cooperation with its credible skill evidence chain, and promotes collaborative innovation.

[0107] The transparent decision log unit is used to record the causal explanations, decision-making basis, and execution results of all key decisions (contract generation, clause adjustment, value allocation, skills certification), and store them on the blockchain; making the entire decision-making process traceable and auditable, while the causal explanation records help employees and organizations understand the decision-making logic and reduce disputes.

[0108] The dispute resolution coordination unit is used to initiate a smart contract-based mediation process when negotiations fail. When employee agents and organizational agents fail to reach an agreement (e.g., disagreement over contract terms or disagreement on contribution assessment results), a smart contract-based mediation process is initiated. By pre-setting mediation rules (e.g., introducing third-party authentication nodes and fair arbitration based on historical data), prolonged disputes are avoided; and by automatically executing mediation results through smart contracts (e.g., adjusting contract terms and recalculating contribution values), dispute resolution efficiency is improved; thereby reducing human intervention and lowering the time and communication costs of dispute resolution.

[0109] Personal data insurance also includes:

[0110] The data usage authorization module is used to receive fine-grained authorization instructions from employees regarding data usage; it enables precise authorization and expiration of data usage, preventing data misuse and fully complying with personal information protection requirements; employees have autonomy over their data usage rights, enhancing system trust.

[0111] The data flow tracking module records all data access and usage behaviors into the blockchain, making the entire data flow traceable and tamper-proof, and enabling rapid identification of the responsible party in the event of a privacy breach; it also meets compliance audit requirements and reduces data management risks.

[0112] The data deletion execution module is used to automatically delete the corresponding data after the authorization expires, avoiding the risk of data residue after the authorization expires.

[0113] The two-way contract protocol engine includes:

[0114] The contract generation module, based on a Bayesian game model, supports multi-round negotiation between employee agents and organizational agents to generate and sign initial bidirectional smart contracts; it realizes a closed loop of contract generation from privacy protection (zero-knowledge proof) to risk assessment (counterfactual simulation) to interest balancing (Nash equilibrium) to automatic execution (smart contract); thus improving the fairness and enforceability of contracts.

[0115] The dynamic adjustment module, based on a dynamic adjustment formula and combined with multi-faceted contribution evaluation results and external feedback, automatically updates contract terms (work objectives, performance standards, incentive coefficients); it solves the rigidity problem of traditional fixed contracts and achieves dynamic matching of contribution and reward; its formulaic adjustment ensures logical transparency and quantifiability, avoiding the subjectivity of manual adjustment.

[0116] The system's overall bidirectional symbiotic intelligent agent decision-making and negotiation system overturns the traditional "organization-led, employee-passive" management paradigm. By constructing an equal decision-making system for employee and organizational intelligent agents, and using technologies such as game theory and counterfactual simulation, it achieves a dynamic balance between organizational interests and employee demands. The autonomous negotiation of intelligent agents, the automatic generation and dynamic adjustment of contracts upgrade human resource management from process-driven to intelligent collaborative-driven, significantly reducing manual intervention. Furthermore, by introducing ecological intelligent agent units and cross-organizational skill verification modules, and building an immutable skill evidence chain based on a consortium blockchain, it achieves cross-organizational skill mutual recognition and talent mobility. It breaks through the management boundaries of a single organization, upgrading the system from an enterprise-level HR tool to an industry-level talent ecosystem platform, improving the efficiency and utilization of talent resource allocation. The various technical modules are not simply superimposed but deeply collaborative (e.g., federated learning provides data support for intelligent agent decision-making, blockchain provides evidence for causal explanation and value allocation, and game theory provides the logical basis for contract generation), resolving core contradictions that traditional HR systems cannot balance, such as privacy and efficiency, fairness and flexibility, and individual and organizational dynamics.

[0117] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A human resource management method based on artificial intelligence, characterized in that, Includes the following steps: S1. By deploying a personal data vault on the employee side, homomorphic encryption algorithms are used to encrypt, store, and process personal data to ensure data sovereignty; S2. Construct and run employee intelligent agents and organizational intelligent agents, wherein the employee intelligent agent runs on the employee side and represents the interests of the employee, and the organizational intelligent agent runs on the organization side and represents the interests of the organization; S3. Through the interaction and negotiation between the employee intelligent agent and the organization intelligent agent, an initial two-way contract is generated and signed based on the counterfactual simulation and game theory model. The two-way contract is a smart contract that can be executed automatically. S4. During the duration of the initial two-way contract, the employee's contribution data in multiple dimensions is monitored and calculated in real time by the employee's intelligent agent to form a multi-dimensional contribution evaluation result of the employee. S5. Based on the multi-factor evaluation results, the terms of the two-way contract are automatically adjusted and updated according to the dynamic adjustment formula; S6. By integrating a federated learning framework with differential privacy and homomorphic encryption, the decision-making models of the employee agent and the organization agent are collaboratively trained under the premise of privacy protection. S7. Based on the causal reasoning engine, perform interpretable analysis on the two-way contract and the multi-factor contribution evaluation results, and based on the Shapley value algorithm, perform dynamic value measurement and allocation on the multi-factor contribution evaluation results.

2. The artificial intelligence-based human resource management method according to claim 1, characterized in that, In step S3, the process of generating and signing the initial two-way contract includes the following steps: S301. The employee agent generates a contract proposal based on local private information and uses zero-knowledge proof to verify the feasibility of the contract proposal to the organization agent. S302. The organizational agent generates a counterfactual simulation based on the organizational goals to evaluate the impact of different contract schemes on the organizational objective function. S303. Through multi-round Bayesian game negotiation, a Nash equilibrium solution that satisfies the constraints of both parties is reached between the employee agent and the organization agent. S304. Automatically generate and sign the smart contract based on the Nash equilibrium solution.

3. The human resource management method based on artificial intelligence according to claim 1, characterized in that, In step S5, the dynamic adjustment formula for automatically adjusting and updating the terms of the two-way contract is as follows: in, For employee intelligent agents With organizational intelligence Between in time The state vector of the established two-way contract includes work objectives, performance standards, and incentive coefficients; For employee intelligent agents In time The actual multi-faceted contribution assessment results; This is a preset threshold vector; This is an additional feedback signal vector from the external environment; Let be the gap function. and These are the corresponding weighting coefficients.

4. The artificial intelligence-based human resource management method according to claim 1, characterized in that, In step S6, the optimization objective of the federated learning framework is: in, Parameters for a shared knowledge representation model optimized for collaboration between employee and organizational agents; The total number of employee agents participating in collaborative learning; For the first The personal work dataset held by each employee agent; for The number of samples in the dataset; For dataset-based The calculated local loss function; For regularization hyperparameters; For regularization terms; The execution process of the federated learning framework includes: each employee agent utilizing local resources... After training the shared knowledge representation model and obtaining the updated shared knowledge representation model, add the updated shared knowledge representation model to the updated shared knowledge representation model. Distributed noise is used to meet differential privacy requirements; and the federated learning coordinator uses a homomorphic encryption algorithm to securely aggregate the encrypted model updates from each employee agent to obtain the updated shared knowledge representation model parameters. .

5. The artificial intelligence-based human resource management method according to claim 1, characterized in that, In step S7, the multi-factor contribution evaluation results are dynamically valued and allocated based on the Shapley value algorithm, including the following steps: S701. The fair value of each employee is calculated using the Shapley value algorithm, expressed as: in, A collection of all employee agents involved in value distribution; For a specific employee agent whose Shapley value is to be calculated, i.e., an employee whose fair value needs to be determined; For set It does not include employee intelligent agents. Any subset of represents a collaborative team; For subset The number of intelligent agents among employees; The total number of employee intelligent entities, i.e., the set Size; Let be the characteristic function, representing the subset of employee intelligent agents When a collaborative team is formed, the total value that the team can create is defined by the results of a multi-faceted contribution assessment. For employee intelligent agents Join the team Afterwards, the new The total value that can be created; For employee intelligent agents For the cooperative team The marginal contribution, i.e., due to the employee's intelligent agent With the addition of the team The increase in total value; This is used to combine weights for all possible collaborating teams. The weighted average reflects the proportion of employee agents in all possible order of addition. The probability of adding at a specific position; For the calculated employee intelligent agent The Shapley value represents the fair share of value that the employee agent deserves throughout the organization's collaborations. S702. Value allocation based on dynamic weights is expressed as follows: in, This can be considered a time step or allocation cycle, as value allocation is dynamic, with each cycle... After the period ends, the weights are recalculated based on the contributions made during that period. Represents employee intelligent agents In the cycle The contribution metric is obtained from the multivariate contribution evaluation results; It is an adjustable sensitivity parameter used to control the strength of the impact of contribution differences on the allocation weights; It is an exponential function; For the calculated employee intelligent agent In the next cycle The value allocation weight that should be obtained; S703, Store value distribution records and contribution proofs on the blockchain.

6. An artificial intelligence-based human resource management system, employing the human resource management method as described in any one of claims 1-5, characterized in that, include: The data sovereignty and privacy computing unit includes a personal data vault deployed on the employee side for local storage and processing of personal data using homomorphic encryption algorithms; And a federated learning coordinator for coordinating federated learning frameworks that integrate differential privacy and homomorphic encryption; The intelligent agent decision negotiation unit includes multiple employee intelligent agents and at least one organizational intelligent agent. Each employee intelligent agent includes a contribution monitoring module for real-time monitoring and calculation of multivariate contribution evaluation results. The organizational intelligent agent includes an organizational objective function and a matching algorithm, as well as a two-way contract protocol engine for interactive negotiation and two-way contract generation between the employee intelligent agents and the organizational intelligent agent. The cognitive value unit includes a causal reasoning engine for performing interpretable analysis of the bidirectional contract and the multi-factor contribution evaluation results based on a causal graph. And a value measurement and allocation module, used to perform dynamic value measurement and allocation based on the Shapley value algorithm.

7. The artificial intelligence-based human resource management system according to claim 6, characterized in that, The personal data insurance also includes: The data usage authorization module is used to receive fine-grained authorization instructions from employees regarding data usage; The data flow tracking module is used to record all data access and usage behaviors into the blockchain; The data deletion execution module is used to automatically delete the corresponding data after the authorization expires.

8. A human resource management system based on artificial intelligence according to claim 6, characterized in that, The system also includes an ecological intelligent agent unit for managing cross-organizational talent mobility and skills ecosystem optimization, and interacting with the organizational intelligent agent; The ecological intelligent agent unit includes a cross-organizational skills verification module, used for: Receive cross-organizational skill certification requests submitted by the employee agent, wherein the cross-organizational skill certification requests include details of the skills to be certified and hash values ​​of the skill certification materials; Invoke the preset cross-organization verification node manager to filter and authorize at least one verification node from the consortium blockchain according to the cross-organization skill certification request; The verification results of the authorized authentication node on the hash value of the skill details and skill proof materials, the digital signature of the authentication node, and the hash value of the skill proof materials are stored together in the consortium blockchain to form an immutable skill evidence chain. The authentication result, which includes the verification conclusion, is synchronously sent to the employee agent that initiated the request and the corresponding organizational agent.

9. A human resource management system based on artificial intelligence according to claim 6, characterized in that, The system also includes a transparent decision log unit and a dispute resolution coordination unit, wherein: The transparent decision log unit is used to record the causal explanations of all key decisions to the blockchain; The dispute resolution coordination unit is used to initiate a smart contract-based mediation process when negotiations fail.

10. A human resource management system based on artificial intelligence according to claim 6, characterized in that, The bidirectional contract protocol engine includes: The contract generation module is used to generate and sign initial two-way contracts based on a game theory model. The dynamic adjustment module is used to automatically adjust and update the terms of effective two-way contracts based on dynamic adjustment formulas.

Images