Dynamic adaptive authority management method and dynamic adaptive authority management system

Abstract

The invention relates to the field of information security, and provides a dynamic adaptive authority management method and a dynamic adaptive authority management system. The method comprises the steps that target data are obtained, the target data comprise user behavior data and environmental factor data, the user behavior data are used for indicating operation behaviors when a user uses equipment, and the environmental factor data are used for indicating the environment when the user uses the equipment; inputting the target data into a first model to obtain a weighted risk score output by the first model; inputting the weighted risk score into a second model to obtain user permission information output by the second model; and presenting the user permission information. Through the scheme, the authority is dynamically adjusted based on the real-time behavior of the user and the environment, the problems of insufficient static authority management adaptability and lack of context perception are solved, and the user experience is improved while the safety is guaranteed.

Description

Technical Field

[0001] This application relates to the field of information security, and in particular to a dynamic adaptive permission management method and a dynamic adaptive permission management system. Background Technology

[0002] With the widespread use of mobile devices in both personal and business settings, device security and data protection have become increasingly important. Users need solutions that ensure security without compromising user experience, while businesses require systems that can flexibly manage employee device permissions and adapt to different work scenarios. Furthermore, with the development of the Internet of Things (IoT) and communication technologies, the demand for security management solutions that can adapt to complex and dynamic network environments is also growing.

[0003] Based on the above needs, there are currently two main types of permission management technologies on the market: In traditional rule-based risk scoring systems, user behavior is scored using a predefined set of rules to assess security risks. In static permission management systems, permissions are granted to users when they install or use the application for the first time, and these permissions remain unchanged thereafter.

[0004] However, the aforementioned access control technologies have shortcomings. For example, access control lacks adaptability; rule-based systems struggle to cope with complex and ever-changing usage scenarios and cannot accurately capture the behavioral characteristics and risk patterns of individual users. Furthermore, they lack context awareness; static access control systems cannot dynamically adjust permissions based on the user's environment and behavioral patterns, leading to either insufficient security or a compromised user experience. Summary of the Invention

[0005] In view of this, this application provides a dynamic adaptive permission management method and a dynamic adaptive permission management system. In this method, user behavior and environmental data are acquired and input into a first model based on an attention mechanism and a long short-term memory (LSTM) neural network to generate a weighted risk score. Then, a second model combines this score with the effectiveness of permission adjustments and behavioral consistency to dynamically determine and present the user's permission level. This achieves dynamic adjustment of permissions based on real-time user behavior and the environment, solving the problems of insufficient adaptability and lack of context awareness in static permission management, thus improving user experience while ensuring security.

[0006] Firstly, a dynamic adaptive access control method is provided, comprising: acquiring target data, which includes user behavior data and environmental factor data, wherein the user behavior data indicates the user's operational behavior when using the device, and the environmental factor data indicates the environment in which the user is using the device; inputting the target data into a first model to obtain a weighted risk score output by the first model, wherein the first model determines the weights and risk scores corresponding to multiple risk categories based on the target data and an attention mechanism, and inputs historical behavior sequences from the user behavior data into an LSTM neural network to obtain the environmental context information of the historical behavior sequence output by the LSTM neural network, and determines the historical risk trend information of the historical behavior sequence; finally, determining the weighted risk score based on the weights and risk scores corresponding to the multiple risk categories, the environmental context information of the historical behavior sequence, and the historical risk trend information; inputting the weighted risk score into a second model to obtain user access control information output by the second model, wherein the second model determines the user access control information based on the weighted risk score, the effectiveness of access control adjustments, and the consistency of user behavior; and presenting the user access control information. This solution enables dynamic adjustment of permissions based on real-time user behavior and environment, solving the problems of insufficient adaptability and lack of context awareness in static permission management, and improving user experience while ensuring security.

[0007] In conjunction with the first aspect, in a possible implementation of the first aspect, the acquisition of target data includes: authenticating the user and collecting the user's target data based on the subscriber identity module (SIM) card.

[0008] This solution utilizes the Super SIM card as the security root and data collection entry point, ensuring the security of user authentication and the reliability of data sources.

[0009] In conjunction with the first aspect, among the possible implementations of the first aspect, presenting the user permission level includes: based on the Super SIM card, packaging and encrypting the user permission information, and providing a standardized application programming interface (API) through a security protocol; in response to a third-party application requesting user permission data from the Super SIM card, sending the encrypted user permission information to the third-party application. This solution provides a standardized security interface through the Super SIM card, ensuring the privacy and security of permission information transmission and facilitating secure integration by third-party applications.

[0010] In conjunction with the first aspect, in a possible implementation of the first aspect, the user permission information includes the user permission level.

[0011] This solution provides clear and actionable permission level information, making it easier for the system to make permission control decisions.

[0012] In conjunction with the first aspect, in a possible implementation of the first aspect, the target data also includes user authentication status information used to indicate whether the user has been authenticated.

[0013] This solution incorporates authentication status into risk assessment factors, thereby strengthening the security foundation of access control.

[0014] In conjunction with the first aspect, in possible implementations of the first aspect, the environmental factor data includes the user's geographical location, time information, and one or more aspects of the network environment.

[0015] This solution comprehensively considers various environmental factors, making risk assessment and permission adjustments more aligned with users' actual usage scenarios.

[0016] In conjunction with the first aspect, in a possible implementation of the first aspect, the first model, based on the target data and an attention mechanism, determines the weights and risk scores corresponding to multiple risk categories by: based on the predefined multiple risk categories, the first model inputs the user behavior data into a deep neural network to obtain one or more user behavior features corresponding to each risk category; the first model encodes the environmental factor data into a graph structure, and based on the attention mechanism, the context vector of the current time step, and the embedding vector of the risk category, determines the weight of each risk category and the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category; based on the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category, the risk score corresponding to each risk category is determined.

[0017] This scheme utilizes deep neural networks and attention mechanisms, combined with graph-structured environment encoding, to achieve multi-dimensional, fine-grained risk category weights and scoring calculations, thereby improving the accuracy and adaptability of risk assessment.

[0018] In conjunction with the first aspect, among the possible implementations of the first aspect, the multiple risk categories include one or more of the following: identity authentication risk, data access risk, and application usage risk.

[0019] This solution covers the main risk areas of mobile device security, making the assessment more comprehensive.

[0020] In conjunction with the first aspect, in a possible implementation of the first aspect, determining the historical risk trend information of the historical behavior sequence includes: performing a nonlinear transformation on the risk scores of multiple past time points in the historical behavior sequence to obtain the transformed risk scores; and obtaining the historical risk trend information of the historical behavior sequence based on the average value of the transformed risk scores of the multiple time points and the tanh function.

[0021] This scheme effectively captures long-term risk change trends by processing historical risk scores through nonlinear transformation and the tanh function, providing important historical references for current risk assessment.

[0022] In conjunction with the first aspect, in a possible implementation of the first aspect, after obtaining the weighted risk score output by the first model, the method further includes: setting a sliding time window including L time steps and storing the risk scores of the L time steps; determining a risk score threshold based on the average and standard deviation of the risk scores of the L time steps; and determining whether the user has abnormal behavior based on the comparison between the weighted risk score and the risk score threshold.

[0023] This scheme, through sliding window and dynamic threshold calculation, enables adaptive detection of abnormal behavior and timely discovery of potential security threats.

[0024] In conjunction with the first aspect, in a possible implementation of the first aspect, the second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior. This includes: the second model determines the effectiveness score of n permission adjustments based on the risk mitigation effectiveness index, the operation smoothness maintenance coefficient, and the resource response optimization metric, where n is a positive integer; and determines the effectiveness of the permission adjustment based on the effectiveness score of each permission adjustment.

[0025] This solution quantifies the effectiveness of past permission adjustments in terms of security, user experience, and efficiency, such as risk mitigation effectiveness index, operational smoothness maintenance coefficient, and resource response optimization metric, making current permission decisions more scientific and effective, and avoiding ineffective or negative adjustments.

[0026] In conjunction with the first aspect, in a possible implementation of the first aspect, the second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior. This includes: the second model determines the consistency of user behavior based on the current values ​​and historical average values ​​of the user's m behavioral characteristics.

[0027] This scheme compares the current values ​​of user behavior characteristics with historical averages to determine whether user behavior deviates from the normal pattern, providing a basis for the stability of behavior patterns in permission adjustments, thereby enhancing the reliability of permission management and user experience.

[0028] In conjunction with the first aspect, in a possible implementation of the first aspect, after determining the user's permission information, the method further includes: generating a permission change instruction based on a comparison between the user's permission information and the user's current permission information.

[0029] This solution automatically generates specific permission change instructions (elevation, demotion, or retention) based on the evaluation results, enabling dynamic and automated adjustment of permissions.

[0030] In conjunction with the first aspect, in a possible implementation of the first aspect, the method further includes: changing the current permission information to the user permission information, obtaining a first weight coefficient before the user permission change and a second weight coefficient after the user weight change; and determining the evaluation result of the user permission change based on the first weight coefficient and the second weight coefficient. This approach uses a comparison of weighting coefficients before and after the change to quantitatively evaluate the effectiveness of permission changes, providing data support for subsequent permission policy optimization.

[0031] In conjunction with the first aspect, among the possible implementations of the first aspect, presenting the user's permission information includes: presenting the user's permission change notification, such as a new permission level or a reason for the permission change, or one or more of these.

[0032] This solution transparently displays permission changes and their reasons to users, improving user experience and system trustworthiness.

[0033] In conjunction with the first aspect, among the possible implementations of the first aspect, the method further includes: presenting the weighted risk score; and, if the weighted risk score changes, presenting the changed weighted risk score and the reason for the change.

[0034] This solution enables users to understand the current risk status and its changes, enhances their security awareness, and helps them understand the basis for system decisions.

[0035] Secondly, a dynamic adaptive permission management system is provided, including: a super SIM card module for acquiring target data, which includes user behavior data and environmental factor data. The user behavior data is used to indicate the user's operating behavior when using the device, and the environmental factor data is used to indicate the environment in which the user is using the device. The holographic adaptive risk derivation and dynamic modeling module is used to input the target data into the first model to obtain the weighted risk score output by the first model. The first model, based on the target data and an attention mechanism, determines the weights and risk scores corresponding to multiple risk categories. It also inputs historical behavior sequences from the user behavior data into an LSTM neural network to obtain the environmental context information of the historical behavior sequence output by the LSTM neural network, as well as the historical risk trend information of the historical behavior sequence. Finally, based on the weights and risk scores corresponding to the multiple risk categories, the environmental context information of the historical behavior sequence, and the historical risk trend information, the weighted risk score is determined. The cross-domain adaptive permission reconstruction and control module is used to input the weighted risk score into a second model to obtain user permission information output by the second model. The second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustments, and the consistency of user behavior. The presentation module is used to present the user permission information.

[0036] Through this solution, the system integrates the security authentication and data collection of the Super SIM card, the holographic risk modeling and assessment (MABRAF), and the cross-domain dynamic permission adjustment (MAPAA) capabilities, thus constructing a complete dynamic permission management system that can respond in real time to changes in the environment and behavior.

[0037] Thirdly, an electronic device is provided, including one or more processors; one or more memories; the one or more memories storing one or more computer programs, the one or more computer programs including instructions that, when executed by the one or more processors, cause a method as described in any of the first aspect to the foregoing possible implementations to be performed.

[0038] Fourthly, a computer-readable storage medium is provided that stores computer instructions that, when executed on a computer, cause a method as described in any of the first aspect to the aforementioned possible implementations to be performed.

[0039] Fifthly, a chip is provided, the chip including a processor and a communication interface for receiving a signal and transmitting the signal to the processor, the processor processing the signal such that a method as described in any of the first aspect to the foregoing possible implementations is executed.

[0040] In a sixth aspect, a computer program product is provided that, when the computer program product is run on a computer, causes the computer to perform a method as described in the first aspect to any of the aforementioned possible implementations.

[0041] It should be understood that the beneficial effects of the second to sixth aspects mentioned above can be referred to the beneficial effects described in the first aspect, and will not be repeated here. Attached Figure Description

[0042] Figure 1 A schematic flowchart of the dynamic adaptive permission management method provided in an embodiment of this application is shown; Figure 2 A schematic diagram of the module flow of the method provided in the embodiments of this application is shown; Figure 3 This illustration shows a schematic diagram of a third-party application security status overview interface provided in an embodiment of this application; Figure 4 This illustration shows a schematic diagram of a risk score change pop-up notification provided in an embodiment of this application; Figure 5 This illustration shows a user permission change notification pop-up window provided in an embodiment of this application; Figure 6 This illustration shows a schematic diagram of a third-party application security status weekly report provided in an embodiment of this application; Figure 7 This is a structural schematic diagram of a device provided in an embodiment of this application; Figure 8 This is a structural schematic diagram of a system on a chip (SoC) provided in an embodiment of this application. Detailed Implementation

[0043] The technical solution of this application is described below with reference to the accompanying drawings.

[0044] As mentioned in the background section, existing mobile device security management technologies suffer from insufficient adaptability and a lack of context awareness.

[0045] To address the aforementioned issues, this application proposes a dynamic adaptive permission management method 100, which can be applied to a dynamic adaptive permission management system. In this method 100, user behavior and environmental data are acquired, and a weighted risk score is generated by a first model based on an attention mechanism and an LSTM neural network. A second model then combines this score with the effectiveness of permission adjustments and behavioral consistency to dynamically determine and present the user's permission level. This achieves dynamic permission adjustment based on real-time user behavior and the environment, solving the problems of insufficient adaptability and lack of context awareness in static permission management, thus improving user experience while ensuring security.

[0046] Figure 1 A schematic flowchart of a dynamic adaptive permission management method 100 provided in an embodiment of this application is shown. Figure 1 As shown, the method 100 includes steps S110 to S140.

[0047] Step S110: Obtain target data.

[0048] Specifically, the target data includes user behavior data and environmental factor data. The user behavior data indicates the user's operational behavior when using the device. The environmental factor data indicates the environment in which the user is using the device.

[0049] In step S110, the dynamic adaptive access control system first collects basic data, or target data, for assessing user risk. This target data serves as the input for subsequent risk modeling and access control decisions, and primarily includes user behavior data and environmental factor data.

[0050] For example, user behavior data can refer to log information generated by a user during the operation of their mobile device, such as a smartphone or tablet, that reflects their operating habits and intentions. For instance, it may include the starting, closing, and background running status of applications; or it may include file or database access records such as read, write, and delete operations; or it may include the target address (URL / IP), port, and protocol type of network requests; or it may include the number of authentication attempts, success / failure status, and authentication methods used, such as password, fingerprint, or facial recognition.

[0051] For example, environmental factor data can refer to information describing the user's current physical or network environment status, which directly affects the context of risk assessment. In embodiments of this application, the environmental factor data includes the user's geographic location, time information, and one or more of the network environment information. For example, geographic location can indicate latitude and longitude coordinates obtained through GPS, base station positioning, or Wi-Fi triangulation, or a location accurate to the building level; time information can indicate the current specific time (year, month, day, hour, minute, second); and the network environment can indicate the type of network the device is connected to, such as home Wi-Fi, corporate intranet, public Wi-Fi, or cellular network, or it can indicate the device's signal strength, such as the Received Signal Strength Indicator (RSSI) value.

[0052] It should be understood that in the embodiments of this application, environmental factors such as geographical location, time information, and network environment are core elements constituting context awareness. The embodiments of this application comprehensively consider the correlation between these three factors. For example, if it is detected that a user logs into their account at an unusual time and location via an insecure network called "Free_WiFi," even if authentication is successful, this combined environment will be marked as a high-risk context. Conversely, if a user performs the same operation on a weekday in a company office building via a dedicated Wi-Fi network, it is considered a low-risk context. This fusion analysis of multi-dimensional environmental data makes risk assessment more closely resemble real-world scenarios.

[0053] Optionally, in embodiments of this application, acquiring the target data includes: authenticating the user and collecting the user's target data based on the Super SIM card. The Super SIM card is a new type of SIM card integrating an enhanced security chip. It not only supports traditional communication functions but also has a built-in independent secure execution environment. For example, in embodiments of this application, when a user inserts the Super SIM card and powers on the device, the system first triggers an authentication process, such as requiring a PIN code or performing biometric verification. After successful authentication, the Super SIM card is activated. Subsequently, the Super SIM card can utilize its secure channel to collect the aforementioned user behavior data and environmental factor data in real time from the operating system's underlying layer or dedicated API interfaces. Because data collection occurs in a hardware-level secure environment, it effectively prevents malicious software from tampering with or eavesdropping, ensuring the authenticity and integrity of the data. For example, when a user attempts to access a sensitive file, the Super SIM card records the behavioral characteristics of this operation, such as the file path, operation type, and the environment in which it occurs, such as location and network, and encrypts and temporarily stores this raw data in its internal secure storage area.

[0054] Optionally, in embodiments of this application, the target data further includes user authentication status information indicating whether the user has passed authentication. This user authentication status information can be used to determine the security starting point of the current session. For example, in embodiments of this application, the user authentication status information can include three states: authenticated, indicating that the user has passed at least one authentication method, such as PIN and fingerprint; unauthenticated, such as when the device has just been powered on or is in a locked state; and authentication failed, such as multiple failed attempts potentially triggering a temporary lockout. In embodiments of this application, this user authentication status information can be generated by the Super SIM card after completing the authentication process and transmitted as part of the target data along with other data to the subsequent processing module. For example, if the user only authenticates via a simple PIN code, their authentication status is recorded as medium, which will be used as a factor in the subsequent risk assessment, affecting the final weighted risk score.

[0055] Step S120: Input the target data into the first model to obtain the weighted risk score.

[0056] Specifically, the first model determines the weights and risk scores corresponding to multiple risk categories based on the target data and the attention mechanism; and inputs the historical behavior sequences in the user behavior data into the LSTM neural network to obtain the environmental context information of the historical behavior sequence output by the LSTM neural network, as well as the historical risk trend information of the historical behavior sequence. Finally, based on the weights and risk scores corresponding to the multiple risk categories, the environmental context information of the historical behavior sequence, and the historical risk trend information, the weighted risk score is determined.

[0057] In this step, the system inputs the target data obtained in step S110 into the first model. In the embodiments of this application, the first model can also be referred to as a holographic adaptive risk derivation and dynamic modeling framework, which calculates a comprehensive weighted risk score through a series of complex algorithms. This score is a core quantitative indicator for measuring the user's current security status.

[0058] Optionally, in embodiments of this application, the first model determines the weights and risk scores corresponding to multiple risk categories based on the target data and the attention mechanism, including: based on the predefined multiple risk categories, the first model inputs the user behavior data into a deep neural network to obtain one or more user behavior features corresponding to each risk category; the first model encodes the environmental factor data into a graph structure, and determines the weight of each risk category and the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category based on the attention mechanism, the context vector of the current time step, and the embedding vector of the risk category; and determines the risk score corresponding to each risk category based on the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category.

[0059] For example, these multiple risk categories are a predefined risk classification system. For instance, in embodiments of this application, three core risk categories can be defined: identity authentication risk, which can be used to monitor the authenticity and strength of user identity authentication, such as weak passwords, logins from different locations, and frequent failed attempts; data access risk, which can be used to monitor unauthorized or abnormal access to sensitive data such as ID photos and financial statements; and application usage risk, which can be used to monitor the installation and use of high-risk applications such as APKs from unknown sources or applications with background monitoring permissions.

[0060] For example, in this process, a deep neural network acts as a feature extractor, receiving raw user behavior data streams such as operation log sequences and automatically learning and outputting user behavior features associated with each risk category through its hidden layers. For instance, for a single file access behavior, the DNN might output a set of feature vectors: [File sensitivity = high, access time = early morning, user recent activity = low].

[0061] For example, a graph structure is a data structure that visualizes environmental factors and their interrelationships. In embodiments of this application, the system can treat geographic location, time information, and network environment as three nodes and establish edges based on their co-occurrence patterns. For instance, time information and network environment form a strongly correlated edge. Then, a graph neural network (GNN) is used to encode the graph, generating an embedding vector for each node, thereby transforming discrete environmental information into continuous numerical vectors.

[0062] The attention mechanism is a key innovation of the first model. Specifically, this first model computes an attention weight that depends on the current context vector from LSTM or environment graph encoding and the embedding vector of the risk category. For example, in a public Wi-Fi environment, the attention weight for the authentication risk category is significantly amplified. Ultimately, the risk score for each risk category is a weighted sum of all its features.

[0063] Optionally, in an embodiment of this application, determining the historical risk trend information of the historical behavior sequence includes: performing a nonlinear transformation on the risk scores of multiple past time points in the historical behavior sequence to obtain the transformed risk scores; and obtaining the historical risk trend information of the historical behavior sequence based on the average value of the transformed risk scores of the multiple time points and the tanh function.

[0064] For example, a historical behavior sequence can refer to a stream of user behavior events arranged chronologically over a past period, such as the past 7 days. Historical risk trend information is used to characterize the direction of evolution of a user's risk level, and this historical risk trend information will be input as long-term memory into the final weighted risk score calculation.

[0065] Optionally, in an embodiment of this application, after obtaining the weighted risk score output by the first model, the method further includes: setting a sliding time window including L time steps and storing the risk scores of the L time steps; determining a risk score threshold based on the average and standard deviation of the risk scores of the L time steps; and determining whether the user has abnormal behavior based on a comparison between the weighted risk score and the risk score threshold.

[0066] The sliding time window is a first-in, first-out queue of length L, used to dynamically maintain the weighted risk scores for the most recent L moments. For example, L can be 24, representing 24 hours. When the weighted risk score at the current moment exceeds the risk score threshold, the system can determine that the user has engaged in abnormal behavior and trigger an alarm or initiate a higher-priority review process. For example, a user whose risk score is usually stable between 30 and 50, if suddenly reaches 85 at a certain point in time, and exceeds their dynamic threshold such as 60, will be immediately identified as abnormal.

[0067] Step S130: Input the weighted risk score into the second model to obtain user permission information.

[0068] Specifically, the second model determines the user's permission information based on the weighted risk score, the effectiveness of permission adjustments, and the consistency of user behavior.

[0069] In step S130, the system inputs the weighted risk score output in step S120 into the second model. In embodiments of this application, the second model may also be referred to as the cross-domain adaptive permission reconstruction and control model.

[0070] Optionally, in an embodiment of this application, the second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior, including: the second model determines the effect score of n permission adjustments based on the risk mitigation effectiveness index, the operation smoothness maintenance coefficient, and the resource response optimization metric, where n is a positive integer; and determines the effectiveness of the permission adjustment based on the effect score of each permission adjustment.

[0071] The effectiveness of permission adjustments refers to the overall impact of past permission changes on actual application. User behavior consistency measures the degree of deviation between current behavior patterns and long-term user habits. The risk mitigation effectiveness index measures the actual contribution of permission adjustments to reducing risk. The operational smoothness maintenance coefficient measures the impact of permission adjustments on user experience. The resource response optimization metric measures the degree to which permission adjustments optimize system resource consumption.

[0072] In step S130, the effectiveness score is a weighted combination of the risk mitigation effectiveness index, the operational smoothness maintenance coefficient, and the resource response optimization metric. By averaging the effectiveness scores of the past n adjustments, the effectiveness of the current permission adjustment can be obtained, which is used to guide the decision-making for this permission adjustment.

[0073] Optionally, in an embodiment of this application, the second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior, including: the second model determines the consistency of user behavior based on the current value and historical average value of the user's m behavioral characteristics.

[0074] Optionally, in an embodiment of this application, after determining the user permission information, the method further includes: generating a permission change instruction based on a comparison between the user permission information and the user's current permission information.

[0075] For example, a permission change instruction can be a control command used to instruct the operating system or application to perform specific permission modification operations. For instance, if newly determined user permission information indicates that the permission level should be elevated from "medium" to "high," an instruction to elevate the permission is generated; otherwise, an instruction to degrade the permission or maintain the status quo is generated.

[0076] Optionally, in embodiments of this application, the method further includes: changing the current permission information to the user permission information, obtaining a first weight coefficient before the user permission change and a second weight coefficient after the user weight change; and determining the evaluation result of the user permission change based on the first weight coefficient and the second weight coefficient.

[0077] Specifically, the first weighting coefficient and the second weighting coefficient are quantitative parameters reflecting the importance of user permissions. The evaluation result can be based on the difference between the first weighting coefficient and the second weighting coefficient. This result can be used for policy optimization and to assess the scope of impact of this permission change.

[0078] Step S140: Present user permission information.

[0079] In other words, step S140 displays the determined user permission information to the outside world in the form of visualization or interface call.

[0080] Optionally, in embodiments of this application, presenting the user permission level includes: based on the Super SIM card, packaging and encrypting the user permission information, and providing a standardized API interface through a security protocol; in response to a third-party application requesting user permission data from the Super SIM card, sending the encrypted user permission information to the third-party application.

[0081] The Super SIM card can use its built-in hardware encryption engine to encrypt authorization information, generating unreadable ciphertext data. Security protocols can refer to communication protocols such as HTTPS and TLS / SSL, ensuring the confidentiality and integrity of data during network transmission. Standardized API interfaces can refer to a set of predefined programming interfaces that conform to industry standards (such as RESTful APIs).

[0082] Optionally, in embodiments of this application, the user permission information includes user permission levels. For example, user permission levels can be represented as a hierarchical label, such as low, medium, high, and restricted, or as numerical levels such as 1 to 10.

[0083] Optionally, in embodiments of this application, presenting the user's permission information includes: presenting the user's permission change notification, including one or more of the following: a new permission level and a reason for the permission change. For example, when a user's permissions are adjusted, the system may display a prompt on the device's lock screen or notification bar: Your permissions have been updated! Current level: Medium. The reason for the permission change may include detecting a connection to an insecure public Wi-Fi network. The reason for the permission change can directly reference the dominant factors in the risk assessment, such as excessively high authentication risk or abnormal data access behavior.

[0084] Optionally, in embodiments of this application, the method further includes: presenting the weighted risk score; and, if the weighted risk score changes, presenting the changed weighted risk score and the reason for the change. For example, the current risk score can be displayed in real time in the form of a dashboard, such as "Current security score: 75 / 100. When the score changes significantly, such as rising from 75 to 85, an automatic push notification is sent: Security score updated! Reason: Multiple login attempts from different locations were detected."

[0085] The following is combined Figures 2 to 6 This application provides a detailed description of its embodiments.

[0086] Figure 2 A schematic diagram of the module flow of the method 100 provided in an embodiment of this application is shown. Wherein, as... Figure 2 As shown, the dynamic adaptive permission management system includes users, a super SIM card, a holographic adaptive risk derivation and dynamic modeling module (hereinafter referred to as the MABRAF module), a cross-domain adaptive permission reconstruction and control module (hereinafter referred to as the MAPAA module), and third-party applications. The following describes embodiments of this method 100 from steps one to six.

[0087] Step 1: Corresponding to step S110, the user performs basic authentication and multi-dimensional environmental data collection through the Super SIM.

[0088] Specifically, users can leverage the inherent authentication capabilities of the Super SIM card to achieve rapid and secure user identification, while comprehensively collecting user behavior and environmental data. This first step aims to provide a rich and reliable data foundation for subsequent analysis, ensuring the system accurately grasps user usage patterns and their environment, thus laying a solid foundation for dynamic permission assessment. Through multi-dimensional data collection, the system's ability to understand user contexts is enhanced.

[0089] For example, in an embodiment of this application, step one may be performed in detail as follows: a) The Super SIM card automatically completes user authentication based on the phone number.

[0090] b) Activate the multi-dimensional data collection module to continuously monitor and record user operation behavior.

[0091] c) The environmental data acquisition submodule acquires the user's geographical location, time information, network environment, and other file data in real time.

[0092] d) Perform preliminary processing on the collected data, including noise reduction, standardization, and anonymization.

[0093] e) Data is encrypted and stored using the secure storage mechanism of the Super SIM card.

[0094] The main outputs of this step one include: user authentication status (pass / fail), processed user behavior dataset, and environmental factor dataset.

[0095] Step 2: Corresponding to step S120, the MABRAF module, based on the collected dataset, applies a multi-dimensional perception focusing architecture using artificial intelligence technology to perform intelligent behavior analysis and dynamic risk assessment, generating a holographic weighted risk score R(t).

[0096] Step two of this paper designs a "Holographic Adaptive Risk Derivation and Dynamic Modeling Framework" (MABRAF), also known as the first model mentioned earlier. Compared with existing traditional rule-based risk scoring systems, this step achieves dynamic weight adjustment, multi-dimensional risk assessment, and real-time behavioral sequence analysis. By integrating attention mechanisms and multi-level risk category assessment, MABRAF effectively solves the problem of insufficient adaptability in existing technologies.

[0097] Specifically, MABRAF can dynamically adjust the importance of different risk categories based on the current context, capture long-term dependencies in user behavior using time-series networks, and provide a more comprehensive risk profile through multi-level assessment. This approach not only significantly improves the accuracy and real-time performance of risk assessment but also enhances the system's adaptability to complex and dynamic security environments, providing a more reliable decision-making basis for subsequent access control.

[0098] MABRAF Holographic Weighted Risk Score The calculation formula is as follows:

[0099] This MABRAF framework calculates a holographic weighted risk score for the current moment by integrating different categories of behavioral risk, environmental context, and historical trends. It can adaptively adjust the weights of various levels and factors, respond accurately to complex behavioral patterns and environmental changes, and take into account the cumulative effects of historical risks, providing a comprehensive, dynamic and predictive risk assessment result.

[0100] The parameters in the formula are explained as follows: Let represent the holographic weighted risk score at time t; σ is the sigmoid function used to map the score to the interval [0,1]. It is the number of risk categories; It is the dynamic weight of the k-th type of risk at time t; It is the score of the k-th type of risk at time t; This indicates the use of LSTM networks to process user spatiotemporal dynamic behavior chain data sequences. The output; m is the size of the time window for considering historical risks (defined by business needs; when t <= m, then...). (A preset initial credit score, set by the business). This represents the holographic weighted risk score at time j in the past. and These are the combined weighting coefficients for environmental context and historical risk.

[0101] The following is an introduction Detailed calculation process.

[0102] First, multi-level dynamic behavioral chain feature extraction is performed. Specifically, K user risk categories are defined, including identity authentication risk (number of login attempts, number of password changes), data access risk (access frequency per unit time, data transfer volume), and application usage risk (average application usage time, application click frequency), etc. Then, for each category, a deep neural network (such as a combination of CNN and LSTM) is used to extract high-level features from the raw behavioral data. Finally, the extracted features are normalized to obtain... Used for later calculations .

[0103] Next, dynamic context-aware weight generation is performed. Specifically, contextual information such as time, location, and device status is encoded into a graph structure, and the weight of each risk category is calculated. and the weight of each behavioral feature The calculation formula is as follows:

[0104] This method implements a dynamic risk category weight calculation based on an attention mechanism. It combines current context information... and risk category characteristics Using nonlinear transformations and softmax normalization, importance scores for each risk category were generated. This method adaptively adjusts the weights of different risk categories based on real-time context, improving the model's responsiveness to complex and dynamic security environments. It allows the system to flexibly focus on the most relevant risk categories under different circumstances, thereby improving the accuracy and timeliness of risk assessment while enhancing the model's interpretability. It is the context vector of the current time step, which can be generated by the GNN algorithm for graph structures; It is the embedding vector of the k-th risk category; , , and It consists of learnable parameter matrices and vectors; This represents the total number of risk categories.

[0105] Next, a single-category dynamic threshold evolution risk score is calculated. Specifically, for each risk category k, a weighted average is used to obtain the risk score for that category. The calculation formula is as follows:

[0106] This formula calculates the risk score for the k-th risk category at time t. It synthesizes the anomalousness of all behavioral characteristics within that category through a weighted summation, where the anomalousness of each characteristic is amplified by an exponential function and standardized using its historical mean and standard deviation. Dynamic adjustment of the weights ensures adaptability to the current context, while normalization of the denominator balances the impact of different numbers of characteristics. This design can keenly capture anomalous behavior while remaining robust to normal fluctuations, thus providing an accurate, dynamic, and interpretable risk assessment mechanism.

[0107] in, It is the score of the k-th type of risk at time t; It is the number of behavioral characteristics included in the k-th type of risk; It is the dynamic weight of the i-th behavioral feature in the k-th risk category at time t; It is the observed value of the i-th behavioral feature in the k-th risk category at time t; and These are the historical mean and standard deviation of the feature, respectively.

[0108]

[0109] in, It is in time Time The first of the risk categories The dynamic weight of each behavioral characteristic determines the importance of that characteristic to risk assessment at the current moment. It is the sigmoid activation function, which maps the input to the (0,1) interval. It is the first Individual behavioral characteristics in time The feature vector includes specific behavioral metrics such as the number of login attempts and the number of password changes. It is time The context vector contains environmental information such as login time and login location. It is a learnable weight vector. It is a learnable bias term that increases the flexibility of the model. This represents a vector concatenation operation, which combines information from different sources into a unified input vector.

[0110] Next, environmental context sequence evolution analysis is performed. Specifically, an LSTM network is used to process the user's historical behavior sequences. (Including user login time, login location, app opening history, etc. over a past period), to obtain .

[0111] Next, a historical nonlinear risk trend analysis is performed. Specifically, the tanh function is first used to perform a nonlinear transformation on the risk scores over the past m time points. Then, the average of the transformed scores is calculated to capture the cumulative effect and changing trend of the risk.

[0112] Then, a holographic weighted risk score is calculated. Specifically, by combining the parameters calculated above, the sigmoid function is used to map the final result to the [0,1] interval, thus obtaining the holographic weighted risk score R(t).

[0113] Finally, adaptive anomaly detection is performed. Specifically, a sliding window of length L is first maintained for each user, storing the risk scores of the most recent L time steps. Then, a dynamic threshold is calculated using the following formula:

[0114] in, It is the average value of the holographic weighted risk score within the sliding window; It is the standard deviation of the holographic weighted risk score within the sliding window; It is a dynamic adjustment factor, calculated using the following formula:

[0115] in, It is the basic adjustment factor. This is the learning rate. The expression within parentheses calculates the proportion of scores that exceed the previous threshold in the most recent L time steps. Represents the holographic weighted risk score for time t.

[0116] After calculating the dynamic threshold θ(t), anomaly detection is performed. If the current risk score R(t) exceeds the dynamic threshold θ(t), it is marked as a potential anomaly. The judgment process is as follows:

[0117] Here, 1 represents the presence of a potential anomaly, and 0 represents the absence of an anomaly.

[0118] The main output of this step is: a holographic weighted risk score. Abnormal behavior detection results (abnormal / no abnormality).

[0119] These outputs are used for displaying the safety status overview interface, providing pop-up alerts for changes in risk scores, and generating weekly safety status reports. For example, Figure 3 This illustration shows a schematic diagram of a third-party application security status overview interface provided in an embodiment of this application. Figure 4 This illustration shows a schematic diagram of a risk score change pop-up notification provided in an embodiment of this application.

[0120] Step 3: Corresponding to step S130, the MAPAA module uses holographic weighted risk scoring. By applying nonlinear transformation and cross-domain data fusion algorithms, adaptive cruise control permission levels are generated. .

[0121] This step designs a cross-domain adaptive permission reconstruction and control algorithm module, also known as the second model mentioned earlier. Compared with existing threshold-based permission management systems, this method achieves multi-dimensional information fusion, non-linear permission adjustment, and a self-optimization mechanism. MAPAA effectively solves the lack of context awareness in existing technologies by comprehensively considering factors such as current risk score, risk change rate, impact of time-series permission tracking, and user behavior stability, and by introducing non-linear transformations such as exponential, logarithmic, and improved sigmoid transformations, thus achieving a better balance between user experience and security. Specifically, MAPAA can dynamically adjust permission levels based on the user's real-time behavior and environment, providing a smooth and personalized permission change experience, while continuously learning and adapting to user usage patterns through a self-optimization mechanism. This method not only improves the accuracy and flexibility of permission management but also maximizes user experience while ensuring security, providing a more intelligent and user-friendly solution for mobile device security management.

[0122] The core formula of MAPAA is as follows:

[0123] This method calculates a comprehensive permission adjustment index by integrating current holographic weighted risk scores, risk entropy evolution trend measurements, temporal permission tracking impact, and robustness characteristics of user nonlinear behavior. It can quickly respond to sudden risks while considering long-term behavioral patterns, achieving a balance between accuracy and stability in permission management. By introducing multiple dimensions and nonlinear transformations, the algorithm provides a more granular and dynamic permission adjustment mechanism, significantly improving the system's adaptability to complex security scenarios.

[0124] The parameters in the formula are explained as follows: This represents the permission level at time t, after processing by the sigmoid function. Mapping to the [0,1] range can directly correspond to the predefined permission levels of the business (different levels are assigned different usage rights, and the specific rights set is predefined by the business). It is the holographic weighted risk score at time t, which can be calculated from step two. It represents the rate of change of the trend of risk entropy increase, and captures the trend of rapid rise or fall of risk. It is a time-series permission tracking impact score, which reflects the effectiveness of past adjustment decisions. It is a robustness index for nonlinear sequences of user behavior, taking into account the long-term behavioral characteristics of users. It is the environmental situation awareness index, which is calculated by taking into account contextual information such as time and location. These are the weighting coefficients of each factor, which are dynamically adjusted through machine learning algorithms. It is an index used to adjust the holographic weighted risk score and the sensitivity to rate of change.

[0125] The specific implementation process of the algorithm is described below.

[0126] First, the rate of change of the risk entropy evolution trend is calculated.

[0127] Using a sliding window for calculation To capture short-term risk trends.

[0128]

[0129] in It refers to the window size. The user's holographic weighted risk score representing time t. The user's comprehensive holographic weighted risk score represents the time period tw.

[0130] Next, an impact analysis of time-series permission tracking is performed. Specifically, the impact of this analysis is calculated. Track and evaluate the effectiveness of past permission adjustments.

[0131]

[0132] in This is the score for the effectiveness of the i-th permission adjustment. It is the time decay factor. It can be calculated using the following formula:

[0133] This formula calculates the overall effect of each permission adjustment by integrating scores from three dimensions: risk mitigation effectiveness index, operational smoothness maintenance coefficient, and resource response optimization metric, while also considering the time factor. It comprehensively assesses the impact of permission adjustments and reflects how the adjustment's effect changes over time. By introducing multiple evaluation dimensions and a time decay factor, this function provides a more comprehensive and dynamic effect evaluation mechanism.

[0134] The parameters in the formula are explained as follows: It is the effect score of the i-th permission adjustment, with a value range of [0, 2]. It is a risk mitigation effectiveness index score, reflecting the changes in application safety status after adjustment. ,in and These are the holographic weighted risk scores before and after the adjustment. It is a smoothness of operation coefficient score, which measures the impact of adjustments on the user's operating experience. ,in and It represents the success rate of m key user operations (predefined by the business logic) before and after the adjustment. It represents the weight of each operation. It is a resource response optimization metric score that assesses the impact of adjustments on the overall system performance. ,in and It is the average system response time before and after the adjustment. , , These are the weights for security, user experience, and system performance, respectively. It is the sigmoid function, used to map scores to the interval [0, 1]. It is the time decay coefficient. This is the time elapsed since the permissions were adjusted.

[0135] Following this, robustness analysis of nonlinear sequences of user behavior is performed. Specifically, the following calculations are performed. This reflects the consistency of user behavior.

[0136]

[0137] in, It is the current value of the j-th behavioral feature at time t (such as the number of login attempts, the number of password changes, the amount of data transferred, the average usage time of the application, etc.). It is its historical average. This indicates the total number of behavioral characteristics.

[0138] Next, the cruise permission level P(t) is calculated. Specifically, it is calculated using the preceding core formula. And mapped to discrete permission levels.

[0139] Finally, the final permission change command is generated. Specifically, a comparison is made. Generate a corresponding change instruction based on the current permission level. This can be calculated using the following method:

[0140] in, It is a permission change instruction, that is, a suggested change in permission level. It represents the current user's permission level. It is a dynamically adjusted change threshold (predefined by the business). It is a symbolic function used to determine the direction of change. This indicates rounding down to the nearest integer.

[0141] according to Generate specific change instructions: If Generate a privilege escalation command; if Generate a demotion command; if No change instructions are generated.

[0142] In addition, this application also designs a self-optimization mechanism that can periodically evaluate the effect of permission adjustments and update the weight coefficients. The formula for this permission assessment is as follows:

[0143] in Represents the weighting coefficient. It's the learning rate. It is the effect evaluation function.

[0144] This step primarily outputs: User permission level. Permission change command.

[0145] These outputs are used in third-party application security status overview interfaces, user permission change notification pop-ups, and third-party application security status weekly reports. For example, Figure 5 This illustration shows a user permission change notification pop-up window provided in an embodiment of this application. Figure 6 This illustration shows a schematic diagram of a third-party application security status weekly report provided in an embodiment of this application.

[0146] Step 4: Corresponding to step S140, the Super SIM card packages the user's cruise permission level and final change instruction into an information body, and provides a security interface and third-party application integration service to the outside world.

[0147] Based on the output of the above steps, the SuperSIM platform provides secure and standardized interfaces for third-party applications, enabling them to conveniently and securely access access level information. This is achieved through an encrypted communication protocol mechanism to ensure the security of data transmission and usage. Comprehensive documentation and sample code lower the integration barrier, promoting wider application adoption and thus expanding the system's influence and practical value.

[0148] The main outputs of step four are: API interface specifications, encrypted communication protocol, SDK, API documentation, sample code, and developer support resources.

[0149] In summary, this application presents an innovative "Holographic Adaptive Risk Derivation and Dynamic Modeling Framework" (MABRAF) calculation method. First, through a super SIM card module, the system collects user behavior data and environmental context information in real time, forming a raw dataset. After entering the preprocessing module, the data undergoes denoising, standardization, and anonymization to ensure data quality and privacy protection. Next, the system utilizes a multi-level neural network to extract multi-level dynamic behavioral chain features and performs weighted evaluation of risk categories through nonlinear transformations (such as Sigmoid and tanh). The weight of each risk category is quantified through dynamic context-aware calculation based on an attention mechanism, allowing the influence of different risk factors to adaptively adjust according to the real-time context. The core risk score R(t) is calculated by comprehensively considering the weighted single-class dynamic threshold evolution risk score of each risk category, the evolution progress of the environmental context sequence, and historical nonlinear risk trend measurement. Simultaneously, the system employs a dynamic weight adjustment mechanism to update the influence of each behavioral feature in real time based on changes in user behavior characteristics and environmental factors.

[0150] The MABRAF algorithm significantly improves the accuracy and adaptability of risk assessment by introducing dynamic weight adjustment and multi-level risk assessment. Specifically, the MABRAF algorithm employs an attention mechanism to dynamically adjust the weights of different risk categories and combines it with a temporal network to process user historical behavior sequences, achieving accurate responses to complex and dynamic security environments. This method not only captures the long-term dependencies of user behavior but also adapts to changes in risk importance in different scenarios, thus providing more accurate and real-time risk assessments than traditional static models when facing complex and ever-changing mobile device usage environments.

[0151] Furthermore, this application also designs an innovative "Cross-Domain Adaptive Permission Restructuring and Control Algorithm" (MAPAA) calculation method. First, the algorithm uses a holographic weighted risk score R(t), combined with current user behavior data and environmental context information, to standardize the risk using a nonlinear transformation, obtaining the risk score at the current moment. Next, it generates a risk entropy evolution trend measure (|dR / dt|) by calculating the rate of change of the risk score, capturing dynamic changes in user behavior and risk fluctuations, and further enhances the sensitivity to risk changes using an exponential weighting method, determining the trend direction through a sign function (sign(dR / dt)). Second, it generates a time-series permission tracking impact degree H(t) by introducing the effects of historical permission adjustments, quantifying the impact of past permission changes on the current state, and using time decay weighting to reflect long-term effects. Then, it calculates the user's nonlinear behavior robustness feature U(t) to quantify the consistency of user behavior patterns. Finally, by comprehensively utilizing these nonlinearly transformed and weighted indicators, the algorithm outputs the final permission adjustment value P(t), which reflects the comprehensive impact of multi-dimensional data on permission management.

[0152] The MAPAA algorithm achieves a more refined and dynamic permission adjustment mechanism by integrating multi-dimensional information and nonlinear transformations. MAPAA not only considers the current risk score but also incorporates factors such as the rate of risk change, historical permission adjustment effects, and user behavior stability. Through exponential, logarithmic, and improved sigmoid function nonlinear transformations, it enhances sensitivity to different input ranges. This design allows MAPAA to maximize user experience while ensuring security, avoiding frequent and abrupt permission changes in traditional systems and providing a smoother, more personalized permission adjustment experience. Furthermore, MAPAA's self-optimization mechanism continuously learns and adapts to user patterns, further improving the accuracy and user-friendliness of permission management.

[0153] The hardware module of this application is described below, which can be used to implement the aforementioned method 100.

[0154] Now for reference Figure 7The diagram shows a block diagram of a device 700 according to one embodiment of this application. Device 700 may include one or more processors 701 coupled to a controller hub 703. In at least one embodiment, the controller hub 703 communicates with the processor 701 via a multi-branch bus such as a front side bus (FSB), a point-to-point interface such as a quickpath interconnect (QPI), or a similar connection 710. The processor 701 executes instructions controlling general types of data processing operations. In one embodiment, the controller hub 703 includes, but is not limited to, a graphics memory controller hub (GMCH) (not shown) and an input / output hub (IOH) (which may be on a separate chip) (not shown), wherein the GMCH includes memory and a graphics controller and is coupled to the IOH.

[0155] Device 700 may also include a coprocessor 702 and a memory 704 coupled to a controller hub 703. Alternatively, one or both of the memory and the GMCH may be integrated within the processor, with memory 704 and coprocessor 702 directly coupled to processor 701 and controller hub 703, which resides on a single chip with the IOH. Memory 704 may be, for example, dynamic random access memory (DRAM), phase change memory (PCM), or a combination of both. In one embodiment, coprocessor 702 is a dedicated processor, such as, for example, a high-throughput MIC processor (many integrated core, MIC), a network or communication processor, a compression engine, a graphics processor, a general-purpose computing on GPU (GPGPU), or an embedded processor, etc. Optional properties of coprocessor 702 are indicated by dashed lines. Figure 7 middle.

[0156] As a computer-readable storage medium, memory 704 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. For example, memory 704 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device such as one or more hard-disk drives (HDD(s)), one or more compact disc (CD) drives, and / or one or more digital versatile disc (DVD) drives.

[0157] In one embodiment, device 700 may further include a network interface controller (NIC) 706. NIC 706 may include a transceiver for providing a radio interface to device 700, thereby enabling communication with any other suitable device (such as a front-end module, antenna, etc.). In various embodiments, NIC 706 may be integrated with other components of device 700. NIC 706 can implement the functionality of the communication unit in the above embodiments.

[0158] Device 700 may further include input / output (I / O) device 705. I / O 705 may include: a user interface designed to enable a user to interact with device 700; a peripheral component interface designed to enable peripheral components to also interact with device 700; and / or sensors designed to determine environmental conditions and / or location information related to device 700.

[0159] It is worth noting that, Figure 7 This is merely an example. That is, although... Figure 7 The diagram shows that device 700 includes multiple devices such as processor 701, controller hub 703, and memory 704. However, in actual applications, devices using the methods of this application may include only a portion of the devices in device 700. For example, it may include only processor 701 and NIC 706. Figure 7 The properties of the optional devices are shown in dashed lines. According to some embodiments of this application, the memory 704, which is a computer-readable storage medium, stores instructions that, when executed on a computer, cause the device 700 to perform the methods according to the above embodiments. Specific details can be found in the methods of the above embodiments, and will not be repeated here.

[0160] Now for reference Figure 8 The diagram shown is a block diagram of a system-on-chip (SoC) 800 according to an embodiment of this application. Figure 8In the diagram, similar components share the same reference numerals. Additionally, dashed boxes are an optional feature for more advanced SoCs. Figure 8 In this SoC 800, the following are included: an interconnect unit 850 coupled to an application processor 810; a system proxy unit 880; a bus controller unit 890; an integrated memory controller unit 840; a group or one or more coprocessors 820, which may include integrated graphics logic, an image processor, an audio processor, and a video processor; a static random access memory (SRAM) unit 830; and a direct memory access (DMA) unit 860. In one embodiment, the coprocessor 820 includes a dedicated processor, such as, for example, a network or communication processor, a compression engine, a GPGPU, a high-throughput MIC processor, or an embedded processor.

[0161] The static random-access memory (SRAM) unit 830 may include one or more computer-readable media for storing data and / or instructions. The computer-readable storage medium may store instructions, specifically, temporary and permanent copies of those instructions. These instructions may include, when executed by at least one unit in the processor, causing the SoC 800 to perform the attention training method according to the above embodiments, as detailed in the methods described above, which will not be repeated here.

[0162] This application provides a computer program product that, when run on an electronic device, causes the electronic device to execute the technical solutions described in the above embodiments. Its implementation principle and technical effects are similar to those of the related embodiments described above, and will not be repeated here.

[0163] This application provides a readable storage medium containing instructions that, when executed by an electronic device, cause the electronic device to perform the technical solution described in the above embodiments. The implementation principle and technical effects are similar and will not be repeated here.

[0164] This application provides a chip for executing instructions. When the chip is running, it executes the technical solutions described in the above embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0165] Various embodiments of the mechanisms disclosed in this application can be implemented in hardware, software, firmware, or combinations of these implementation methods. Embodiments of this application can be implemented as computer programs or program code executable on a programmable system, the programmable system including at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device.

[0166] Program code can be applied to input instructions to execute the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, the processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application-specific integrated circuit (ASIC), or a microprocessor.

[0167] The program code can be implemented using a high-level procedural language or an object-oriented programming language to communicate with the processing system. Assembly language or machine language can also be used when needed. In fact, the mechanisms described in this application are not limited to any particular programming language. In either case, the language can be a compiled language or an interpreted language.

[0168] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored thereon on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or through other computer-readable media. Therefore, machine-readable media may include any mechanism for storing or transmitting information in a machine-readable (e.g., computer-readable) form, including but not limited to floppy disks, optical disks, CD-ROMs, compact disc read-only memory (CD-ROMs), magneto-optical disks, read-only memory (ROM), random-access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic cards or optical cards, flash memory, or tangible machine-readable storage for transmitting information (e.g., carrier waves, infrared signals, digital signals, etc.) using the Internet in the form of electrical, optical, acoustic, or other forms of propagated signals. Therefore, machine-readable media includes any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a machine-readable (e.g., computer-readable) form.

[0169] In the accompanying drawings, some structural or methodological features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be necessary. Rather, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the accompanying drawings. Furthermore, including structural or methodological features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may be omitted or may be combined with other features.

[0170] It should be noted that all units / modules mentioned in the device embodiments of this application are logical units / modules. Physically, a logical unit / module can be a physical unit / module, a part of a physical unit / module, or a combination of multiple physical units / modules. The physical implementation of these logical units / modules themselves is not the most important factor; the combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed in this application. Furthermore, to highlight the innovative aspects of this application, the above-described device embodiments of this application have not introduced units / modules that are not closely related to solving the technical problems proposed in this application. This does not mean that the above-described device embodiments do not contain other units / modules.

[0171] It should be noted that in the examples and description of this patent, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0172] Although this application has been illustrated and described with reference to certain preferred embodiments thereof, those skilled in the art should understand that various changes in form and detail may be made thereto without departing from the spirit and scope of this application.

Claims

1. A dynamic adaptive permission management method, characterized in that, include: Acquire target data, which includes user behavior data and environmental factor data. The user behavior data is used to indicate the user's operating behavior when using the device, and the environmental factor data is used to indicate the environment in which the user is using the device. The target data is input into a first model to obtain a weighted risk score output by the first model. The first model determines the weights and risk scores corresponding to multiple risk categories based on the target data and an attention mechanism. The historical behavior sequences in the user behavior data are input into an LSTM neural network to obtain the environmental context information of the historical behavior sequences output by the LSTM neural network and to determine the historical risk trend information of the historical behavior sequences. Finally, the weighted risk score is determined based on the weights and risk scores corresponding to the multiple risk categories, the environmental context information of the historical behavior sequences, and the historical risk trend information. The weighted risk score is input into the second model to obtain the user permission information output by the second model. The second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior. Present the user permission information.

2. The method according to claim 1, characterized in that, The acquisition of target data includes: The user is authenticated and the target data of the user is collected based on the Super SIM card.

3. The method according to claim 2, characterized in that, The presentation of the user permission information includes: Based on the super SIM card, the user permission information is packaged and encrypted, and a standardized API interface is provided through a security protocol; In response to a third-party application requesting user permission data from the Super SIM card, the encrypted user permission information is sent to the third-party application.

4. The method according to any one of claims 1 to 3, characterized in that, The user permission information includes the user permission level.

5. The method according to any one of claims 1 to 3, characterized in that, The target data also includes user authentication status information to indicate whether the user has passed authentication.

6. The method according to any one of claims 1 to 3, characterized in that, The environmental factor data includes the user's geographical location, time information, and one or more factors related to the network environment.

7. The method according to any one of claims 1 to 3, characterized in that, The first model, based on the target data and the attention mechanism, determines the weights and risk scores corresponding to multiple risk categories, including: Based on the predefined multiple risk categories, the first model inputs the user behavior data into a deep neural network to obtain one or more user behavior features corresponding to each risk category; The first model encodes the environmental factor data into a graph structure, and determines the weight of each risk category among the multiple risk categories and the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category based on the attention mechanism, the context vector of the current time step and the embedding vector of the risk category. Based on the weight of each user behavior feature among the one or more user behavior features corresponding to each risk category, a risk score is determined for each of the multiple risk categories.

8. The method according to claim 7, characterized in that, The multiple risk categories include one or more of the following: identity authentication risk, data access risk, and application usage risk.

9. The method according to any one of claims 1 to 3, characterized in that, The historical risk trend information for determining the historical behavior sequence includes: A nonlinear transformation is performed on the risk scores of multiple past time points in the historical behavior sequence to obtain the transformed risk scores; Based on the average of the transformed risk scores at the multiple time points and the tanh function, the historical risk trend information of the historical behavior sequence is obtained.

10. The method according to any one of claims 1 to 3, characterized in that, After obtaining the weighted risk score output by the first model, the method further includes: Set up a sliding time window with L time steps and store the risk scores for the L time steps, where L is a positive integer; The risk score threshold is determined based on the average and standard deviation of the risk scores over the L time steps. The user's abnormal behavior is determined by comparing the weighted risk score with the risk score threshold.

11. The method according to any one of claims 1 to 3, characterized in that, The second model, based on the weighted risk score, the effectiveness of permission adjustments, and the consistency of user behavior, determines the user permission information, including: The second model determines the effectiveness score of n permission adjustments based on the risk mitigation effectiveness index, the operational smoothness maintenance coefficient, and the resource response optimization metric, where n is a positive integer; The effectiveness of each permission adjustment is determined by scoring its impact.

12. The method according to any one of claims 1 to 3, characterized in that, The second model, based on the weighted risk score, the effectiveness of permission adjustments, and the consistency of user behavior, determines the user permission information, including: The second model determines the consistency of the user's behavior based on the current values ​​and historical average values ​​of the user's m behavioral features, where m is a positive integer.

13. The method according to any one of claims 1 to 3, characterized in that, After determining the user permission information, the method further includes: Based on the comparison between the user's permission information and the user's current permission information, a permission change instruction is generated.

14. The method according to claim 13, characterized in that, The method further includes: Based on the current permission information, change it to the user permission information, and obtain the first weight coefficient before the user permission change and the second weight coefficient after the user weight change; The evaluation result of the user permission change is determined based on the first weighting coefficient and the second weighting coefficient.

15. The method according to any one of claims 1 to 3, characterized in that, The presentation of the user permission information includes: The system displays a notification of changes to the user's permissions, including one or more of the following: the new permission level and the reason for the permission change.

16. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Present the weighted risk score; If the weighted risk score changes, the revised weighted risk score and the reason for the change will be presented.

17. A dynamic adaptive permission management system, characterized in that, include: The Super SIM card module is used to acquire target data, which includes user behavior data and environmental factor data. The user behavior data is used to indicate the user's operating behavior when using the device, and the environmental factor data is used to indicate the environment in which the user is using the device. The holographic adaptive risk derivation and dynamic modeling module is used to input the target data into a first model to obtain a weighted risk score output by the first model. The first model determines the weights and risk scores corresponding to multiple risk categories based on the target data and an attention mechanism. It also inputs the historical behavior sequences in the user behavior data into an LSTM neural network to obtain the environmental context information of the historical behavior sequences output by the LSTM neural network and to determine the historical risk trend information of the historical behavior sequences. Finally, the weighted risk score is determined based on the weights and risk scores corresponding to the multiple risk categories, the environmental context information of the historical behavior sequences, and the historical risk trend information. The cross-domain adaptive permission reconstruction and control module is used to input the weighted risk score into the second model to obtain the user permission information output by the second model. The second model determines the user permission information based on the weighted risk score, the effectiveness of permission adjustment, and the consistency of user behavior. The presentation module is used to present the user permission information.

18. An electronic device, characterized in that, It includes one or more processors; one or more memories; said one or more memories storing one or more computer programs, said one or more computer programs including instructions that, when executed by said one or more processors, cause the method of any one of claims 1 to 16 to be performed.

19. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed on a computer, cause the method as described in any one of claims 1 to 16 to be performed.

20. A chip, characterized in that, The chip includes a processor and a communication interface, the communication interface being used to receive signals and transmit the signals to the processor, the processor processing the signals such that the method as described in any one of claims 1 to 16 is executed.

21. A computer program product, characterized in that, When the computer program product is run on a computer, the computer performs the method as described in any one of claims 1 to 16.

Images