An agent data dynamic matching method based on user authority

Abstract

The application discloses a user permission-based intelligent agent data dynamic matching method, and belongs to the technical field of artificial intelligence. User identity authentication is performed through face recognition technology, replacing traditional static passwords, greatly improving the security and anti-counterfeiting capability of identity authentication. Meanwhile, the authentication result is bound with a dynamic data access role, realizing a flexible access control mode of one-time authentication and dynamic authorization, which can instantly match the most suitable permission according to the real-time identity of the user and the access target, improving the flexibility and response speed of the system.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence technology, specifically relating to a method for dynamic matching of intelligent agent data based on user permissions. Background Technology

[0002] With the rapid development of artificial intelligence and intelligent agent technologies, intelligent agent systems have been widely applied in finance, healthcare, and smart devices. However, ensuring the security and controllability of intelligent agent data during access has become a pressing issue. Traditional authentication methods based on usernames and passwords are vulnerable to attacks and cannot dynamically adapt to changes in user permissions. Furthermore, existing technologies lack sufficient dynamic monitoring and security detection mechanisms for access behavior, making it difficult to effectively prevent unauthorized access or malicious operations. Summary of the Invention

[0003] This application provides a method for dynamic matching of intelligent agent data based on user permissions, in order to solve the technical problem that the existing technology based on password authentication results in poor security detection.

[0004] This application provides a method for dynamic matching of intelligent agent data based on user permissions, including: Obtain user-generated agent data requests; wherein, the agent data requests include at least a user's facial image and the target agent data that the user wants to access; A pre-set face recognition model is used to identify the user's face image in the data request of the intelligent agent to determine the user's corresponding data access role; Determine the matching between the data access role and the access role requirements corresponding to the target intelligent agent data to obtain the data role matching result; If the data role matching result is successful, the user is allowed to access the target intelligent agent data within the user permissions corresponding to the data access role.

[0005] In one possible implementation, it also includes: During the process of the user accessing the target intelligent agent data, the network traffic characteristics corresponding to the user are collected, and the network traffic characteristics are identified by a pre-set deep learning model to determine the network security detection result. Based on the network security detection results, the user's access behavior is controlled.

[0006] In one possible implementation, a pre-set face recognition model is used to identify the user's face image in the intelligent agent's data request, and to determine the user's corresponding data access role, including: A pre-set face recognition model is used to identify the user's face image in the data request of the intelligent agent to obtain the user's corresponding identity authentication result; Based on the identity authentication result, query the pre-set association table of identity and data access role to obtain the data access role corresponding to the user.

[0007] In one possible implementation, the method for pre-setting the face recognition model includes: Construct a face recognition model and initialize the hyperparameters of the face recognition model to obtain multiple different parameter vectors; The parameter vector is trained sequentially using information weighted learning strategy, rotation information learning strategy, chain combination learning strategy and global information learning strategy until the maximum number of training times is reached to obtain the target optimal vector; The face recognition model is preset based on the target optimal vector.

[0008] In one possible implementation, a face recognition model is constructed, and the hyperparameters of the face recognition model are initialized to obtain multiple different parameter vectors, including: A face recognition model is constructed using a convolutional neural network. The parameters of the face recognition model are randomly initialized between the upper limit and the lower limit. The initialized parameters are then combined into a vector to obtain a parameter vector. Multiple different parameter vectors are obtained repeatedly.

[0009] In one possible implementation, the information-weighted learning strategy includes:

[0010]

[0011]

[0012]

[0013]

[0014]

[0015]

[0016] in, Indicates the first t The training process of the 1st time j A parameter vector j =1,2,…,J, where J represents the total number of parameter vectors. Indicates the first j The parameter vector after weighted learning of each piece of information. Indicates the first weighted weight. Indicates the second weighting weight. Indicates the third weighting weight. This represents the first weighted learning information. This represents the second weighted learning information. This represents the third weighted learning information. Describes the first random vector. Describes the second random vector. Represents the third random vector. Represents the first random number between (0,1). This represents the second random number between (0,1). This represents a third random number between (0,1). This represents the fitness of the first random vector. This represents the fitness of the second random vector. This represents the fitness of the third random vector. Represents pi (π). Let denote the natural constant e as the base, and let cos denote the cosine function. The influence factor is expressed as a constant term.

[0017] In one possible implementation, the rotation information learning strategy includes:

[0018]

[0019] in, Indicates the first t During the training process, the first i The parameter vector after weighted learning of each piece of information. i =1,2,…,J, Indicates the first i The parameter vector after learning rotational information. This represents the current optimal vector. Indicates inertia weight, Let b represent the natural constant and b represent the spiral path factor. This represents the spiral search step size control factor between (-1, 1). This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. This represents the inverse cosine function.

[0020] In one possible implementation, the chain-combination learning strategy includes:

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] in, Indicates the first t During the training process, the first n The parameter vector after learning rotational information. n =1,2,…,J, Indicates the first n The parameter vector after chain combination learning. This represents the disturbance factor. Represents the first combined learning factor. This represents the second combined learning factor. Indicates the first n The parameter vector after learning +1 rotation information, and n When it is J, then The parameter vector is learned from random rotation information. Represents the average parameter vector. Indicates the learning step size in chaining. Indicates the direction of chain learning. Represents the sine function. Represents the corresponding parameter vectors The maximum value, Represents the corresponding parameter vectors The maximum value, Indicates a constant factor. This represents the fourth random number between (0,1). The learning cycle parameter represents the learning period of chain combinatorial learning. This represents the fifth random number between (0,1). Indicates the maximum number of training iterations. This represents the maximum value of the disturbance factor. This represents the minimum value of the disturbance factor.

[0027] In one possible implementation, the global information learning strategy includes:

[0028]

[0029] in, Indicates the first t During the training process, the first m The parameter vector after chain combination learning. m =1,2,…,J, Indicates the first m The parameter vector after learning global information. Let represent the parameter vector after the first random chain combination learning. This represents the parameter vector after the second random chain combination learning. This means that the mean is 0, and the mean is 0. The Gaussian coefficient of variation is generated from the Gaussian variation of the variance. This represents the Gaussian variation control factor. This represents the maximum value of the Gaussian variation control factor. This represents the minimum value of the Gaussian variation control factor.

[0030] In one possible implementation, determining the matching between the data access role and the access role requirements corresponding to the target agent data to obtain the data role matching result includes: Determine whether the data access role is among the access role requirements corresponding to the target agent data. If so, the data role matching result is determined to be a successful match; otherwise, the data role matching result is determined to be a failed match.

[0031] This application provides a user-permission-based intelligent agent data dynamic matching method. It uses facial recognition technology for user authentication, replacing traditional static passwords and significantly improving the security and anti-forgery capabilities of authentication. Simultaneously, it binds the authentication result to dynamic data access roles, achieving a flexible access control mode of one-time authentication and dynamic authorization. This allows for the instant matching of the most appropriate permissions based on the user's real-time identity and access goals, enhancing the system's flexibility and response speed. Attached Figure Description

[0032] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0033] Figure 1 A flowchart illustrating a method for dynamic matching of intelligent agent data based on user permissions, provided in an embodiment of this application.

[0034] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0035] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0036] The embodiments of this application are described in detail below with reference to the accompanying drawings.

[0037] like Figure 1 As shown, this application implements a method for dynamic matching of intelligent agent data based on user permissions, including: S101. Obtain the intelligent agent data request generated by the user. The intelligent agent data request includes at least the user's facial image and the target intelligent agent data that the user wants to access.

[0038] S102. Use a pre-set face recognition model to identify the user's face image in the data request of the intelligent agent and determine the data access role corresponding to the user.

[0039] This application uses facial recognition technology for user authentication, replacing traditional static passwords and significantly improving the security and anti-forgery capabilities of authentication. Simultaneously, it binds the authentication result to dynamic data access roles, achieving a flexible access control mode of "one-time authentication, dynamic authorization." This allows for the instant matching of the most appropriate permissions based on the user's real-time identity and access objectives, enhancing the system's flexibility and responsiveness.

[0040] It is worth noting that different users can be assigned different data access roles by the administrator, and different data access roles have different access permissions to the data, thereby enabling the control of user permissions.

[0041] S103. Determine the matching between the data access role and the access role requirements corresponding to the target intelligent agent data to obtain the data role matching result.

[0042] For the same type of target intelligent agent data, different data access roles can be assigned different user permissions. Therefore, by using the same target intelligent agent data, different data access roles can perform different data access operations, thereby achieving dynamic management. Here, user permissions refer to the data access operation permissions corresponding to a data access role. For example, the user permission of a certain data access role can be access to the target intelligent agent data.

[0043] S104. If the data role matching result is successful, then the user is allowed to access the target intelligent agent data within the user permissions corresponding to the data access role.

[0044] This application establishes a mapping relationship of "identity-role-permission," enabling fine-grained permission management of agent data. Different data access roles can be assigned different operation permissions (such as read-only, read-write, modification, download, etc.) and data access scopes (such as specific models, data within specific time periods, etc.), ensuring that users can only operate within their authorized scope, effectively preventing permission abuse and data leakage, and meeting the needs of agent data for high security and fine-grained control.

[0045] In one possible implementation, it also includes: During the process of the user accessing the target intelligent agent data, the network traffic characteristics corresponding to the user are collected, and the network traffic characteristics are identified by a pre-set deep learning model to determine the network security detection result. Based on the network security detection results, the user's access behavior is controlled.

[0046] This application not only performs strict identity and permission verification at the access point, but also introduces a real-time security detection mechanism based on network traffic characteristics during user access. By analyzing user behavior patterns through deep learning models, it can promptly detect potential threats such as abnormal access, data theft, and malicious operations, and dynamically interrupt or restrict access based on the detection results. This achieves full-cycle security protection from "entry control" to "process monitoring," effectively making up for the shortcomings of traditional solutions that lack continuous monitoring.

[0047] Optionally, the network security detection result can be either "network secure" or "network insecure." If the network is secure, monitoring continues. If the network is insecure, the user's access behavior can be controlled based on the network security detection result. Control measures may include: prohibiting the user from accessing the agent's data for a certain period; as the number of times the user is detected as having network insecurity increases, the prohibition period gradually lengthens; if the number of network insecurity detections exceeds a preset limit, the user's access can be permanently prohibited. After being prohibited, the user can be unblocked by the administrator. It is worth noting that using a pre-set deep learning model for network security identification is a relatively common technique and will not be elaborated upon here.

[0048] In one possible implementation, a pre-set face recognition model is used to identify the user's face image in the intelligent agent's data request, and to determine the user's corresponding data access role, including: A pre-set face recognition model is used to identify the user's face image in the data request of the intelligent agent to obtain the user's corresponding identity authentication result; Based on the identity authentication result, query the pre-set association table of identity and data access role to obtain the data access role corresponding to the user.

[0049] Facial images of users can be pre-collected, and the facial recognition model can be trained using these images and the user's identity tags, enabling the model to identify the user. Furthermore, by setting a cosine similarity table to establish the association between identity and data access role, the user's corresponding data access role can be determined, thereby further determining whether the user can access the target agent's data.

[0050] In one possible implementation, the method for pre-setting the face recognition model includes: Construct a face recognition model and initialize the hyperparameters of the face recognition model to obtain multiple different parameter vectors; The parameter vector is trained sequentially using information weighted learning strategy, rotation information learning strategy, chain combination learning strategy and global information learning strategy until the maximum number of training times is reached to obtain the target optimal vector; The face recognition model is preset based on the target optimal vector.

[0051] The target optimal vector refers to the parameter vector obtained after learning the global information with the highest fitness during the last training process. The parameter vector in the target optimal vector can be used as the final parameters of the face recognition model.

[0052] This application's embodiments significantly improve the overall performance of a face recognition model through an innovative multi-strategy collaborative training method. First, an information-weighted learning strategy enhances key information extraction, a rotational information learning strategy improves robustness to search poses, a chain-based combined learning strategy optimizes feature fusion efficiency, and a global information learning strategy enhances the model's overall generalization ability. These four strategies, trained sequentially, complement each other, enabling the model to maintain high-precision recognition even in complex scenarios. The combination of hyperparameter initialization and multi-parameter vector training effectively avoids local optima, and the accurate acquisition of the target optimal vector significantly improves the quality of the model's presets. Ultimately, this significantly enhances the model's recognition accuracy, robustness, and generalization ability, providing solid technical support for the reliable application of face recognition technology in demanding scenarios such as security and finance.

[0053] Optionally, after each training iteration of the parameter vector, out-of-bounds handling can be performed on the parameters in the parameter vector to ensure parameter validity.

[0054] In one possible implementation, a face recognition model is constructed, and the hyperparameters of the face recognition model are initialized to obtain multiple different parameter vectors, including: A face recognition model is constructed using a convolutional neural network. The parameters of the face recognition model are randomly initialized between the upper limit and the lower limit. The initialized parameters are then combined into a vector to obtain a parameter vector. Multiple different parameter vectors are obtained repeatedly.

[0055] For example, the weights of a convolutional neural network generally have a lower bound of 0 and an upper bound of 1. Therefore, the weights can be randomly initialized between (0,1), and the initialized weights can be combined into a vector to obtain the parameter vector.

[0056] In one possible implementation, the information-weighted learning strategy includes:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] in, Indicates the first t The training process of the 1st time j A parameter vector j =1,2,…,J, where J represents the total number of parameter vectors. Indicates the first j The parameter vector after weighted learning of each piece of information. Indicates the first weighted weight. Indicates the second weighting weight. Indicates the third weighting weight. This represents the first weighted learning information. This represents the second weighted learning information. This represents the third weighted learning information. Describes the first random vector. Describes the second random vector. Represents the third random vector. Represents the first random number between (0,1). This represents the second random number between (0,1). This represents a third random number between (0,1). This represents the fitness of the first random vector. This represents the fitness of the second random vector. This represents the fitness of the third random vector. Represents pi (π). Let denote the natural constant e as the base, and let cos denote the cosine function. The influence factor, represented as a constant term, can be set to 5.

[0064] Optionally, the fitness of the first random vector can be less than the fitness of the second random vector, and the fitness of the second random vector can be less than the fitness of the third random vector. The fitness can be obtained by first obtaining the loss function value corresponding to the parameter vector, and then taking the reciprocal of the loss function value to obtain the fitness of the parameter vector. To avoid the denominator being zero when taking the reciprocal, the loss function value can be incremented by 1 first.

[0065] The information weighted learning strategy provided in this application can effectively learn random information from parameter vectors. This learning method can increase the solution space traversal capability in the early stage of the algorithm and enhance the local search capability in the later stage of the algorithm, making the algorithm more robust.

[0066] In one possible implementation, the rotation information learning strategy includes:

[0067]

[0068] in, Indicates the first t During the training process, the first i The parameter vector after weighted learning of each piece of information. i =1,2,…,J, Indicates the first i The parameter vector after learning rotational information. This represents the current optimal vector. Indicates inertia weight, represents the natural constant, and b represents the spiral path factor, which can be set to 1 or 2. This represents the spiral search step size control factor between (-1, 1). This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. This represents the inverse cosine function.

[0069] The rotation information learning strategy provided in this application further enhances the local search capability of the algorithm, enabling the parameter vector to perform a more detailed local search after learning the information corresponding to other parameter vectors, thereby improving the convergence capability of the algorithm.

[0070] In one possible implementation, the chain-combination learning strategy includes:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] in, Indicates the first t During the training process, the first n The parameter vector after learning rotational information. n =1,2,…,J, Indicates the first n The parameter vector after chain combination learning. This represents the disturbance factor. Represents the first combined learning factor. This represents the second combined learning factor. Indicates the first n The parameter vector after learning +1 rotation information, and n When it is J, then The parameter vector is learned from random rotation information. Represents the average parameter vector. Indicates the learning step size in chaining. Indicates the direction of chain learning. Represents the sine function. Represents the corresponding parameter vectors The maximum value, Represents the corresponding parameter vectors The maximum value, This represents a constant factor, and is set to 5. This represents the fourth random number between (0,1). This represents the learning cycle parameter for chain-based combinatorial learning, and is set to 1.5. This represents the fifth random number between (0,1). Indicates the maximum number of training iterations. This represents the maximum value of the disturbance factor, and is set to 0.95; This represents the minimum value of the disturbance factor, and is set to 0.4.

[0077] The chain-based combinatorial learning strategy provided in this application can effectively enable parameter vectors to form a chain search in the solution space. This chain composed of parameter vectors is irregular, and each parameter vector is searched in a spiral manner, thereby forming a chain-based combinatorial learning, which greatly improves the efficiency and ability of solution space exploration, thereby improving the global search capability of the algorithm. Furthermore, as the algorithm converges, the convergence accuracy will become higher and higher, ensuring the convergence of the algorithm.

[0078] In one possible implementation, the global information learning strategy includes:

[0079]

[0080] in, Indicates the first t During the training process, the first m The parameter vector after chain combination learning. m =1,2,…,J, Indicates the first m The parameter vector after learning global information. Let represent the parameter vector after the first random chain combination learning. This represents the parameter vector after the second random chain combination learning. This means that the mean is 0, and the mean is 0. The Gaussian coefficient of variation is generated from the Gaussian variation of the variance. This represents the Gaussian variation control factor. This represents the maximum value of the Gaussian variation control factor, and is set to 0.8; This represents the minimum value of the Gaussian variation control factor, and is set to 0.01.

[0081] The global information learning strategy provided in this application can improve the global search capability, enabling the algorithm to escape local optima. As the algorithm progresses, the mutation range gradually shrinks, allowing the algorithm to converge normally in the later stages and improving the accuracy of the algorithm.

[0082] Traditional gradient descent is prone to getting stuck in local optima and is sensitive to initialization. However, the embodiments of this application effectively avoid these defects by using multi-parameter vector initialization and collaborative training with four learning strategies, thereby improving the accuracy of face recognition.

[0083] In one possible implementation, determining the matching between the data access role and the access role requirements corresponding to the target agent data to obtain the data role matching result includes: Determine whether the data access role is among the access role requirements corresponding to the target agent data. If so, the data role matching result is determined to be a successful match; otherwise, the data role matching result is determined to be a failed match.

[0084] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0086] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0087] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0088] Those skilled in the art will understand that all or part of the steps in the above facts and methods can be implemented by a program instructing related hardware. The program or the program described therein can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: at this time, the corresponding method steps are introduced. The storage medium can be ROM / RAM, magnetic disk, optical disk, etc.

[0089] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dynamic matching of intelligent agent data based on user permissions, characterized in that, include: Obtain user-generated agent data requests; wherein, the agent data requests include at least a user's facial image and the target agent data that the user wants to access; A pre-set face recognition model is used to identify the user's face image in the data request of the intelligent agent to determine the user's corresponding data access role; Determine the matching between the data access role and the access role requirements corresponding to the target intelligent agent data to obtain the data role matching result; If the data role matching result is successful, the user is allowed to access the target intelligent agent data within the user permissions corresponding to the data access role; The method for pre-setting the face recognition model includes: Construct a face recognition model and initialize the hyperparameters of the face recognition model to obtain multiple different parameter vectors; The parameter vector is trained sequentially using information weighted learning strategy, rotation information learning strategy, chain combination learning strategy and global information learning strategy until the maximum number of training times is reached to obtain the target optimal vector; The face recognition model is preset based on the target optimal vector; The rotation information learning strategy includes: in, Indicates the first t During the training process, the first i The parameter vector after weighted learning of each piece of information. i =1,2,…,J, Indicates the first i The parameter vector after learning rotational information. This represents the current optimal vector. Indicates inertia weight, Let b represent the natural constant and b represent the spiral path factor. This represents the spiral search step size control factor between (-1, 1). Let π represent the mathematical constant pi, and cos represent the cosine function. This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. Indicates the maximum number of training iterations. Represents the inverse cosine function; The chain-based combination learning strategy includes: in, Indicates the first t During the training process, the first n The parameter vector after learning rotational information. n =1,2,…,J, Indicates the first n The parameter vector after chain combination learning. This represents the disturbance factor. Represents the first combined learning factor. This represents the second combined learning factor. Indicates the first n The parameter vector after learning +1 rotation information, and n When it is J, then The parameter vector is learned from random rotation information. Represents the average parameter vector. Indicates the learning step size in chaining. Indicates the direction of chain learning. Represents the sine function. Represents the corresponding parameter vectors The maximum value, Represents the corresponding parameter vectors The maximum value, Indicates a constant factor. This represents the fourth random number between (0,1). The learning cycle parameter represents the learning period of chain combinatorial learning. This represents the fifth random number between (0,1). This represents the maximum value of the disturbance factor. This represents the minimum value of the disturbance factor. 2.The method of claim 1, wherein, Also includes: During the process of the user accessing the target intelligent agent data, the network traffic characteristics corresponding to the user are collected, and the network traffic characteristics are identified by a pre-set deep learning model to determine the network security detection result. Based on the network security detection results, the user's access behavior is controlled. 3.The method of claim 1, wherein, A pre-built facial recognition model is used to identify the user's facial image in the data request from the intelligent agent, and to determine the user's corresponding data access role, including: A pre-set face recognition model is used to identify the user's face image in the data request of the intelligent agent to obtain the user's corresponding identity authentication result; Based on the identity authentication result, query the pre-set association table of identity and data access role to obtain the data access role corresponding to the user. 4.The method of claim 1, wherein, Construct a face recognition model and initialize its hyperparameters, obtaining multiple different parameter vectors, including: A face recognition model is constructed using a convolutional neural network. The parameters of the face recognition model are randomly initialized between the upper limit and the lower limit. The initialized parameters are then combined into a vector to obtain a parameter vector. Multiple different parameter vectors are obtained repeatedly.

5. The method of claim 1, wherein, The information weighted learning strategy includes: in, Indicates the first t The training process of the 1st time j A parameter vector j =1,2,…,J, where J represents the total number of parameter vectors. Indicates the first j The parameter vector after weighted learning of each piece of information. Indicates the first weighted weight. Indicates the second weighting weight. Indicates the third weighting weight. This represents the first weighted learning information. This represents the second weighted learning information. This represents the third weighted learning information. Describes the first random vector. Describes the second random vector. Represents the third random vector. Represents the first random number between (0,1). This represents the second random number between (0,1). This represents a third random number between (0,1). This represents the fitness of the first random vector. This represents the fitness of the second random vector. This represents the fitness of the third random vector. Represents pi (π). Let denote the natural constant e as the base, and let cos denote the cosine function. The influence factor is expressed as a constant term. 6.The method of claim 5, wherein, The global information learning strategy includes: in, Indicates the first t During the training process, the first m The parameter vector after chain combination learning. m =1,2,…,J, Indicates the first m The parameter vector after learning global information. This represents the parameter vector after the first random chain combination learning. This represents the parameter vector after learning the second random chain combination. This means that the mean is 0, and the mean is 0. The Gaussian coefficient of variation is generated from the Gaussian variation of the variance. This represents the Gaussian variation control factor. This represents the maximum value of the Gaussian variation control factor. This represents the minimum value of the Gaussian variation control factor.

7. The method of claim 1, wherein, Determine the matching between the data access role and the access role requirements corresponding to the target intelligent agent data to obtain the data role matching result, including: Determine whether the data access role is among the access role requirements corresponding to the target agent data. If so, the data role matching result is determined to be a successful match; otherwise, the data role matching result is determined to be a failed match.

Images