Human resource authority management method and system

Abstract

The invention belongs to the technical field of authority management, and discloses a human resource authority management method and system. The method comprises the following steps: in a cloud data center, constructing a behavior pattern baseline model, a user portrait generation model and a dynamic permission adjustment model; using a data acquisition device to acquire a real-time service mode and real-time behavior data of the user, and uploading the real-time service mode and the real-time behavior data to a cloud data center; extracting real-time behavior data characteristics of the real-time behavior data, and performing abnormal behavior analysis on the real-time behavior data characteristics by using a behavior mode baseline model according to a real-time service mode; and generating a real-time user portrait by using a user portrait generation model according to the real-time behavior data characteristics, and generating real-time human resource permission data by using a permission dynamic adjustment model according to the real-time user portrait. According to the invention, the problems of low practicability, large labor cost investment, low efficiency and low intelligent degree in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of authority management, and in particular relates to a human resources authority management method and system. Background Art

[0002] Human resource rights management refers to the process of setting and controlling user rights in a human resource information system or human resource management system in an enterprise or organization. This is an important part of ensuring data security, employee privacy protection, and compliance with relevant laws and regulations.

[0003] The existing human resources rights management technology has the following defects: 1) Existing technologies usually use static permission management, that is, once permission configuration is set, it is rarely changed. This cannot adapt to the needs of changes in employee responsibilities or organizational structure adjustments, resulting in permissions that may not match the actual needs of users and low practicality; 2) The existing technology usually requires administrators to adjust permissions manually, which requires a lot of manpower. This process is time-consuming and error-prone, especially in large organizations, resulting in low efficiency of human resource permission management. 3) The existing technology has a low level of intelligence and lacks the functions and technologies of real-time monitoring of user behavior, abnormal analysis, and automatic adjustment of permissions, and cannot meet the technical requirements of this technical field. Summary of the invention

[0004] In order to solve the problems of low practicality, high manpower cost, low efficiency and low intelligence in the prior art, the present invention aims to provide a human resource authority management method and system.

[0005] The technical solution adopted by the present invention is: A human resources authority management method comprises the following steps: In the cloud data center, we build a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model. Using a data collection device to collect real-time business model and real-time behavior data of users, and uploading the real-time business model and real-time behavior data to a cloud data center; In the cloud data center, the real-time behavior data features are extracted, and according to the real-time business model, the behavior pattern baseline model is used to perform abnormal behavior analysis on the real-time behavior data features; Based on the characteristics of real-time behavioral data, use the user portrait generation model to generate real-time user portraits. Based on the real-time user portraits, use the authority dynamic adjustment model to generate real-time human resources authority data.

[0006] Furthermore, in the cloud data center, a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model are constructed, including the following steps: In the cloud data center, a number of historical behavior data of users under different historical business modes are collected, and the number of historical behavior data are preprocessed to obtain a number of preprocessed historical behavior data; Based on several pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a group intelligence optimized logistic regression algorithm, and several historical behavior data features are generated; Based on several historical behavior data features, a deep learning algorithm is used to build a user portrait generation model and generate several historical user portraits; Based on several historical user portraits, an enhanced reinforcement learning algorithm is used to build a dynamic permission adjustment model, and generate several historical dynamic permission adjustment experiences.

[0007] Furthermore, the behavioral pattern baseline model is constructed based on the RF-ICPO-LR algorithm; The deep learning algorithm is built on the LSTM-Elman algorithm; The dynamic permission adjustment model is built based on the ML-DQN algorithm.

[0008] Furthermore, based on a number of pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a swarm intelligence optimized logistic regression algorithm, and a number of historical behavior data features are generated, including the following steps: According to the pre-processed historical behavior data under each historical business model, the corresponding historical behavior data features are extracted using the CNN network; Use the RF algorithm to obtain the importance scores of several feature components in each historical behavior data feature, and select several historical key feature components based on the importance scores; Based on several historical key feature components, the LR algorithm is used to construct the initial behavior pattern baseline model under each historical business model; Use the ICPO algorithm to optimize the initial historical model parameters of the initial behavior pattern baseline model to obtain the optimized behavior pattern baseline model under each historical business model; According to a number of pre-processed historical behavior data, the optimized behavior pattern baseline model is optimized and trained to obtain a final behavior pattern baseline model, and a number of historical behavior data features are generated.

[0009] Furthermore, based on several historical behavior data features, a deep learning algorithm is used to construct a user portrait generation model, and several historical user portraits are generated, including the following steps: Divide several historical behavior data features to obtain pre-training sets, training sets, and test sets, and use the LSTM-Elman algorithm to build an initial user portrait generation model; Using the pre-training set, pre-train the classifier built based on the Elman algorithm in the initial user portrait generation model to obtain a pre-trained user portrait generation model; Use the training set to optimize the pre-trained user portrait generation model to obtain an optimized user portrait generation model; Use the test set to test the optimized user portrait generation model. If the model test results meet the requirements, the final user portrait generation model is output; otherwise, continue the optimization training.

[0010] Furthermore, based on several historical user portraits, an enhanced reinforcement learning algorithm is used to build a dynamic permission adjustment model and generate several historical permission dynamic adjustment experiences, including the following steps: The permission dynamic adjustment problem is used as the simulation environment of the DQN algorithm, the permission dynamic adjustment problem is divided into several subtasks, and meta-feature parameters are defined in the historical user portrait; According to the meta-feature parameters, several historical meta-features of historical user portraits are extracted, and according to several historical meta-features, ML algorithms are used for training under several subtasks to obtain an initial meta-learning module and several historical meta-learning experiences; Initialize the deep Q network of the DQN algorithm using the initial meta-learning module, and update the initial meta-learning module to obtain an updated meta-learning module; Set the environment state, action space, and reward function of the DQN algorithm, build the agent and experience replay pool, and obtain the initial permission dynamic adjustment model; Based on several historical user portraits, the initial permission dynamic adjustment model is trained and optimized to obtain the final permission dynamic adjustment model and several historical permission dynamic adjustment experiences; Store several historical meta-learning experiences and several historical permission dynamic adjustment experiences into the experience replay pool of the final permission dynamic adjustment model.

[0011] Furthermore, in the cloud data center, the real-time behavior data features of the real-time behavior data are extracted, and according to the real-time business model, the behavior pattern baseline model is used to perform abnormal behavior analysis on the real-time behavior data features, including the following steps: In the cloud data center, a CNN network is used to extract the real-time behavior data features of the real-time behavior data; According to the real-time business model, several real-time key feature components of the real-time behavior data characteristics are extracted, and the corresponding target behavior pattern baseline model is matched; Use the target behavior pattern baseline model to perform abnormal behavior analysis on several real-time key feature components to obtain real-time abnormal behavior analysis results; If the real-time abnormal behavior analysis result shows that abnormal behavior exists, the connection of the data acquisition device is disconnected, and a real-time abnormal behavior alarm is sent to the corresponding data acquisition device.

[0012] Further, based on the real-time behavior data characteristics, a user portrait generation model is used to generate a real-time user portrait, and based on the real-time user portrait, a permission dynamic adjustment model is used to generate real-time human resource permission data, including the following steps: Input the real-time behavior data features into the user portrait generation model, and use the user portrait generation model to generate a real-time user portrait; According to the real-time user portrait, the meta-learning module is used to adjust the permission dynamic adjustment model to obtain the adjusted permission dynamic adjustment model; According to the real-time user portrait, the adjusted permission dynamic adjustment model is used to generate a permission dynamic adjustment strategy to obtain real-time human resources permission data.

[0013] Furthermore, according to the real-time user portrait, the meta-learning module is used to adjust the permission dynamic adjustment model to obtain an adjusted permission dynamic adjustment model, which includes the following steps: According to the meta-feature parameters, several real-time meta-features of the real-time user portrait are extracted, and according to the real-time meta-features, corresponding real-time meta-learning experiences are matched in several historical meta-learning experiences in the experience replay pool; Based on the real-time meta-learning experience, a meta-learning module is used to generate real-time network parameters of the deep Q network of the permission dynamic adjustment model; According to the real-time network parameters, the deep Q network of the authority dynamic adjustment model is adjusted to obtain the adjusted authority dynamic adjustment model.

[0014] A human resources authority management system is used to implement a human resources authority management method. The system includes a cloud data center and several data acquisition devices. The several data acquisition devices are respectively communicated with the cloud data center. The cloud data center includes a model building unit, an abnormal behavior analysis unit, and a portrait generation and authority adjustment unit that are connected in sequence.

[0015] The beneficial effects of the present invention are: The present invention provides a human resource authority management method and system, and proposes a dynamic authority adjustment mechanism, which can automatically adjust the authority according to the real-time behavior data characteristics of the user, solves the problem of lack of flexibility of static authority management in the prior art, and improves practicality; by collecting the user's business model and behavior data in real time, a real-time user portrait is generated in the cloud data center, which helps to more accurately reflect the user's current status and needs, thereby realizing personalized authority management and improving the efficiency and accuracy of dynamic authority adjustment; a behavior pattern baseline model is constructed for abnormal behavior analysis of real-time behavior data characteristics, thereby improving the intelligence level, security and prevention of potential risks of authority management; a dynamic authority adjustment model is constructed, combining the advantages of meta-learning and reinforcement learning, so that the model can quickly adapt to new user portraits and automatically adjust human resource authorities, further improving the intelligence level while avoiding manual adjustment of human resource authorities, reducing labor cost investment, and improving user experience and management efficiency.

[0016] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the human resources authority management method in the present invention.

[0018] Figure 2 It is a structural diagram of the human resources authority management system in the present invention. DETAILED DESCRIPTION

[0019] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0020] Embodiment 1: like Figure 1 As shown, this embodiment provides a human resources authority management method, including the following steps: S1: In the cloud data center, build a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model, including the following steps: S1-1: In the cloud data center, a number of historical behavior data of the user under different historical business modes are collected, and the number of historical behavior data are preprocessed to obtain a number of preprocessed historical behavior data; Behavioral data includes login logs, operation records, working hours, task completion status and other data. Preprocessing includes data cleaning, format conversion and normalization. While eliminating duplicate and erroneous data, standardization conversion is performed to eliminate magnitude differences and noise, and convert it into a format that can be recognized by subsequent models. S1-2: Based on a number of pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a swarm intelligence optimized logistic regression algorithm, and a number of historical behavior data features are generated; The behavioral pattern baseline model was constructed based on the Random Forest (RF)-Improved Crested Porcupine Optimizer (ICPO)-Logistic Regression (LR) algorithm; The RF algorithm screens the feature components of the input behavior data through the internal Classification and Regression Tree (CART), which can process a large number of feature components and select the most stable and discriminative key feature components; the logistic regression algorithm is one of the commonly used methods for establishing a baseline model, especially when it is necessary to predict binary results. The logistic regression model can provide a probability estimate of each observation belonging to a certain category, which makes it very suitable for establishing a behavior pattern baseline. The ICPO algorithm optimizes the model parameters of the logistic regression algorithm, provides the optimal initial model parameters, and improves the accuracy and training rate of the behavior pattern baseline model; Based on several pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a swarm intelligence optimized logistic regression algorithm, and several historical behavior data features are generated, including the following steps: S1-2-1: Based on the pre-processed historical behavior data under each historical business model, a convolutional neural network (CNN) is used to extract the corresponding historical behavior data features; S1-2-2: Use the RF algorithm to obtain the importance scores of several feature components in each historical behavior data feature, and select several historical key feature components based on the importance scores, including the following steps: S1-2-2-1: Use the CART tree trained by the RF algorithm to extract the feature contribution of several feature components in the historical behavior data features; The formula is:

[0021] In the formula, For the Feature contribution of feature components; For the Feature components in random forest The feature contribution of each tree; is the CART tree indicator; is the characteristic component indicator; is the total number of CARTs;

[0022] In the formula, CART tree node for random forest m ,node and nodes The Gini index of CART tree node m Medium Category The proportion of is the total number of categories; m , , is the node indicator; is the category indicator; S1-2-2-2: normalizing the feature contributions of several feature components to obtain corresponding normalized feature contributions; The formula is:

[0023] In the formula, It is the contribution of the feature after normalization; J is the total number of characteristic components; S1-2-2-3: Generate feature selection standard values ​​for several feature components based on the normalized feature contribution; The formula is:

[0024] In the formula, For the Feature selection criterion values ​​for feature components; For the The normalized feature contribution of the feature components; is the characteristic component indicator; S1-2-2-4: The feature selection standard value of the feature component is used as the importance score, and according to the importance score, the first M feature components are selected as the historical key feature components; the key feature components include login features, operation features, working time features, task completion features, etc.; S1-2-3: Based on several historical key feature components, use the LR algorithm to build an initial behavior pattern baseline model under each historical business model; The formula is:

[0025] In the formula, For the Under the historical business model, given the historical behavior data characteristics The behavior pattern baseline function; is the intercept term; is the coefficient of each historical key characteristic component; are M historical key feature components; It is an indicator of the historical key characteristic component; is the total number of historical key characteristic components; is a natural constant; An indicator of historical business patterns; S1-2-4: Use the ICPO algorithm to optimize the initial historical model parameters of the initial behavior pattern baseline model to obtain the optimized behavior pattern baseline model under each historical business model, including the following steps: S1-2-4-1: Taking minimizing the prediction error of the initial behavior pattern baseline model as the optimization goal, set the fitness function of the ICPO algorithm, and set the encoding vector of the ICPO individual, ICPO population parameters and the maximum number of iterations; The formula for the fitness value is:

[0026] In the formula, is the fitness function; is the mean square error function; For the The true value of samples; For the The predicted value of samples; is the sample indicator amount; N is the total number of samples; S1-2-4-2: According to the coding vector of the ICPO individual and the ICPO population parameters, the Circle chaotic mapping sequence is used for initialization to generate several initial ICPO individuals of the initial ICPO population; The formula is:

[0027] In the formula, is the initial ICPO individual of the Circle chaos map, i.e. the initial solution; is the randomly generated initial ICPO individual; It is the ICPO individual indicator; S1-2-4-3: Introduce a cyclic population reduction mechanism to limit the number of individuals in the ICPO population parameters and obtain updated ICPO population parameters for the next iteration; The formula is:

[0028] In the formula, For the The number of individuals in the ICPO population parameter of the iteration; For the The number of individuals in the ICPO population parameter of the iteration; is the minimum number of individuals in the ICPO population parameter; Evaluate arguments for functions; Evaluate loop parameters for a function; Evaluate loop parameters for the maximum function; t is the indicator of the number of iterations; S1-2-4-4: Calculate the initial fitness value of the initial ICPO individuals in the initial ICPO population according to the fitness function; S1-2-4-5: According to the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population; The formula for the first defense strategy is:

[0029] In the formula, For the updated ICPO individuals within the first defense range; The initial ICPO individual within the first defense range; is a random number based on normal distribution; is a random value in the interval [0,1]; It is the optimal solution within the first defense range; is the vector generated between the true optimal solution within the first defense range and the optimal solution randomly selected from the ICPO population; It is the ICPO individual indicator; is the iteration indicator; The formula for the second defense strategy is:

[0030] In the formula, For the updated ICPO individuals within the second defense range; The initial ICPO individual within the second defense range; is the search upper limit vector of the second defense range; is a random value in the interval [0,1]; Respectively Initial ICPO individuals; All are [1, ]; is the vector generated between the true optimal solution within the second defense range and the optimal solution randomly selected from the ICPO population; The formula for the third defense strategy is:

[0031] In the formula, For the updated ICPO individuals within the third defense range; The initial ICPO individual within the third defense range; is the search upper limit vector of the third defense range; Respectively Initial ICPO individuals; is [1, ]; The odor diffusion factor defined for the fitness function; It is a defense factor; Control parameters for search direction; The formula for the fourth defense strategy is:

[0032] In the formula, For the updated ICPO individuals within the fourth defense range; The initial ICPO individual within the fourth defense range; It is the optimal solution within the fourth defense range; All are random values ​​in the interval [0,1]; It is a defense factor; Control parameters for search direction; is the average force affecting the search direction; is the convergence speed factor; S1-2-4-6: Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population; The formula is:

[0033] In the formula, It is a dynamically reversed ICPO individual; is the decreasing inertia coefficient; are the maximum and minimum values ​​of the vector space respectively; For the updated ICPO individuals; S1-2-4-7: According to the fitness function, calculate the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population, take the ICPO individual with the minimum fitness value as the optimal individual, and retain the optimal individual; S1-2-4-8: If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output to obtain the optimal initial historical model parameters of the initial behavior pattern baseline model, including the intercept term and the coefficients of each historical key feature component; S1-2-4-9: Optimize the initial behavior pattern baseline model according to the optimal initial historical model parameters to obtain the optimized behavior pattern baseline model under each historical business model; S1-2-5: Based on a number of pre-processed historical behavior data, the optimized behavior pattern baseline model is optimized and trained to obtain a final behavior pattern baseline model, and a number of historical behavior data features are generated; S1-3: Based on several historical behavior data features, use deep learning algorithms to build a user portrait generation model and generate several historical user portraits; The deep learning algorithm is based on the Long Short-Term Memory (LSTM)-Elman algorithm; LSTM can capture long-distance dependencies and keep the memory of previous information in the sequence, which can help the model better understand the temporal relationship in the sequence and extract deep features from historical behavior data features. The Elman network can be pre-trained through data to achieve accurate prediction of new input behavior data features, and remember the previous state through feedback connection, which enhances the model's understanding of time series dynamics. Based on several historical behavior data features, a deep learning algorithm is used to build a user portrait generation model and generate several historical user portraits, including the following steps: S1-3-1: Divide several historical behavior data features to obtain pre-training set, training set and test set, and use LSTM-Elman algorithm to build the initial user portrait generation model; Using the pre-training set, pre-train the classifier built based on the Elman algorithm in the initial user portrait generation model to obtain a pre-trained user portrait generation model; S1-3-2: Use the training set to optimize the pre-trained user portrait generation model to obtain an optimized user portrait generation model; S1-3-3: Use the test set to test the optimized user portrait generation model. If the model test result meets the requirements, output the final user portrait generation model. Otherwise, continue the optimization training. S1-4: Based on several historical user portraits, use the enhanced reinforcement learning algorithm to build a dynamic permission adjustment model and generate several historical dynamic permission adjustment experiences; The permission dynamic adjustment model is built based on the Meta-Learning (ML)-Deep Q Network (DQN) algorithm; Using meta-learning to initialize the parameters of the DQN model enables it to learn quickly based on a small amount of data, improving the training efficiency of the permission dynamic adjustment model; Based on several historical user portraits, an enhanced reinforcement learning algorithm is used to build a dynamic permission adjustment model and generate several historical dynamic permission adjustment experiences, including the following steps: S1-4-1: Use the permission dynamic adjustment problem as the simulation environment of the DQN algorithm, divide the permission dynamic adjustment problem into several subtasks, and define meta-feature parameters in the historical user portraits; the subtasks correspond to each historical business model, define the goal of meta-learning, that is, quickly adapt to the permission management scenario of the new business model, and determine the meta-learning tasks, such as generating parameters for the DQN model; S1-4-2: According to the meta-feature parameters, extract several historical meta-features of the historical user portraits, and use the ML algorithm to train under several subtasks based on the several historical meta-features to obtain the initial meta-learning module and several historical meta-learning experiences; S1-4-3: Use the initial meta-learning module to initialize the deep Q network of the DQN algorithm, and update the initial meta-learning module to obtain an updated meta-learning module; S1-4-4: Set the environment state of the DQN algorithm (corresponding to the user profile including the user's behavior pattern, historical activities, risk score, etc.), action space (such as increasing, decreasing or maintaining the current user's permissions) and reward function, and build the intelligent agent and experience replay pool to obtain the initial permission dynamic adjustment model; S1-4-5: Based on several historical user portraits, the initial permission dynamic adjustment model is trained and optimized to obtain the final permission dynamic adjustment model and several historical permission dynamic adjustment experiences; S1-4-6: Store several historical meta-learning experiences and several historical permission dynamic adjustment experiences into the experience replay pool of the final permission dynamic adjustment model; S2: using a data collection device to collect the user's real-time business model and real-time behavior data, and uploading the real-time business model and real-time behavior data to a cloud data center; S3: In the cloud data center, extract the real-time behavior data features of the real-time behavior data, and perform abnormal behavior analysis on the real-time behavior data features using the behavior pattern baseline model according to the real-time business model, including the following steps: S3-1: In the cloud data center, use the CNN network to extract the real-time behavior data features of the real-time behavior data; S3-2: Extract several real-time key feature components of real-time behavior data features according to the real-time business model, and match the corresponding target behavior model baseline model; S3-3: Use the target behavior pattern baseline model to perform abnormal behavior analysis on several real-time key feature components to obtain real-time abnormal behavior analysis results, including the following steps: S3-3-1: Input several real-time key feature components into the target behavior pattern baseline model, compare the real-time key feature components with the feature reference values ​​of the target behavior pattern baseline model, and obtain a real-time anomaly score; S3-3-2: If the real-time anomaly score exceeds the preset anomaly score threshold, then output the real-time abnormal behavior analysis result with abnormal behavior, otherwise, output the real-time abnormal behavior analysis result without abnormal behavior; S3-4: If the real-time abnormal behavior analysis result shows that abnormal behavior exists, disconnect the data acquisition device and send a real-time abnormal behavior alarm to the corresponding data acquisition device; S4: Based on the real-time behavior data characteristics, a user portrait generation model is used to generate a real-time user portrait. Based on the real-time user portrait, a permission dynamic adjustment model is used to generate real-time human resource permission data, including the following steps: S4-1: Input the real-time behavior data features into the user portrait generation model, and use the user portrait generation model to generate a real-time user portrait; S4-2: According to the real-time user portrait, the meta-learning module is used to adjust the permission dynamic adjustment model to obtain the adjusted permission dynamic adjustment model, including the following steps: S4-2-1: extracting several real-time meta-features of the real-time user portrait according to the meta-feature parameters, and matching the corresponding real-time meta-learning experience in several historical meta-learning experiences in the experience replay pool according to the real-time meta-features; S4-2-2: Based on the real-time meta-learning experience, use the meta-learning module to generate the real-time network parameters of the deep Q network of the permission dynamic adjustment model; S4-2-3: According to the real-time network parameters, the deep Q network of the authority dynamic adjustment model is adjusted to obtain an adjusted authority dynamic adjustment model; S4-3: Based on the real-time user portrait, the adjusted permission dynamic adjustment model is used to generate a permission dynamic adjustment strategy to obtain real-time human resource permission data, including the following steps: S4-3-1: According to the real-time user portrait, the corresponding several target historical permission dynamic adjustment experiences are matched in the experience playback pool of the adjusted permission dynamic adjustment model, and the action space of the adjusted permission dynamic adjustment model is updated according to the several target historical permission dynamic adjustment experiences to obtain the updated action space ,in, For the updated Action value, is the action indicator, is the total number of action space dimensions; S4-3-2: Analyze the real-time user portrait, update the adjusted permissions, dynamically adjust the state space of the model, and obtain the updated state space ,in, For the updated Status value, is the state indicator, is the total number of state space dimensions; S4-3-3: inputting the updated state space into the adjusted deep Q network of the adjusted authority dynamic adjustment model, and generating a Q value for each possible action in the updated action space using the adjusted deep Q network according to the updated state space; S4-3-4: Use the agent to obtain the reward value of each possible action in the updated action space according to the reward function, and update the Q value of the possible action according to the reward value to obtain the updated Q value of each possible action; The formula is:

[0034] In the formula, The updated status value and the updated action value The corresponding updated Q value; Status value and action value The corresponding predicted Q value; is the learning rate; is the highest predicted Q value; It is a comprehensive indicator; To update the parameters; S4-3-5: Repeat the above Q-value updating steps until the number of iterations reaches the iteration number threshold, and use the greedy strategy to select the possible action with the highest updated Q-value as the execution action; S4-3-6: Generate real-time dynamic adjustment of permissions based on the execution action, and perform dynamic adjustment of permissions on the historical human resource permission data of the corresponding user based on the real-time dynamic adjustment of permissions to obtain real-time human resource permission data.

[0035] Embodiment 2: like Figure 2As shown, this embodiment provides a human resources authority management system for implementing a human resources authority management method. The system includes a cloud data center and several data collection devices, the several data collection devices are respectively connected to the cloud data center for communication, and the cloud data center includes a model building unit, an abnormal behavior analysis unit, and a portrait generation and authority adjustment unit connected in sequence; A data collection device is used to collect the user's real-time business model and real-time behavior data, and upload the real-time business model and real-time behavior data to the cloud data center; A model building unit, used to build a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model in a cloud data center; An abnormal behavior analysis unit, used to extract real-time behavior data features of real-time behavior data, and perform abnormal behavior analysis on the real-time behavior data features using a behavior pattern baseline model according to a real-time business pattern; The portrait generation and authority adjustment unit is used to generate a real-time user portrait based on the real-time behavior data characteristics using a user portrait generation model, and to generate real-time human resource authority data based on the real-time user portrait using a dynamic authority adjustment model.

[0036] The present invention provides a human resource authority management method and system, and proposes a dynamic authority adjustment mechanism, which can automatically adjust the authority according to the real-time behavior data characteristics of the user, solves the problem of lack of flexibility of static authority management in the prior art, and improves practicality; by collecting the user's business model and behavior data in real time, a real-time user portrait is generated in the cloud data center, which helps to more accurately reflect the user's current status and needs, thereby realizing personalized authority management and improving the efficiency and accuracy of dynamic authority adjustment; a behavior pattern baseline model is constructed for abnormal behavior analysis of real-time behavior data characteristics, thereby improving the intelligence level, security and prevention of potential risks of authority management; a dynamic authority adjustment model is constructed, combining the advantages of meta-learning and reinforcement learning, so that the model can quickly adapt to new user portraits and automatically adjust human resource authorities, further improving the intelligence level while avoiding manual adjustment of human resource authorities, reducing labor cost investment, and improving user experience and management efficiency.

[0037] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A human resources authority management method, characterized by: The steps include: In the cloud data center, we build a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model. Using a data collection device to collect real-time business model and real-time behavior data of users, and uploading the real-time business model and real-time behavior data to a cloud data center; In the cloud data center, the real-time behavior data features are extracted, and according to the real-time business model, the behavior pattern baseline model is used to perform abnormal behavior analysis on the real-time behavior data features; Based on the characteristics of real-time behavioral data, use the user portrait generation model to generate real-time user portraits. Based on the real-time user portraits, use the authority dynamic adjustment model to generate real-time human resources authority data.

2. A human resources authority management method according to claim 1, characterized in that: In the cloud data center, building a behavior pattern baseline model, a user portrait generation model, and a permission dynamic adjustment model includes the following steps: In the cloud data center, a number of historical behavior data of users under different historical business modes are collected, and the number of historical behavior data are preprocessed to obtain a number of preprocessed historical behavior data; Based on several pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a group intelligence optimized logistic regression algorithm, and several historical behavior data features are generated; Based on several historical behavior data features, a deep learning algorithm is used to build a user portrait generation model and generate several historical user portraits; Based on several historical user portraits, an enhanced reinforcement learning algorithm is used to build a dynamic permission adjustment model, and generate several historical dynamic permission adjustment experiences.

3. A human resources authority management method according to claim 2, characterized in that: The behavior pattern baseline model is constructed based on the RF-ICPO-LR algorithm; The deep learning algorithm is built based on the LSTM-Elman algorithm; The dynamic permission adjustment model is built based on the ML-DQN algorithm.

4. A human resources authority management method according to claim 3, characterized in that: Based on several pre-processed historical behavior data under different historical business models, a behavior pattern baseline model is constructed using a swarm intelligence optimized logistic regression algorithm, and several historical behavior data features are generated, including the following steps: According to the pre-processed historical behavior data under each historical business model, the corresponding historical behavior data features are extracted using the CNN network; Use the RF algorithm to obtain the importance scores of several feature components in each historical behavior data feature, and select several historical key feature components based on the importance scores; Based on several historical key feature components, the LR algorithm is used to construct the initial behavior pattern baseline model under each historical business model; Use the ICPO algorithm to optimize the initial historical model parameters of the initial behavior pattern baseline model to obtain the optimized behavior pattern baseline model under each historical business model; According to a number of pre-processed historical behavior data, the optimized behavior pattern baseline model is optimized and trained to obtain a final behavior pattern baseline model, and a number of historical behavior data features are generated.

5. A human resource authority management method according to claim 3, characterized in that: Based on several historical behavior data features, a deep learning algorithm is used to build a user portrait generation model and generate several historical user portraits, including the following steps: Divide several historical behavior data features to obtain pre-training sets, training sets, and test sets, and use the LSTM-Elman algorithm to build an initial user portrait generation model; Using the pre-training set, pre-train the classifier built based on the Elman algorithm in the initial user portrait generation model to obtain a pre-trained user portrait generation model; Use the training set to optimize the pre-trained user portrait generation model to obtain an optimized user portrait generation model; Use the test set to test the optimized user portrait generation model. If the model test results meet the requirements, the final user portrait generation model is output; otherwise, continue the optimization training.

6. A human resource authority management method according to claim 3, characterized in that: Based on several historical user portraits, an enhanced reinforcement learning algorithm is used to build a dynamic permission adjustment model and generate several historical dynamic permission adjustment experiences, including the following steps: The permission dynamic adjustment problem is used as the simulation environment of the DQN algorithm, the permission dynamic adjustment problem is divided into several subtasks, and meta-feature parameters are defined in the historical user portrait; According to the meta-feature parameters, several historical meta-features of historical user portraits are extracted, and according to several historical meta-features, ML algorithms are used for training under several subtasks to obtain an initial meta-learning module and several historical meta-learning experiences; Initialize the deep Q network of the DQN algorithm using the initial meta-learning module, and update the initial meta-learning module to obtain an updated meta-learning module; Set the environment state, action space, and reward function of the DQN algorithm, build the agent and experience replay pool, and obtain the initial permission dynamic adjustment model; Based on several historical user portraits, the initial permission dynamic adjustment model is trained and optimized to obtain the final permission dynamic adjustment model and several historical permission dynamic adjustment experiences; Store several historical meta-learning experiences and several historical permission dynamic adjustment experiences into the experience replay pool of the final permission dynamic adjustment model.

7. A human resource authority management method according to claim 4, characterized in that: In the cloud data center, the real-time behavior data features are extracted, and according to the real-time business model, the behavior pattern baseline model is used to perform abnormal behavior analysis on the real-time behavior data features, including the following steps: In the cloud data center, a CNN network is used to extract the real-time behavior data features of the real-time behavior data; According to the real-time business model, several real-time key feature components of the real-time behavior data characteristics are extracted, and the corresponding target behavior pattern baseline model is matched; Use the target behavior pattern baseline model to perform abnormal behavior analysis on several real-time key feature components to obtain real-time abnormal behavior analysis results; If the real-time abnormal behavior analysis result shows that abnormal behavior exists, the connection of the data acquisition device is disconnected, and a real-time abnormal behavior alarm is sent to the corresponding data acquisition device.

8. A human resource authority management method according to claim 6, characterized in that: According to the real-time behavior data characteristics, a user portrait generation model is used to generate a real-time user portrait. According to the real-time user portrait, a permission dynamic adjustment model is used to generate real-time human resource permission data, including the following steps: Input the real-time behavior data features into the user portrait generation model, and use the user portrait generation model to generate a real-time user portrait; According to the real-time user portrait, the meta-learning module is used to adjust the permission dynamic adjustment model to obtain the adjusted permission dynamic adjustment model; According to the real-time user portrait, the adjusted permission dynamic adjustment model is used to generate a permission dynamic adjustment strategy to obtain real-time human resources permission data.

9. A human resources authority management method according to claim 8, characterized in that: According to the real-time user portrait, the meta-learning module is used to adjust the permission dynamic adjustment model to obtain the adjusted permission dynamic adjustment model, including the following steps: According to the meta-feature parameters, several real-time meta-features of the real-time user portrait are extracted, and according to the real-time meta-features, corresponding real-time meta-learning experiences are matched in several historical meta-learning experiences in the experience replay pool; Based on the real-time meta-learning experience, a meta-learning module is used to generate real-time network parameters of the deep Q network of the permission dynamic adjustment model; According to the real-time network parameters, the deep Q network of the authority dynamic adjustment model is adjusted to obtain the adjusted authority dynamic adjustment model.

10. A human resources authority management system, used to implement the human resources authority management method according to any one of claims 1 to 9, characterized in that: The system includes a cloud data center and several data acquisition devices, wherein the several data acquisition devices are respectively connected to the cloud data center in communication, and the cloud data center includes a model building unit, an abnormal behavior analysis unit, and a portrait generation and authority adjustment unit which are connected in sequence.

Images