Data access control method and system based on artificial intelligence

Abstract

The invention relates to the technical field of machine learning, in particular to a data access control method and system based on artificial intelligence, and the method comprises the steps: S1, a preprocessing module obtains a plurality of historical access requests in advance, generates historical representative data corresponding to each historical access request, and sends the historical representative data to a server; new historical representative data also continues to be generated based on the generated historical representative data. And S2, the practice module performs practice in sequence based on all the historical representative data to generate a plurality of control models, and the control module performs sequential storage on the plurality of control models generated by the practice module according to a practice generation sequence. And S3, the control module receives the real-time access request from the user module, generates real-time representative data corresponding to the real-time access request, and performs control processing on the real-time representative data by using a plurality of sequentially stored control models. The intelligence of the access control method can be improved.

Description

Technical Field

[0001] The present application relates to the field of machine learning technology, and in particular to a data access control method and system based on artificial intelligence. Background Art

[0002] With the acceleration of digital transformation, data has become the core asset of enterprises. However, traditional access control methods are difficult to adapt to complex access scenarios. A new access control method is needed to better protect data security.

[0003] Chinese patent application publication number CN105229662A discloses an access control device, method, and program. These devices extract hierarchical elements of the hierarchy level that are paired with the identified hierarchical elements from information in an organizational information management unit, and extract hierarchical elements of the hierarchy level that are paired with the extracted hierarchical elements. The devices then compare the identified and extracted hierarchical elements with hierarchical elements defined by a task allocation management unit to determine whether access is permitted. Furthermore, Chinese patent application publication number CN113505090A discloses an access control method and device. The method comprises: in response to receiving a request from a first application to access a universal integrated circuit card (UICC), reading an access control condition file and an access control extension file in the UICC; the access control extension file is used to indicate at least whether an application authorized to access the UICC is permitted to provide access control conditions; determining a first access control condition based on the access control condition file and the access control extension file; and processing the first application's request to access the UICC based on the first access control condition. However, the access control methods disclosed in these two patent applications lack intelligence. Summary of the Invention

[0004] This application provides a preprocessing module to generate historical representative data corresponding to historical access requests, generates new historical representative data based on the historical representative data, provides an exercise module to sequentially exercise and generate several control models based on all the historical representative data, provides a control module to generate real-time representative data corresponding to real-time access requests, and uses several control models to perform control processing on the real-time representative data. This application aims to improve the intelligence of the access control method.

[0005] This application provides an artificial intelligence-based data access control method, including the following steps:

[0006] S1. A preprocessing module obtains several historical access requests in advance, and for each historical access request, the preprocessing module generates historical representative data corresponding to the historical access request, and further generates new historical representative data based on the generated historical representative data;

[0007] S2. The practice module sequentially generates a plurality of control models based on all historical representative data, and the control module sequentially stores the plurality of control models generated by the practice module in the order of practice generation;

[0008] S3. The control module receives a real-time access request from the user module, generates real-time representative data corresponding to the real-time access request, and performs control processing on the real-time representative data using a plurality of control models stored sequentially.

[0009] As a preferred technical solution of the present application, the training module sequentially trains and generates several control models based on all historical representative data, including the following steps:

[0010] S21, the practice module sets corresponding identification data for each historical representative data, and uses each historical representative data and the identification data corresponding to each historical representative data to form each practice data group;

[0011] S22, the training module uses all the training data sets to train and generate a control model, and for each training data set, the training module uses the historical representative data in the training data set as input data of the control model generated by the training to obtain output data;

[0012] S23. The practice module counts the total number of output data that are identical to the identification data in the corresponding practice data group, calculates the ratio of the total number to the total number of output data, and determines whether the ratio is greater than a preset ratio threshold. If yes, all steps are terminated; if not, change processing is performed on all practice data groups, and the process jumps to S22.

[0013] As a preferred technical solution of the present application, the practice module performs change processing on all practice data groups, including the following steps:

[0014] S231, the training module determines a training data group from all training data groups, and determines whether the output data corresponding to the determined training data group is the same as the identification data in the determined training data group. If they are not the same, the output data corresponding to the determined training data group is added to the historical representative data in the determined training data group. If they are the same, the process proceeds to the next step.

[0015] S232, the practice module generates real-time data and determines whether the real-time data is greater than a preset control data. If so, the output data corresponding to the determined practice data group is added to the historical representative data in the determined practice data group. If not, the module proceeds to the next step.

[0016] S233: The training module scrambles the output data corresponding to the determined training data group to obtain scrambled output data, adds the scrambled output data to the historical representative data in the determined training data group, and determines whether there is an undetermined training data group. If yes, the process jumps to S231; otherwise, the process proceeds to the next step.

[0017] S234, the practice module determines whether the end condition is met. If not, the control data is modified and the process jumps to S231. If the end condition is met, all steps are ended.

[0018] As a preferred technical solution of the present application, the exercise module determines whether the end condition is met, including: the exercise module counts the first number of exercise data groups that have been changed due to the execution of S231, and the second number of exercise data groups that have been changed due to the execution of S232, and determines whether the difference between the first number and the second number is less than a preset difference threshold. If so, it is determined that the end condition is met; if not, it is determined that the end condition is not met.

[0019] As an optimal technical solution of the present application, the practice module performs modification processing on the control data, including: when the second number is greater than the first number, the practice module increases the control data; when the second number is less than the first number, the practice module decreases the control data.

[0020] As a preferred technical solution of the present application, the control module uses a plurality of control models stored sequentially to perform control processing on real-time representative data, including the following steps:

[0021] S31, the control module determines the first control model stored in sequence;

[0022] S32, the control module uses the real-time representative data as input data of the determined control model to obtain output data, and adds the output data to the real-time representative data;

[0023] S33, the control module determines whether there is an undetermined control model. If so, it determines the next control model according to the storage order and jumps to S32. If not, it determines the final output data based on the output data of all control models and determines whether the real-time access request is allowed based on the final output data.

[0024] As a preferred technical solution of the present application, the pre-processing module further generates new historical representative data based on the generated historical representative data, including the following steps:

[0025] S11, the preprocessing module performs PCA transformation processing on the generated historical representative data to obtain a plurality of characteristic values, and for each characteristic value, the preprocessing module sets a weight value corresponding to the characteristic value, and calculates the product of the characteristic value and the weight value corresponding to the characteristic value;

[0026] S12, the preprocessing module calculates the sum of all products to obtain a basic value, and the preprocessing module generates a number of random values based on a Gaussian distribution, and the preprocessing module further sets a ratio value;

[0027] S13. For each random value, the preprocessing module calculates the sum of the basic value and the product of the random value and the proportional value to obtain intermediate data, and the preprocessing module performs PCA inverse transformation on the intermediate data to obtain new historical representative data.

[0028] This application also provides an artificial intelligence-based data access control system, including the following modules:

[0029] a pre-processing module, configured to obtain a number of historical access requests in advance, and generate historical representative data corresponding to each historical access request, and further generate new historical representative data based on the generated historical representative data;

[0030] A practice module, used to practice and generate several control models in sequence based on all historical representative data;

[0031] The user module is used to collect real-time access requests from users and send the real-time access requests to the control module;

[0032] The control module is used to sequentially store several control models generated by the exercise module in the order of exercise generation, and is used to receive real-time access requests from the user module, generate real-time representative data corresponding to the real-time access requests, and use the several control models stored in sequence to perform control processing on the real-time representative data.

[0033] Compared with the prior art, the beneficial effects of this application are at least as follows:

[0034] In the technical solution provided by the present application, first, the pre-processing module obtains several historical access requests in advance, and for each historical access request, the pre-processing module generates historical representative data corresponding to the historical access request, and continues to generate new historical representative data based on the generated historical representative data. Secondly, the practice module sequentially generates several control models based on all the historical representative data, and the control module sequentially stores the several control models generated by the practice module in the order of practice generation. Finally, the control module receives real-time access requests from the user module, generates real-time representative data corresponding to the real-time access requests, and the control module uses the several control models stored sequentially to perform control processing on the real-time representative data. Through the present application, the intelligence of the access control method can be improved, so that the access control method can adapt to complex access scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 This is a flowchart of the data access control method based on artificial intelligence in an embodiment of the present application;

[0037] Figure 2 Generate several flow charts of control models in the embodiments of this application;

[0038] Figure 3 This is a flow chart of controlling and processing real-time representative data in an embodiment of the present application;

[0039] Figure 4 Schematic diagram of an artificial intelligence-based data access control system in an embodiment of the present application. DETAILED DESCRIPTION

[0040] The embodiments of the present application provide a data access control method and system based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0041] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 The data access control method based on artificial intelligence in the embodiment of the present application includes the following main steps:

[0042] S1. The preprocessing module obtains several historical access requests in advance, and for each historical access request, the preprocessing module generates historical representative data corresponding to the historical access request, and further generates new historical representative data based on the generated historical representative data;

[0043] S2. The practice module sequentially generates a plurality of control models based on all historical representative data, and the control module sequentially stores the plurality of control models generated by the practice module in the order of practice generation;

[0044] S3. The control module receives a real-time access request from the user module, generates real-time representative data corresponding to the real-time access request, and performs control processing on the real-time representative data using a plurality of control models stored sequentially.

[0045] Specifically, in order to provide an intelligent access control method that can adapt to complex access scenarios, in S1, the pre-processing module collects several historical access requests in advance. The historical access requests may include information such as user permissions, the sensitivity of the accessed data, and the access frequency. The various information included in the historical access requests varies according to the access scenarios. For each historical access request, the pre-processing module generates historical representative data corresponding to the historical access request. The historical representative data can be in the form of a vector, and each value in the vector corresponds to a type of information in the historical access request. For each historical access request, the pre-processing module continues to generate new historical representative data based on the generated historical representative data. This is done to ensure that the number of all historical representative data generated is sufficient, so that the practice module can generate an accurate control model. In S2, the practice module sequentially practices and generates several control models based on all the historical representative data, and sends the generated several control models to the control module. The control module sequentially stores the several control models generated by the practice module in the order in which they were practiced. In S3, the control module receives a real-time access request from the user module and generates real-time representative data corresponding to the real-time access request. The type of information included in the real-time access request is the same as the type of information included in the historical access request. The method of generating the real-time representative data is the same as the method of generating the historical representative data. The control module uses several control models stored sequentially to perform control processing on the real-time representative data and ultimately decides whether to allow the real-time access request.

[0046] Furthermore, the training module sequentially trains and generates several control models based on all historical representative data, including the following steps:

[0047] S21, the practice module sets corresponding identification data for each historical representative data, and uses each historical representative data and the identification data corresponding to each historical representative data to form each practice data group;

[0048] S22, the training module uses all the training data sets to train and generate a control model, and for each training data set, the training module uses the historical representative data in the training data set as input data of the control model generated by the training to obtain output data;

[0049] S23. The practice module counts the total number of output data that are identical to the identification data in the corresponding practice data group, calculates the ratio of the total number to the total number of output data, and determines whether the ratio is greater than a preset ratio threshold. If yes, all steps are terminated; if not, change processing is performed on all practice data groups and jumps to S22.

[0050] Specifically, refer to Figure 2As shown, it is described how the practice module sequentially practices and generates several control models based on all historical representative data. In S21, the practice module sets corresponding identification data for each historical representative data, and uses each historical representative data and the identification data corresponding to each historical representative data to form each practice data group. That is, each practice data group is composed of historical representative data and the identification data corresponding to the historical representative data. The identification data represents the risk level of the historical access request corresponding to the historical representative data. The risk level is an integer greater than zero, and the larger the integer, the higher the risk level. In S22, the practice module uses all the practice data groups to practice and generate a control model. The control model can be a support vector machine model in the prior art. After practicing and generating the control model, for each practice data group, the practice module uses the historical representative data in the practice data group as the input data of the control model generated by the practice and obtains the output data. In S23, the practice module counts the total number of output data that are identical to the identification data in the corresponding practice data group. The practice data group corresponding to the output data refers to the practice data group that is input into the control model generated by the practice to obtain the output data. The practice module calculates the ratio of the total number to the total number of output data, and determines whether the ratio is greater than a preset ratio threshold. The ratio threshold is set according to the actual application scenario. If it is greater than, no new control model will be generated by practice, and all steps will be terminated. If it is not greater than, change processing will be performed on all practice data groups, and jump to S22 to continue execution to practice generating a new control model.

[0051] Furthermore, the practice module performs change processing on all practice data groups, including the following steps:

[0052] S231, the training module determines a training data group from all training data groups, and determines whether the output data corresponding to the determined training data group is the same as the identification data in the determined training data group. If they are not the same, the output data corresponding to the determined training data group is added to the historical representative data in the determined training data group. If they are the same, the module proceeds to the next step.

[0053] S232: The practice module generates real-time data and determines whether the real-time data is greater than a preset control data. If so, the output data corresponding to the determined practice data group is added to the historical representative data in the determined practice data group. If not, the module proceeds to the next step.

[0054] S233: The training module scrambles the output data corresponding to the determined training data group to obtain scrambled output data, adds the scrambled output data to the historical representative data in the determined training data group, and determines whether there is an undetermined training data group. If yes, the process jumps to S231; otherwise, the process continues to the next step.

[0055] S234. The practice module determines whether the end condition is met. If not, the control data is modified and the process jumps to S231. If the end condition is met, all steps are ended.

[0056] Specifically, the process of the practice module performing change processing on all practice data groups is described. In step S231, the practice module determines a practice data group from all practice data groups and determines whether the output data corresponding to the determined practice data group is identical to the identification data in the determined practice data group. If not, the output data corresponding to the determined practice data group is added to the historical representative data in the determined practice data group. If they are identical, the process continues with step S232. In step S232, the practice module generates real-time data, which is an arbitrary value between zero and one, and determines whether the real-time data is greater than a preset reference data, the initial value of which is also between zero and one. If so, the output data corresponding to the determined practice data group is added to the historical representative data in the determined practice data group. If the historical representative data is in the form of a vector, the vector is incremented by one value. If not, the process continues with step S233. In S233, the training module performs scrambling processing on the output data corresponding to the determined training data group, thereby obtaining scrambled output data. The scrambling processing includes: generating an arbitrary value between -1 and +1, adding the output data corresponding to the determined training data group and the generated arbitrary value to obtain the scrambled output data, adding the scrambled output data to the historical representative data in the determined training data group, and then determining whether there is an undetermined training data group. If so, the module jumps to S231, continues to determine an undetermined training data group and executes subsequent steps. If not, the module continues to S234. In S234, the training module determines whether the termination condition is met. If not, the module performs modification processing on the control data and jumps to S231 to continue execution. It should be noted that when continuing execution, all training data groups are restored to the state before the modification processing. If the condition is met, all steps are terminated.

[0057] Through the above method, the training module uses a training data set with more features each time it generates a new control model, which can provide additional information for the new control model and help improve the accuracy of the new control model.

[0058] Furthermore, the exercise module determines whether the termination condition is met, including: the exercise module counts a first number of exercise data groups that have been changed due to the execution of S231, and a second number of exercise data groups that have been changed due to the execution of S232, and determines whether the difference between the first number and the second number is less than a preset difference threshold. If so, it is determined that the termination condition is met; if not, it is determined that the termination condition is not met.

[0059] Specifically, the process of the practice module judging whether the end condition is met is introduced. The practice module counts the first number of the practice data groups that have been changed due to the execution of S231, and counts the second number of the practice data groups that have been changed due to the execution of S232, and judges whether the gap between the first number and the second number is less than a preset gap threshold. The gap is the absolute value of the difference between the first number and the second number. If it is less than, the accuracy of the new control model generated by practicing with all the practice data groups that have undergone the change processing will be better, and it is judged that the end condition is met. If it is not less than, it is judged that the end condition is not met.

[0060] Furthermore, the practice module performs modification processing on the control data including: when the second number is greater than the first number, the practice module increases the control data; when the second number is less than the first number, the practice module decreases the control data.

[0061] Specifically, it introduces how the exercise module performs changes to the control data. If the second number counted is greater than the first number counted, a method should be adopted to reduce the second number counted next time, that is, adaptively increase the control data. If the second number is less than the first number, a method should be adopted to increase the second number counted next time, that is, adaptively reduce the control data.

[0062] The practice module executes the above S234 in order to find suitable control data. In addition, it can also execute the above S231 to S233 after each adaptive modification of the control data, and use the new control model generated by practicing all the practice data groups that have undergone the change processing, and then use the control data corresponding to the control model with the highest accuracy as the suitable control data.

[0063] Furthermore, the control module uses a plurality of sequentially stored control models to perform control processing on the real-time representative data, including the following steps:

[0064] S31, the control module determines the first control model stored in sequence;

[0065] S32, the control module uses the real-time representative data as input data of the determined control model to obtain output data, and adds the output data to the real-time representative data;

[0066] S33. The control module determines whether there is an undetermined control model. If so, it determines the next control model according to the storage order and jumps to S32. If not, it determines the final output data based on the output data of all control models and determines whether the real-time access request is allowed based on the final output data.

[0067] Specifically, refer to Figure 3 As shown, the control module uses several sequentially stored control models to control and process real-time representative data. In S31, the control module determines the first sequentially stored control model. In S32, the control module uses the real-time representative data as input data for the determined control model to obtain output data, and then adds the output data to the real-time representative data. In S33, the control module determines whether there is an undetermined control model. If so, it determines the next control model according to the storage order and jumps to S32 to continue execution. If not, the control module determines the final output data based on the output data of all control models. The output data that appears the most times in the output data of all control models can be regarded as the final output data. Then, based on the final output data, it determines whether the real-time access request is allowed. If the risk level indicated by the final output data is high, the real-time access request can be rejected.

[0068] Furthermore, the pre-processing module further generates new historical representative data based on the generated historical representative data, including the following steps:

[0069] S11, the preprocessing module performs PCA transformation processing on the generated historical representative data to obtain a number of characteristic values, and for each characteristic value, the preprocessing module sets a weight value corresponding to the characteristic value, and calculates the product of the characteristic value and the weight value corresponding to the characteristic value;

[0070] S12, the preprocessing module calculates the sum of all products to obtain a basic value, and the preprocessing module generates a number of random values based on a Gaussian distribution, and the preprocessing module also sets a ratio value;

[0071] S13. For each random value, the preprocessing module calculates the sum of the basic value and the product of the random value and the proportional value to obtain intermediate data, and the preprocessing module performs PCA inverse transformation processing on the intermediate data to obtain new historical representative data.

[0072] Specifically, the process of the preprocessing module continuing to generate new historical representative data based on the generated historical representative data is introduced. In S11, the preprocessing module performs PCA transformation processing on the generated historical representative data, with the purpose of obtaining a number of characteristic values. The number of characteristic values represents the main characteristic directions of the historical representative data. For each characteristic value, the preprocessing module sets a weight value corresponding to the characteristic value, and calculates the product of the characteristic value and the weight value corresponding to the characteristic value. The same weight value can be set for each characteristic value, and of course, different weight values can also be set for each characteristic value. In S12, the preprocessing module calculates the sum of all products to obtain a basic value. The preprocessing module generates a number of random values based on a Gaussian distribution. The number of random values is the same as the number of characteristic values. Specifically, each random value is randomly generated from a Gaussian distribution with a mean of 0 and a standard deviation of 0.1. The preprocessing module also sets a ratio value, which can be 0.5. In S13, for each random value, the preprocessing module calculates the sum of the base value and the product of the random value and the scale value, thereby obtaining intermediate data. The intermediate data is also in the form of a vector. The preprocessing module also performs an inverse PCA transform on the intermediate data to ultimately obtain new historical representative data. Through the above method, new historical representative data that is relatively similar to the historical representative data can be generated.

[0073] According to another aspect of the embodiment of the present application, Figure 4 As shown, the present application also provides an artificial intelligence-based data access control system, including a preprocessing module, a practice module, a user module, and a control module to implement the artificial intelligence-based data access control method described above.

[0074] The functions of each module are as follows:

[0075] a pre-processing module, configured to obtain a number of historical access requests in advance, and generate historical representative data corresponding to each historical access request, and further generate new historical representative data based on the generated historical representative data;

[0076] A practice module, used to practice and generate several control models in sequence based on all historical representative data;

[0077] The user module is used to collect real-time access requests from users and send the real-time access requests to the control module;

[0078] The control module is used to sequentially store several control models generated by the exercise module in the order of exercise generation, and is used to receive real-time access requests from the user module, generate real-time representative data corresponding to the real-time access requests, and use the several control models stored in sequence to perform control processing on the real-time representative data.

[0079] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0080] If the integrated unit is implemented in the form of 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 the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0081] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The data access control method based on artificial intelligence is characterized by: The method comprises the following steps: S1. A preprocessing module obtains several historical access requests in advance, and for each historical access request, the preprocessing module generates historical representative data corresponding to the historical access request, and further generates new historical representative data based on the generated historical representative data; S2. The practice module sequentially generates a plurality of control models based on all historical representative data, and the control module sequentially stores the plurality of control models generated by the practice module in the order of practice generation; S3. The control module receives a real-time access request from the user module, generates real-time representative data corresponding to the real-time access request, and performs control processing on the real-time representative data using a plurality of control models stored sequentially.

2. The method according to claim 1, characterized in that The training module generates several control models based on all historical representative data, including the following steps: S21, the practice module sets corresponding identification data for each historical representative data, and uses each historical representative data and the identification data corresponding to each historical representative data to form each practice data group; S22, the training module uses all the training data sets to train and generate a control model, and for each training data set, the training module uses the historical representative data in the training data set as input data of the control model generated by the training to obtain output data; S23. The practice module counts the total number of output data that are identical to the identification data in the corresponding practice data group, calculates the ratio of the total number to the total number of output data, and determines whether the ratio is greater than a preset ratio threshold. If yes, all steps are terminated; if not, change processing is performed on all practice data groups, and the process jumps to S22.

3. The method according to claim 2, characterized in that The exercise module performs change processing on all exercise data groups, including the following steps: S231, the training module determines a training data group from all training data groups, and determines whether the output data corresponding to the determined training data group is the same as the identification data in the determined training data group. If they are not the same, the output data corresponding to the determined training data group is added to the historical representative data in the determined training data group. If they are the same, the process proceeds to the next step. S232, the practice module generates real-time data and determines whether the real-time data is greater than a preset control data. If so, the output data corresponding to the determined practice data group is added to the historical representative data in the determined practice data group. If not, the module proceeds to the next step. S233: The training module scrambles the output data corresponding to the determined training data group to obtain scrambled output data, adds the scrambled output data to the historical representative data in the determined training data group, and determines whether there is an undetermined training data group. If yes, the process jumps to S231; otherwise, the process proceeds to the next step. S234, the practice module determines whether the end condition is met. If not, the control data is modified and the process jumps to S231. If the end condition is met, all steps are ended.

4. The method according to claim 3, characterized in that The exercise module determines whether the termination condition is met, including: the exercise module counts the first number of exercise data groups that have been changed due to the execution of S231, and the second number of exercise data groups that have been changed due to the execution of S232, and determines whether the difference between the first number and the second number is less than a preset difference threshold. If so, it is determined that the termination condition is met; if not, it is determined that the termination condition is not met.

5. The method according to claim 4, characterized in that The practice module performs modification processing on the control data, including: when the second number is greater than the first number, the practice module increases the control data; when the second number is less than the first number, the practice module decreases the control data.

6. The method according to claim 1, characterized in that The control module uses a plurality of control models stored sequentially to perform control processing on real-time representative data, including the following steps: S31, the control module determines the first control model stored in sequence; S32, the control module uses the real-time representative data as input data of the determined control model to obtain output data, and adds the output data to the real-time representative data; S33, the control module determines whether there is an undetermined control model. If so, it determines the next control model according to the storage order and jumps to S32. If not, it determines the final output data based on the output data of all control models and determines whether the real-time access request is allowed based on the final output data.

7. The method according to claim 1, characterized in that The pre-processing module further generates new historical representative data based on the generated historical representative data, including the following steps: S11, the preprocessing module performs PCA transformation processing on the generated historical representative data to obtain a plurality of characteristic values, and for each characteristic value, the preprocessing module sets a weight value corresponding to the characteristic value, and calculates the product of the characteristic value and the weight value corresponding to the characteristic value; S12, the preprocessing module calculates the sum of all products to obtain a basic value, and the preprocessing module generates a number of random values based on a Gaussian distribution, and the preprocessing module further sets a ratio value; S13. For each random value, the preprocessing module calculates the sum of the basic value and the product of the random value and the proportional value to obtain intermediate data, and the preprocessing module performs PCA inverse transformation processing on the intermediate data to obtain new historical representative data.

8. An artificial intelligence-based data access control system for implementing the method according to any one of claims 1 to 7, characterized in that: Includes the following modules: a pre-processing module, configured to obtain a number of historical access requests in advance, and generate historical representative data corresponding to each historical access request, and further generate new historical representative data based on the generated historical representative data; A practice module, used to practice and generate several control models in sequence based on all historical representative data; The user module is used to collect real-time access requests from users and send the real-time access requests to the control module; The control module is used to sequentially store several control models generated by the exercise module in the order of exercise generation, and is used to receive real-time access requests from the user module, generate real-time representative data corresponding to the real-time access requests, and use the several control models stored in sequence to perform control processing on the real-time representative data.

Images