Privacy protection and data security processing method and device of expert agent, terminal, medium and product

Abstract

The application provides a privacy protection and data security processing method and device of an expert Agent, a terminal, a medium and a product. The method is applied to an expert Agent and includes the following steps: obtaining a target task and corresponding task attribute information; performing a permission judgment operation based on a pre-constructed hybrid permission model according to the task attribute information to obtain a corresponding permission judgment result; in the case that the permission judgment result is that the permission is possessed, determining target data to be called according to the target task; obtaining and performing security processing on the target data based on a hierarchical privacy protection rule according to a hierarchical label of the target data to obtain processed target data; and completing the target task based on the processed target data. The application effectively protects data privacy and security under the premise of ensuring normal operation of the expert Agent through data classification and a privacy protection and data security mechanism.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, terminal, medium and product for privacy protection and data security processing of an expert agent. Background Technology

[0002] Expert agents are increasingly being used in fields such as financial risk control and medical diagnosis. During operation, expert agents frequently process user privacy data and core business data, while also interacting with external systems and users to perform operations such as knowledge retrieval and data processing.

[0003] Currently, expert agents face the following data privacy and security risks: sensitive data is not processed according to risk levels, which can easily lead to the leakage of highly sensitive data; during the interaction between expert agents and external systems and users, most data is transmitted by calling network interfaces (APIs), and the calling instances generally use public cloud virtual machines, which poses a risk of interception and leakage; the granularity of data access control is coarse, a dynamic permission management mechanism has not been established, and there is a lack of full-process behavior auditing for knowledge calling and data processing operations, making it impossible to trace the source of security incidents. Summary of the Invention

[0004] In view of the shortcomings of the prior art described above, the purpose of this application is to provide a method, device, terminal, medium and product for privacy protection and data security processing of expert agents, so as to solve the problem of data privacy and security risks existing in existing expert agents.

[0005] To achieve the above and other related objectives, a first aspect of this application provides a privacy protection and data security processing method for an expert agent, applied to an expert agent, comprising: acquiring a target task and corresponding task attribute information; performing a permission judgment operation based on the task attribute information and a pre-built hybrid permission model to obtain a corresponding permission judgment result; if the permission judgment result indicates that permission is granted, determining target data to be invoked based on the target task; wherein the target data has a hierarchical label assigned based on a multi-dimensional weighted evaluation rule; acquiring and security processing the target data based on the hierarchical label of the target data and a hierarchical privacy protection rule to obtain processed target data; and completing the target task based on the processed target data.

[0006] In some embodiments of the first aspect of this application, the hybrid permission model includes: a role layer and an attribute layer; wherein, the role layer defines multiple roles and basic permissions for multiple roles; the attribute layer defines multiple attributes; and the attribute layer further includes multiple decision rules defined based on multiple attributes and multiple roles.

[0007] In some embodiments of the first aspect of this application, the method of assigning graded labels to target data based on multi-dimensional weighted evaluation rules includes: setting multiple evaluation indicators and determining the score of the target data under each evaluation indicator; using the analytic hierarchy process (AHP) to determine the weight of each evaluation indicator; calculating the score of the target data based on the score of the target data under each evaluation indicator and the weight of each evaluation indicator; and obtaining graded labels for the target data based on pre-obtained data grading rules and the score of the target data and assigning them to the target data.

[0008] In some embodiments of the first aspect of this application, the types of the graded labels include: extremely sensitive data, highly sensitive data, moderately sensitive data, and low sensitive data.

[0009] In some embodiments of the first aspect of this application, the hierarchical privacy protection rules include: if the hierarchical label of the target data is extremely sensitive data, then the target data is obtained through end-to-end encryption and decrypted in a local trusted execution environment; if the target task is an analysis task, then the decrypted target data is hashed; wherein the end-to-end encryption method uses a rolling key change; if the hierarchical label of the target data is highly sensitive data, then the target data is subjected to synonymous hashing and random hashing; if the hierarchical label of the target data is moderately sensitive data, then the target data is subjected to interval replacement; if the target task is an analysis task, then a perturbation conforming to a normal distribution is added to the replaced target data; the data after adding perturbation satisfies a pre-set statistical characteristic deviation; if the hierarchical label of the target data is low sensitive data, then the target data is subjected to redundant field deletion and format standardization.

[0010] In some embodiments of the first aspect of this application, the method further includes: acquiring full-process operation data; using pre-set security rules in the rule engine and a pre-trained audit model, obtaining abnormal operation identification results based on the full-process operation data, and issuing corresponding alarms based on the abnormal operation identification results; wherein, the audit model is obtained by training a combined model from the training dataset; the combined model consists of a random forest model and an isolated forest model.

[0011] To achieve the above and other related objectives, a second aspect of this application provides a privacy protection and data security processing device for an expert agent, disposed within the expert agent, comprising: an acquisition module for acquiring a target task and corresponding task attribute information; a permission judgment module for performing a permission judgment operation based on the task attribute information and a pre-built hybrid permission model to obtain a corresponding permission judgment result; a call confirmation module for determining target data to be called based on the target task if the permission judgment result indicates that permission is granted; wherein the target data has a hierarchical label assigned based on multi-dimensional weighted evaluation rules; and a privacy protection and data security processing module for acquiring and security processing the target data based on the hierarchical label of the target data and hierarchical privacy protection rules to obtain processed target data; and completing the target task based on the processed target data.

[0012] To achieve the above and other related objectives, a third aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the privacy protection and data security processing method for the expert agent.

[0013] To achieve the above and other related objectives, a fourth aspect of this application provides a computer program product comprising computer program code, which, when executed on a computer, enables the computer to implement the privacy protection and data security processing method of the expert agent.

[0014] To achieve the above and other related objectives, a fifth aspect of this application provides an electronic terminal, including a memory, a processor, and a computer program stored in the memory; the processor executes the computer program to implement the privacy protection and data security processing method of the expert agent.

[0015] As described above, the privacy protection and data security processing method, device, terminal, medium and product of the expert agent in this application have the following beneficial effects: This application effectively protects data privacy and security while ensuring the normal operation of the expert agent function through data classification and privacy protection and data security mechanisms. Attached Figure Description

[0016] Figure 1 The diagram shown is a flowchart illustrating the privacy protection and data security processing method for an expert agent in one embodiment of this application.

[0017] Figure 2 The diagram shown is a schematic block diagram illustrating the privacy protection and data security processing method of an expert agent in one embodiment of this application.

[0018] Figure 3 The diagram shown is a structural schematic of an electronic terminal according to an embodiment of this application. Detailed Implementation

[0019] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0020] To facilitate understanding of the embodiments of this application, firstly, in conjunction with Figure 1 Detailed explanation. Figure 1 This document illustrates a flowchart of a privacy protection and data security processing method for an expert agent according to an embodiment of the present invention. The privacy protection and data security processing method for an expert agent in this embodiment is applied to an expert agent. It should be understood that an expert agent is an intelligent agent possessing specific domain expertise and decision-making capabilities. The privacy protection and data security processing method for an expert agent mainly includes the following steps:

[0021] Step S11: Obtain the target task and its corresponding task attribute information.

[0022] In one embodiment, in the use case of the expert agent, the user inputs a task to the expert agent according to their needs, and the expert agent processes the received task accordingly. The target task is the task input by the user to the expert agent according to their needs. The types of target tasks include, but are not limited to, consultation and answering tasks, data analysis tasks, model training tasks, document processing tasks, etc.

[0023] In one embodiment, the task attribute information includes, but is not limited to, user level, user role type, user department, data level, data domain, access IP address, access time, etc.

[0024] Step S12: Based on the target task and the user identity information, perform permission judgment operation according to the pre-built hybrid permission model to obtain the corresponding permission judgment result.

[0025] In one embodiment, the hybrid permission model is an ABAC-RBAC hybrid permission model (attribute-based access control + role-based access control). The hybrid permission model embeds the expert agent's knowledge retrieval and data processing interfaces, performing permission verification before the expert agent processes tasks. It should be understood that the expert agent's knowledge retrieval and data processing interfaces serve as a bridge connecting the agent with external knowledge sources and data resources; essentially, they are standardized interaction protocols and functional entry points.

[0026] The hybrid access control model comprises a role layer and an attribute layer. The role layer defines multiple roles for expert agents. For example, roles can be "Data Analyst," "Agent Maintenance Personnel," or "Regular User." Each role is associated with basic permissions. For instance, a "Data Analyst" can access the knowledge base but cannot modify core algorithms. The attribute layer defines various attributes, along with multiple decision rules based on roles and attributes. Attributes include subject attributes, access object attributes, and environment attributes. Subject attributes include user level and role type; access object attributes include data level and data domain; environment attributes include access IP address and access time. For example, a decision rule could be: "Only 'Agent Maintenance Personnel' (role attribute) with an IP address on the enterprise intranet (environment attribute) is allowed to access highly sensitive data (access object attribute) during working hours (environment attribute)." Decision rules can be stored using Extensible Markup Language (XML) to support dynamic updates.

[0027] In one embodiment, based on task attribute information and a pre-built hybrid permission model, a permission judgment operation is performed to obtain the corresponding permission judgment result, including:

[0028] The role layer checks whether the user's role has the basic permissions to execute the target task based on the user role type in the task attribute information; if it does, the attribute layer is checked; if it does not, a permission judgment result of no permission is generated.

[0029] The attribute layer selects one or more decision rules related to the target task from a set of predefined decision rules, and then matches the task attribute information with the selected decision rules. If the task attribute information matches all of the selected decision rules, a permission judgment result with permission is generated; otherwise, a permission judgment result without permission is generated.

[0030] It should be noted that this embodiment uses a hybrid permission model that combines roles and attributes to achieve dynamic permission adjustment and avoid permission abuse.

[0031] To better illustrate the process of permission determination using a hybrid permission model, three specific implementation examples are provided.

[0032] Example 1: Permission verification process for the first type of expert agent task.

[0033] The task acquired by the expert agent model is to analyze data A. Data A is highly sensitive data. This task is initiated by user B, whose associated role is "ordinary role," and who belongs to department C. User B initiates the task via the company intranet during working hours.

[0034] First, the role layer performs the verification. User B's role is "regular user." "Regular users" do not have permission to access highly sensitive data, so a permission failure result is generated, and an insufficient permission message is returned to the user. The permission verification process ends.

[0035] Example 2: Permission verification process for the second type of expert agent task.

[0036] The task acquired by the expert agent model is to analyze data A. Data A is highly sensitive data. This task is initiated by user B, whose associated role is "data analyst," and who belongs to department C. User B initiates the task outside of working hours via the company intranet.

[0037] First, the role layer performs the verification. User B's role is "Data Analyst". "Data Analyst" has the right to access highly sensitive data, so the verification proceeds to the attribute layer.

[0038] The relevant decision rule is: only "data analysts" whose IP address is on the corporate intranet are allowed to access highly sensitive data during working hours. If User B initiates the request outside of working hours, the decision rule is not met, resulting in a permission failure judgment result, and an insufficient permission message is returned to the user. The permission verification process ends.

[0039] Example 3: Permission verification process for the third type of expert agent task.

[0040] The task acquired by the expert agent model is to analyze data A. Data A is highly sensitive data. This task is initiated by user B, whose associated role is "data analyst," and who belongs to department C. User B initiates the task outside of working hours via the company intranet.

[0041] First, the role layer performs the verification. User B's role is "Data Analyst". "Data Analyst" has the right to access highly sensitive data, so the verification proceeds to the attribute layer.

[0042] The relevant decision-making rule is: only "data analysts" whose IP addresses are on the corporate intranet are allowed to access highly sensitive data during working hours. User B meets all the conditions defined in the decision-making rule, obtains the permission judgment result, and can proceed with the subsequent data analysis process.

[0043] Step S13: If the permission judgment result indicates that the user has the necessary permissions, determine the target data to be invoked based on the target task.

[0044] In one embodiment, the target data to be invoked is the data required by the expert agent to complete the target task. The data processed by the expert agent is mainly divided into two categories: user privacy data and core business data. It should be noted that both user privacy data and core business data are obtained legally, compliantly, and with authorization.

[0045] In one embodiment, the target data has a hierarchical label assigned based on a multi-dimensional weighted evaluation rule. The method for assigning hierarchical labels to the target data based on the multi-dimensional weighted evaluation rule includes: setting multiple evaluation indicators and determining the score of the target data under each evaluation indicator; using the analytic hierarchy process (AHP) to determine the weight of each evaluation indicator; calculating the score of the target data based on the score of the target data under each evaluation indicator and the weight of each evaluation indicator; and obtaining the hierarchical label of the target data based on the pre-obtained data hierarchical rules and the score of the target data, and assigning it to the target data.

[0046] Specifically, three core evaluation indicators are set: data breach impact (indicator 1), data uniqueness (indicator 2), and compliance requirement level (indicator 3). For data breach impact, the target data is scored according to its impact on user rights and business operations after the breach, resulting in an impact score. For data uniqueness, the target data is scored according to its irreplaceability, resulting in a uniqueness score. For example, if the data is unique, the score is 10 points; if the data can be repeatedly generated, the score is 2 points. For compliance requirement level, the target data is scored according to its protection level. For example, if the data requires mandatory protection, the score is 10 points; if the data is publicly available, the score is 1 point.

[0047] Furthermore, using the Analytic Hierarchy Process (AHP), the weight of each evaluation indicator is determined: First, a three-layer structure is established: the target layer (core decision-making objectives), the criteria layer (key factors influencing the objectives), and the alternative layer (specific alternative solutions or underlying indicators). The core objective of the target layer is to determine the weights of the three core evaluation indicators, while the criteria layer includes these three core evaluation indicators. For the core objective of the target layer, the three core evaluation indicators are compared pairwise using a 1-9 scale (1 - equally important, 3 - slightly important, 5 - significantly important, 7 - strongly important, 9 - extremely important, reciprocal for reverse comparison), resulting in a judgment matrix. The product M of the elements in each row of the judgment matrix is ​​then calculated. i Let i = 1, 2, 3. Calculate M. i The cube root of the value is normalized to obtain the initial weight W. A W B W C Calculate the product of the judgment matrix and the initial weights, and then calculate the largest eigenvalue based on the product. Calculate the consistency index based on the largest eigenvalue, and then look up the average random consistency index from a table based on the consistency index. Divide the consistency index by the average random consistency index to obtain the consistency ratio. If the consistency ratio is ≤0.1, adjust the initial weights W... A W B W C As the final weight. W A W B W C These represent the weights of the three indicators. It should be noted that this is only a brief explanation of the Analytic Hierarchy Process (AHP); for specific implementation details, please refer to existing methods, which will not be elaborated upon further.

[0048] Furthermore, the score of the target data is calculated using the formula "Indicator 1 × Weight 1 + Indicator 2 × Weight 2 + Indicator 3 × Weight 3". The data grading rules define multiple score intervals and corresponding grading labels for each interval. Based on the target data's score, the score interval to which the target data belongs in the data grading rules is determined, and the grading label corresponding to the determined score interval is the grading label for the target data. The types of grading labels include: extremely sensitive data, highly sensitive data, moderately sensitive data, and low sensitive data.

[0049] For example, data classification rules include: extremely sensitive data (8-10 points), highly sensitive data (6-7 points), moderately sensitive data (3-5 points), and low-sensitivity data (1-2 points). If the target data scores 9 points, then the target data is classified as extremely sensitive data.

[0050] Step S14: Based on the hierarchical label of the target data, acquire and securely process the target data according to the hierarchical privacy protection rules to obtain processed target data; based on the processed target data, complete the target task.

[0051] In one embodiment, hierarchical privacy protection rules are constructed for data with different security levels and in combination with the type of target task. This ensures that the de-identified data meets the functional requirements of the expert agent (such as data availability and consistency of statistical features) while reducing the risk of sensitive data leakage.

[0052] Based on the hierarchical labels of the target data, the target data is acquired and processed securely according to hierarchical privacy protection rules. This includes: determining the acquisition method and security processing method of the target data according to the hierarchical labels of the target data within the hierarchical privacy protection rules, and acquiring and processing the target data according to the determined acquisition method and security processing method. The expert agent then completes the target task based on the processed target data.

[0053] In one embodiment, the tiered privacy protection rules include:

[0054] For target data categorized as "extremely sensitive," the module storing the target data and the expert agent transmit data via end-to-end encryption. End-to-end encryption refers to an encryption technique where only the communicating parties can read the information, and intermediate links cannot decrypt it. End-to-end encryption employs a rolling key change mechanism, where the encryption key is automatically updated at set intervals. The rolling key change mechanism is generated, stored, and updated by the Hardware Security Module (HSM). The Hardware Security Module is physically isolated from the host where the expert agent is deployed and communicates with the expert agent only through an encrypted interface to prevent key leakage. It should be understood that the Hardware Security Module (HSM) is a physical device (or embedded chip) specifically designed to protect sensitive data, keys, and encryption operations.

[0055] After receiving encrypted target data, the expert agent decrypts and uses it within a local Trusted Execution Environment (TEE, such as Intel SGX). The decrypted data resides only in the TEE's memory and is not persisted to storage. The memory data is automatically cleared within 100ms after the operation is completed. It should be understood that non-persistent storage means that data is not written to persistent storage media such as local disks or server hard drives during transmission and processing; it is only temporarily processed in memory and then cleared.

[0056] When highly sensitive data is used for agent model training, data analysis, and other analytical tasks, a hash algorithm is used to convert the decrypted data into a fixed-length hash value (e.g., facial data generates a 64-bit string after hashing). The hash value is only used for model feature matching and cannot be reverse-engineered from the original data. A hash value is a fixed-length string of characters obtained after processing the original data using a hash algorithm. Hash values ​​can be used for quickly comparing features, for example, determining whether two data sets are identical (if the hash values ​​match, the original data is likely identical), or as feature identifiers for retrieval and matching scenarios, which is far more efficient than directly processing the original data. Furthermore, hash algorithms are unidirectional; a hash value can be calculated from the original data, but the original data cannot be derived from the hash value. For example, the hash value of "123" might be "a1b2c3," but "123" cannot be obtained from "a1b2c3." This is a key characteristic of hash algorithms, ensuring the security of the original data. Even if the hash value is leaked, the original information cannot be recovered.

[0057] For target data labeled as highly sensitive, a portion of the target data undergoes synonymous hashing, while another portion undergoes random hashing. For example, if the target data is a mobile phone number, the first three digits (number segment) and the middle four digits (area code) are processed using synonymous hashing (the same number segment / area code corresponds to a fixed hash value, such as "135" being a fixed hash of "aefsc", and "0379" being a fixed hash of "fdac"). The last four digits are processed using random hashing (the same last digit generates different hash values ​​each time, such as "1234" being hashed as "xqwe1" or "zrty2"). The processed mobile phone number is: "aefscfdacxqwe1".

[0058] In interactive scenarios where user interaction is required, a mask (such as "aefsc****xqwe1") can be overlaid on the processed target data. When the expert agent performs data correlation analysis internally, it directly uses the processed data (without overlaying the mask), ensuring a balance between data correlation and security.

[0059] It should be noted that both synonymous hashing and random hashing use irreversible hashing algorithms (such as BLAKE3), which can reverse the parsing of the original data without a key, and different types of highly sensitive data use different hash seeds to avoid cross-data type association leakage.

[0060] For target data categorized as moderately sensitive, data with clearly defined numerical ranges are replaced with data intervals. For example, a user's actual spending amount of "156 yuan" is replaced with "100-200 yuan", and the number of times a user visits a doctor "3 times" is replaced with "1-5 times". For text-based data, semantic replacement is used, such as replacing "visited on October 15, 2025" with "visited in Q4, 2025", and "type 2 diabetes" with "metabolic chronic diseases".

[0061] Furthermore, when sensitive data is used for analytical tasks such as agent statistical analysis and model training, a perturbation conforming to a normal distribution is added to the replaced target data, with the perturbation amplitude ≤5%. The data after perturbation must meet the statistical characteristic deviation (such as mean deviation, variance deviation), with the statistical characteristic deviation ≤2%, to ensure that it does not affect the accuracy of agent model training and the reliability of business analysis results.

[0062] For target data with the classification label of low-sensitivity data, perform field filtering on the low-sensitivity data and delete unnecessary fields. For example, in the Agent operation log, only keep "operation time (accurate to the second), operation type (such as knowledge base data query), operation result (success, failure)" and delete irrelevant fields such as "user redundancy identifier (such as temporary session ID) and Agent internal process ID".

[0063] It should be noted that this invention achieves differentiated protection of sensitive data through tiered privacy protection rules and ensures transmission security through secure interaction protocols. The processed data retains its statistical characteristics and business relevance. The secure interaction protocols and access controls do not affect the core functions of the expert agent, such as diagnosis and decision-making. The agent response latency increases by ≤100ms, meeting the needs of practical applications.

[0064] In one embodiment, the privacy protection and data security processing method for the expert agent further includes: collecting full-process operation data of the expert agent through tracking technology. The full-process operation data includes, but is not limited to, the operation subject ID (e.g., user, system number), operation type (e.g., knowledge retrieval, data reading, data modification), operation object information (e.g., data ID, data level), operation time, operation result (success or failure), access IP, device identifier, etc. The collection frequency is real-time (latency ≤ 100ms), and the log storage adopts an immutable blockchain structure (each log block contains the hash value of the previous block). It should be understood that tracking technology refers to pre-setting monitoring points (i.e., tracking points) in the expert agent code to record the expert agent's operation data.

[0065] Furthermore, a rules engine is built, which contains multiple pre-defined security rules. It should be understood that security rules can be set according to actual use cases and needs; no restrictions are imposed here. For example, security rules could be: "Failed permission verification 5 times consecutively" and "Access to extremely sensitive data outside of working hours."

[0066] A training dataset is constructed by obtaining historical full-process operation data from multiple expert agents. This training dataset is then used to train a combined model of a random forest and an isolation forest model to obtain the auditing model. In one implementation of the combined model, the isolation forest first filters and processes abnormal data (e.g., removing or correcting it), and then the cleaned data is input into the random forest for core prediction, or the results of both are weighted and fused. Further, a rule engine and the auditing model are used to analyze the full-process operation data to obtain abnormal operation identification results. Three levels of alarm thresholds are set: low-level, medium-level, and high-level alarm thresholds. Each level corresponds to a different alarm method (e.g., SMS, email, or pop-up notifications on the operations platform). Alarm information includes details of the abnormal operation (e.g., operation subject, operation time, and abnormality type) and is associated with a log blockchain address for one-click tracing of the entire operation process.

[0067] It should be noted that this embodiment combines a combined model and a rule engine, which can identify both known anomalies and detect unknown risks, with a false positive rate of ≤5%. This invention achieves full-process control of operations through access control and auditing, covering the entire lifecycle of data storage, transmission, and use.

[0068] In the embodiments of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.

[0069] It should be noted that, in the embodiments of this application, the words "exemplary" or "for example" indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0070] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0071] Figure 2 This is a schematic block diagram of the privacy protection and data security processing device for the expert agent provided in the embodiments of this application. Figure 2 As shown, the privacy protection and data security processing device 200 of the expert agent includes:

[0072] Module 201 is used to acquire the target task and its corresponding task attribute information;

[0073] The permission judgment module 202 is used to perform permission judgment operations based on the task attribute information and a pre-built hybrid permission model to obtain the corresponding permission judgment result.

[0074] The call confirmation module 203 is used to determine the target data to be called based on the target task if the permission judgment result indicates that the user has the necessary permissions; wherein the target data has hierarchical labels assigned based on multi-dimensional weighted evaluation rules.

[0075] The privacy protection and data security processing module 204 is used to acquire and securely process the target data based on the hierarchical label of the target data and hierarchical privacy protection rules to obtain processed target data; and to complete the target task based on the processed target data.

[0076] It should be understood that the specific process of each module performing the above-mentioned steps has been described in detail in the above method embodiments, and will not be repeated here for the sake of brevity.

[0077] It should also be understood that the module division in the embodiments of this application is illustrative and only represents a logical functional division; in actual implementation, there may be other division methods. Furthermore, the functional modules in the various embodiments of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0078] In one embodiment, the hybrid permission model includes a role layer and an attribute layer; wherein, the role layer defines multiple roles and their basic permissions; the attribute layer defines multiple attributes; and the attribute layer further includes multiple decision rules based on the multiple attributes and the multiple roles.

[0079] In one embodiment, the method of assigning graded labels to target data based on multi-dimensional weighted evaluation rules includes: setting multiple evaluation indicators and determining the score of the target data under each evaluation indicator; using the analytic hierarchy process (AHP) to determine the weight of each evaluation indicator; calculating the score of the target data based on the score of the target data under each evaluation indicator and the weight of each evaluation indicator; and obtaining graded labels for the target data based on pre-obtained data grading rules and the score of the target data and assigning them to the target data.

[0080] In one embodiment, the types of the hierarchical labels include: extremely sensitive data, highly sensitive data, moderately sensitive data, and low sensitive data.

[0081] In one embodiment, the hierarchical privacy protection rules include: if the hierarchical label of the target data is extremely sensitive data, then the target data is obtained through end-to-end encryption and decrypted in a local trusted execution environment; if the target task is an analysis task, then the decrypted target data is hashed; wherein the end-to-end encryption method uses a rolling key change; if the hierarchical label of the target data is highly sensitive data, then the target data is subjected to synonym hashing and random hashing; if the hierarchical label of the target data is moderately sensitive data, then the target data is subjected to interval replacement; if the target task is an analysis task, then a perturbation conforming to a normal distribution is added to the replaced target data; the data after adding perturbation satisfies a pre-set statistical characteristic deviation; if the hierarchical label of the target data is low sensitive data, then redundant fields are deleted and the format is standardized.

[0082] In one embodiment, the privacy protection and data security processing device of the expert agent further includes: an audit module, used to acquire full-process operation data; using the security rules preset in the rule engine and the pre-trained audit model, to obtain abnormal operation identification results based on the full-process operation data, and to issue corresponding alarms based on the abnormal operation identification results; wherein, the audit model is obtained by training a combined model from the training dataset; the combined model consists of a random forest model and an isolated forest model.

[0083] Figure 3 This is a schematic block diagram of the electronic terminal provided in an embodiment of this application. Figure 3As shown, the electronic terminal includes at least one processor 301, a memory 302, at least one network interface 303, and a user interface 305. The various components in the device are coupled together via a bus system 304. It is understood that the bus system 304 is used to implement communication between these components. In addition to a data bus, the bus system 304 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 3 The general will label all buses as bus systems.

[0084] The user interface 305 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0085] It is understood that memory 302 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable categories of memory.

[0086] In this embodiment of the invention, the memory 302 is used to store various types of data to support the operation of the electronic terminal 300. Examples of this data include: any executable program for operation on the electronic terminal 300, such as the operating system 3021 and application programs 3022; the operating system 3021 contains various system programs, such as the framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. The application program 3022 may contain various applications, such as media players, browsers, etc., for implementing various application services. The privacy protection and data security processing method for the expert agent provided in this embodiment of the invention can be included in the application program 3022.

[0087] The methods disclosed in the above embodiments of the present invention can be applied to processor 301, or implemented by processor 301. Processor 301 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 301 or by instructions in the form of software. The processor 301 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 301 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. General-purpose processor 301 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of the present invention can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0088] In an exemplary embodiment, the electronic terminal 300 may be used by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs) to execute the aforementioned method.

[0089] According to the method provided in the embodiments of this application, this application also provides a computer program product, which includes: computer program code, which, when run on a computer, causes the computer to execute... Figure 1 The privacy protection and data security processing method of the expert agent in the illustrated embodiment.

[0090] According to the method provided in the embodiments of this application, this application also provides a computer-readable storage medium storing program code, which, when executed on a computer, causes the computer to perform... Figure 1 The privacy protection and data security processing method of the expert agent in the illustrated embodiment.

[0091] As used in this specification, the terms "component," "module," "system," etc., are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or a computer. As illustrated, applications running on computing devices and computing devices can both be components. One or more components may reside in a process and / or an execution thread, and components may be located on a single computer and / or distributed among two or more computers. Furthermore, these components can be executed from various computer-readable media on which various data structures are stored. Components can communicate, for example, via local and / or remote processes based on signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system, and / or a network, such as the Internet interacting with other systems via signals).

[0092] Those skilled in the art will recognize that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0093] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0094] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0095] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0096] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0097] In the above embodiments, the functions of each functional unit can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. A computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs, DVDs), or semiconductor media (e.g., solid-state disks, SSDs, etc.).

[0098] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0100] In summary, this application provides a method, apparatus, terminal, medium, and product for privacy protection and data security processing of an expert agent. This method, applied to an expert agent, includes: acquiring a target task and its corresponding task attribute information; performing a permission judgment operation based on a pre-built hybrid permission model according to the task attribute information to obtain the corresponding permission judgment result; determining the target data to be invoked based on the target task if the permission judgment result indicates that permission is granted; acquiring and securely processing the target data according to its hierarchical tags and hierarchical privacy protection rules to obtain processed target data; and completing the target task based on the processed target data. This application, through data hierarchical classification and privacy protection and data security mechanisms, effectively protects data privacy and security while ensuring the normal operation of the expert agent. Therefore, this application effectively overcomes the various shortcomings of existing technologies and has high industrial application value.

[0101] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for privacy protection and data security processing of an expert agent, characterized in that, Applied to expert agents, including: Obtain the target task and its corresponding task attribute information; the task attribute information includes user level, user role type, user department, data level, data domain, access IP address, and access time; Based on the task attribute information, and using a pre-built hybrid permission model, a permission judgment operation is performed to obtain the corresponding permission judgment result. The hybrid permission model includes a role layer and an attribute layer. The role layer defines multiple roles and their basic permissions. The attribute layer defines multiple attributes. The attribute layer also includes multiple decision rules defined based on the various attributes and roles. Based on the task attribute information and a pre-built hybrid permission model, permission judgment operations are performed to obtain the corresponding permission judgment results. These include: The role layer checks whether the user's role has the basic permission to execute the target task based on the user role type in the task attribute information; if so, the attribute layer performs verification; if not, a permission judgment result of "no permission" is generated; the attribute layer selects one or more decision rules related to the target task from a set of predefined decision rules, and then matches the task attribute information with these rules; if all one or more of the selected decision rules match successfully, a permission judgment result of "permitted" is generated; otherwise, a permission judgment result of "no permission" is generated. If the permission determination result indicates that the user has the necessary permissions, the target data to be invoked is determined based on the target task; wherein, the target data has hierarchical labels assigned based on multi-dimensional weighted evaluation rules. Based on the hierarchical labels of the target data, the target data is acquired and security-processed according to hierarchical privacy protection rules to obtain processed target data; based on the processed target data, the target task is completed. The tiered privacy protection rules include: If the target data is labeled as extremely sensitive, the target data is obtained through end-to-end encryption and decrypted in a local trusted execution environment; if the target task is an analysis task, the decrypted target data is hashed; wherein, the end-to-end encryption method uses a rolling key change. If the target data is labeled as highly sensitive, then synonym hashing and random hashing are performed on the target data. If the target data is labeled as moderately sensitive, then the target data is replaced by intervals; if the target task is an analysis task, then a perturbation conforming to a normal distribution is added to the replaced target data; the data after adding perturbation satisfies the pre-defined statistical characteristic deviation. If the target data's hierarchical labels are low-sensitivity data, then redundant fields will be removed and the format will be standardized.

2. The method for privacy protection and data security processing of the expert agent according to claim 1, characterized in that, Based on multi-dimensional weighted evaluation rules, the methods for assigning hierarchical labels to target data include: Set multiple evaluation indicators and determine the scores of the target data under each evaluation indicator; The weights of each evaluation indicator are determined using the analytic hierarchy process (AHP). The score of the target data is calculated based on the scores of the target data under each evaluation indicator and the weight of each evaluation indicator. Based on the pre-obtained data grading rules and the score of the target data, the grading labels of the target data are obtained and assigned to the target data.

3. The method for privacy protection and data security processing of the expert agent according to claim 2, characterized in that, The types of hierarchical labels include: extremely sensitive data, highly sensitive data, moderately sensitive data, and low sensitive data.

4. The method for privacy protection and data security processing of the expert agent according to claim 1, characterized in that, The method further includes: acquiring full-process operation data; using the pre-set security rules in the rule engine and the pre-trained audit model, obtaining abnormal operation identification results based on the full-process operation data, and issuing corresponding alarms based on the abnormal operation identification results; wherein, the audit model is obtained by training a combined model from the training dataset; the combined model consists of a random forest model and an isolated forest model.

5. A privacy protection and data security processing device for an expert agent, characterized in that, Located within the expert agent, it includes: The acquisition module is used to acquire the target task and its corresponding task attribute information; the task attribute information includes user level, user role type, user department, data level, data domain, access IP address, and access time. The permission judgment module is used to perform permission judgment operations based on the task attribute information and a pre-built hybrid permission model to obtain the corresponding permission judgment result. The hybrid permission model includes a role layer and an attribute layer. The role layer defines multiple roles and their basic permissions. The attribute layer defines multiple attributes. The attribute layer also includes multiple decision rules defined based on the multiple attributes and roles. Based on the task attribute information and a pre-built hybrid permission model, permission judgment operations are performed to obtain the corresponding permission judgment results. These include: The role layer checks whether the user's role has the basic permission to execute the target task based on the user role type in the task attribute information; if so, the attribute layer performs verification; if not, a permission judgment result of "no permission" is generated; the attribute layer selects one or more decision rules related to the target task from a set of predefined decision rules, and then matches the task attribute information with these rules; if all one or more of the selected decision rules match successfully, a permission judgment result of "permitted" is generated; otherwise, a permission judgment result of "no permission" is generated. The call confirmation module is used to determine the target data to be called based on the target task if the permission judgment result indicates that the user has the necessary permissions; wherein the target data has hierarchical labels assigned based on multi-dimensional weighted evaluation rules. The privacy protection and data security processing module is used to acquire and securely process the target data based on the hierarchical tags of the target data and hierarchical privacy protection rules to obtain processed target data; and to complete the target task based on the processed target data. The tiered privacy protection rules include: If the target data is labeled as extremely sensitive, the target data is obtained through end-to-end encryption and decrypted in a local trusted execution environment; if the target task is an analysis task, the decrypted target data is hashed; wherein, the end-to-end encryption method uses a rolling key change. If the target data is labeled as highly sensitive, then synonym hashing and random hashing are performed on the target data. If the target data is labeled as moderately sensitive, then the target data is replaced by intervals; if the target task is an analysis task, then a perturbation conforming to a normal distribution is added to the replaced target data; the data after adding perturbation satisfies the pre-defined statistical characteristic deviation. If the target data's hierarchical labels are low-sensitivity data, then redundant fields will be removed and the format will be standardized.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 4.

7. A computer program product, characterized in that, The computer program product includes computer program code that, when run on a computer, causes the computer to implement the method as described in any one of claims 1 to 4.

8. An electronic terminal, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the method according to any one of claims 1 to 4.

Images