A digital product content protection system based on artificial intelligence

Abstract

The present invention relates to the field of digital content security and artificial intelligence technology, and discloses a digital product content protection system based on artificial intelligence, the system comprising a data collection module, a feature extraction module, an anomaly detection module, a risk assessment module and a decision execution module. Compared with the prior art, which mostly adopts a single rule verification for digital product recharge interface requests, especially under the conditions of frequent user operations or complex behaviors, it is impossible to effectively prevent the technical problem of malicious tampering of digital content through digital product recharge behaviors, this application realizes dynamic and accurate protection of digital product content by combining multi-dimensional behavior feature extraction and intelligent risk assessment, thereby avoiding the problems of inaccurate abnormal behavior judgment and delayed tampering detection in the prior art, and improving the reliability and real-time performance of digital product content protection.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital content security and artificial intelligence, and in particular relates to a digital product content protection system based on artificial intelligence. Background Art

[0002] At present, the content protection of digital products including mobile phone charges, traffic, online payment wallets, etc. mainly relies on static encryption or rule verification methods, but there are still significant deficiencies in dealing with complex recharge behavior scenarios and potential malicious operations. For example, the existing technology mostly uses a single rule to judge the legitimacy of the recharge interface, which cannot dynamically adapt to the diverse operation behaviors of users, and is prone to missed detection, misjudgment or delayed response when facing high-frequency recharge requests or abnormal recharge behaviors, such as continuous recharge or large recharge in a short period of time. In addition, in actual applications, especially in scenarios involving multiple parties such as operators and third-party recharge platforms, the recharge chain is complex, and attackers may tamper with the content of digital products by forging recharge interface requests, thereby damaging the content of digital products, such as account balances, virtual rights, user identity authentication information, etc.

[0003] Malicious recharge behavior is often accompanied by illegal tampering of digital product content, such as inflating account balances, forging virtual assets, or improper occupation. If abnormal recharge requests are not intercepted in time, the core content of digital products, such as virtual rights or digital records in user accounts, will be tampered with, causing direct economic losses to platform operations. Therefore, improving the accuracy and real-time nature of recharge interception is a key step in achieving digital product content protection.

[0004] Therefore, there is an urgent need for an AI-based digital product content protection system to achieve closed-loop management from recharge behavior detection to product content protection. While effectively intercepting abnormal recharge behavior, it also protects the integrity and security of digital product content in real time. Summary of the invention

[0005] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to propose a digital product content protection system based on artificial intelligence, aiming to solve the technical problem that the existing technology mostly uses a single rule verification for digital product recharge interface requests, especially under conditions of frequent user operations or complex behaviors, and cannot effectively prevent malicious tampering of digital content through digital product recharge behaviors.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a digital product content protection system based on artificial intelligence, comprising:

[0007] A data collection module is used to collect the user's digital product recharge behavior data, the recharge behavior data includes the recharge amount of the i-th recharge, the recharge time interval, and the recharge frequency in the preset time window;

[0008] The collected recharge behavior data is segmented according to the preset time window to obtain discrete data of the recharge behavior data, and the discrete data is converted into time series data. The time series data includes a recharge amount series, a recharge time interval series, and a recharge frequency series;

[0009] The feature extraction module is used to extract multi-dimensional behavior features from the obtained time series data, including:

[0010] The CNN convolutional neural network is used to extract the features of the recharge amount sequence and obtain the distribution pattern vector F of the recharge amount sequence. A ;

[0011] The recharge time interval sequence is feature extracted through the RNN recurrent neural network to obtain the change vector h of the recharge time interval sequence t ;

[0012] The statistical feature extraction method is used to extract the features of the recharge frequency sequence, and the recharge frequency feature vector F of the recharge frequency sequence is obtained. F ;

[0013] The FFT fast Fourier transform is used to analyze the recharge time interval sequence and obtain the periodic characteristic vector X k ;

[0014] Comprehensive distribution pattern vector F A , change vector h t , recharge frequency feature vector F F and the periodic eigenvector X k , and obtain the multi-dimensional behavior feature vector [F A ,h t , F F , X k ];

[0015] The anomaly detection module is used to use the SVM support vector machine to detect the multi-dimensional behavior feature vector [F A ,h t , F F , X k ] to perform anomaly detection, the anomaly detection formula is:

[0016] y=sgn(ω·x+b0)

[0017] Among them, y is the abnormal detection result value, 1 represents normal, -1 represents abnormal, and x is the multidimensional behavior feature vector [F A ,h t , FF , X k ], ω is the weight coefficient, b0 is the bias term, and sgn is the symbolic activation function;

[0018] The risk assessment module is used to perform risk scoring based on the detection results of the anomaly detection module:

[0019] R=α·P+β·y

[0020] Among them, R is the risk score, α and β are weight factors, and P is based on the periodic feature vector X in the feature extraction module. k Determine, represents the periodic feature score, y is the anomaly detection result value in the anomaly detection module;

[0021] The decision execution module is used to determine whether the current digital product recharge operation is passed according to the obtained risk score R, and pre-set a high risk threshold K1 and a low risk threshold K2, and K1 is greater than K2:

[0022] When the risk score is greater than or equal to the high risk threshold K1, the recharge operation is judged as a high risk behavior, the recharge operation is frozen and the user is notified;

[0023] When the risk score R is less than or equal to the low risk threshold K2, the recharge operation is completed directly;

[0024] When the risk score is less than the high-risk threshold K1 and greater than K2, the recharge operation is judged to be a medium-risk behavior, triggering multi-factor verification. The recharge operation is completed after successful verification, and the recharge operation is rejected after failed verification.

[0025] Preferably, in the feature extraction module, the distribution pattern vector F of the recharge amount sequence is obtained. A The extraction formula is:

[0026] F A =ReLU(W A ·W h +b)

[0027] Among them, F A is the distribution pattern vector, ReLU is an activation function, W A is the recharge amount sequence, W h is the convolution kernel, which is used to capture the local amount change characteristics, and b is the bias term.

[0028] Preferably, in the feature extraction module, a change vector h of the recharge time interval sequence is obtained. t The extraction formula is:

[0029] h t =tanh(x(n)·W Δ +W n ·h t-1+p)

[0030] Among them, h t is the change vector of the time interval, tanh is an activation function, x(n) is the recharge time interval sequence, W Δ is the time interval weight matrix, W n is the weight matrix of the hidden state, which is used to adjust the influence of the previous time step on the characteristics of the current time step, h t-1 is the hidden state of the previous time step, and p is the bias term.

[0031] Preferably, in the feature extraction module, the recharge frequency feature vector F F The mean μ of the recharge frequency sequence F and standard deviation σ F Composition, F F =[μ,σ F ].

[0032] Preferably, in the feature extraction module, the periodic feature vector X is obtained k The formula is:

[0033]

[0034] Where x(n) is the recharge time interval sequence, N is the length of the time interval sequence, and k represents the index of the discrete frequency. is a kernel function used to convert time domain data into frequency domain.

[0035] Preferably, in the risk assessment module, P is the periodic feature score, and according to the periodic feature vector X k OK, the associated formula is:

[0036]

[0037] Where P is the periodic feature score, |X k | is the modulus of the periodic eigenvector, ω k is the weight factor, k represents the index of the discrete frequency, and K is the maximum frequency index range.

[0038] Preferably, in the decision execution module, the multi-factor authentication includes a first factor account password authentication, a second factor dynamic verification code authentication, and a third factor biometric authentication.

[0039] The beneficial effect of the present invention is that compared with the prior art that mostly adopts a single rule verification for digital product recharge interface requests, especially under conditions where user operations are frequent or behaviors are complex, it is impossible to effectively prevent the technical problem of malicious tampering of digital content through digital product recharge behaviors. This application combines multi-dimensional behavior feature extraction with intelligent risk assessment to achieve dynamic and accurate protection of digital product content, thereby avoiding the problems of inaccurate abnormal behavior judgment and delayed tampering detection in the prior art, and improving the reliability and real-time performance of digital product content protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 This is a flow chart of a first embodiment of an artificial intelligence-based digital product content protection system provided by the present invention.

[0042] Figure 2 A schematic diagram of a device of an artificial intelligence-based digital product content protection system provided by the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Embodiment 1: Figure 1 As shown, it is a flow chart of the first embodiment of the digital product content protection system based on artificial intelligence of the present invention, and the first embodiment of the digital product content protection system based on artificial intelligence of the present invention is proposed.

[0045] In a first embodiment, the artificial intelligence-based digital product content protection system includes:

[0046] A data collection module is used to collect the user's digital product recharge behavior data, the recharge behavior data includes the recharge amount of the i-th recharge, the recharge time interval, and the recharge frequency in the preset time window;

[0047] The collected recharge behavior data is segmented according to the preset time window to obtain discrete data of the recharge behavior data, and the discrete data is converted into time series data. The time series data includes a recharge amount series, a recharge time interval series, and a recharge frequency series;

[0048] It should be noted that the data collection module automatically captures and records these data based on the user's recharge behavior for subsequent feature extraction and anomaly detection.

[0049] It is understandable that through complete data collection, the user's recharge behavior pattern can be accurately reflected, such as the size distribution of recharge amounts, the regularity of time intervals, and the density characteristics of recharge frequency. This information is the basis for detecting abnormal recharge behavior.

[0050] It should be understood that in order to improve data processing efficiency, the system will segment the collected recharge behavior data according to preset time windows. The data in each time window will be classified into a discrete data set to form a behavior data record in a continuous time period.

[0051] For example, assuming that the system's time window is 1 hour, a user recharges three times within an hour, 100 yuan, 200 yuan, and 300 yuan respectively, and the corresponding time intervals are 1800 seconds and 3600 seconds, then:

[0052] The recharge amount sequence is {100,200,300};

[0053] The recharge time interval sequence is {1800,3600};

[0054] The recharge frequency is 3 (times / hour).

[0055] The feature extraction module is used to extract multi-dimensional behavior features from the obtained time series data, including:

[0056] The CNN convolutional neural network is used to extract the features of the recharge amount sequence and obtain the distribution pattern vector F of the recharge amount sequence. A ;

[0057] The recharge time interval sequence is feature extracted through the RNN recurrent neural network to obtain the change vector h of the recharge time interval sequence t ;

[0058] The statistical feature extraction method is used to extract the features of the recharge frequency sequence, and the recharge frequency feature vector F of the recharge frequency sequence is obtained. F ;

[0059] The FFT fast Fourier transform is used to analyze the recharge time interval sequence and obtain the periodic characteristic vector X k ;

[0060] Comprehensive distribution pattern vector F A , change vector h t , recharge frequency feature vector F F and the periodic eigenvector X k , and obtain the multi-dimensional behavior feature vector [F A ,h t , F F , X k ];

[0061] It should be noted that the effect of extracting features from the recharge amount sequence through the CNN convolutional neural network is to extract local pattern characteristics in the user recharge amount sequence, such as the behavior pattern of continuous high-amount recharge or small-amount high-frequency recharge; the effect of extracting features from the recharge time interval sequence through the RNN recurrent neural network is to extract the dynamic change rules in the recharge time interval sequence, such as continuously shortened time intervals or fixed time interval recharge behavior; the effect of extracting features from the recharge frequency sequence using the statistical feature extraction method is to extract the global distribution characteristics of the recharge frequency, such as the average behavior and fluctuation range of high-frequency recharge; the effect of using the FFT fast Fourier transform to analyze the recharge time interval sequence to obtain a periodic feature vector is to extract periodic features in the time interval, such as fixed time interval recharge behavior.

[0062] It should be understood that by integrating multi-dimensional features, the system can comprehensively describe the characteristics of users' recharge behavior, thereby improving the accuracy and robustness of anomaly detection. The comprehensive features not only consider single dimensions such as recharge amount, time interval and frequency, but also supplement the global perspective of behavior through periodic characteristics, which is suitable for behavioral analysis in complex scenarios.

[0063] The anomaly detection module is used to use the SVM support vector machine to detect the multi-dimensional behavior feature vector [F A ,h t , F F , X k ] to perform anomaly detection, the anomaly detection formula is:

[0064] y=sgn(ω·x+b0)

[0065] Among them, y is the abnormal detection result value, 1 represents normal, -1 represents abnormal, and x is the multidimensional behavior feature vector [F A ,h t , F F , X k ], ω is the weight coefficient, b0 is the bias term, and sgn is the symbolic activation function;

[0066] It should be noted that the weight coefficient ω represents the importance of each feature dimension in the support vector machine model. The weight is determined through model training. The higher the weight value, the greater the contribution of the feature in classification. b0 is a bias term used to adjust the position of the classification boundary to ensure the adaptability of the classifier to positive and negative samples.

[0067] It can be understood that the support vector machine constructs an optimal classification hyperplane by maximizing the classification interval, so that the sample points of different categories are as far away from the classification boundary as possible.

[0068] It should be understood that the input vector contains multi-dimensional features of the user's recharge behavior, and each feature has a synergistic effect. For example, if the recharge amount is abnormally high, but the time interval and frequency are normal, it may be judged as non-abnormal behavior; if the periodic feature shows significant periodicity, it may be identified as automated behavior.

[0069] The risk assessment module is used to perform risk scoring based on the detection results of the anomaly detection module:

[0070] R=α·P+β·y

[0071] Among them, R is the risk score, α and β are weight factors, and P is based on the periodic feature vector X in the feature extraction module. k Determine, represents the periodic feature score, y is the anomaly detection result value in the anomaly detection module;

[0072] It should be understood that, unlike traditional single rule verification, this module can dynamically adjust the risk score based on real-time recharge behavior to adapt to frequent or abnormal recharge scenarios. For high-frequency recharge behaviors or sudden large-amount recharge behaviors in a short period of time, the module can quickly give a risk score to prevent the risk of content tampering caused by delayed response.

[0073] It should be understood that the periodic feature scoring extracts fixed patterns in user behavior based on the periodic feature vector, and the anomaly detection result extracts the abnormality of the behavior. The combination of the two realizes a dual evaluation from regularity to suddenness:

[0074] Periodic features improve the ability to identify automated recharge behaviors, such as fixed time intervals or significant periodic patterns;

[0075] The abnormal detection results supplement the accurate judgment of sudden high-risk behaviors.

[0076] For example, in complex scenarios involving operators or third-party recharge platforms, this module can combine the multi-dimensional features of recharge behavior to improve the accuracy of abnormal behavior identification.

[0077] The decision execution module is used to determine whether the current digital product recharge operation is passed according to the obtained risk score R, and pre-set a high risk threshold K1 and a low risk threshold K2, and K1 is greater than K2:

[0078] When the risk score is greater than or equal to the high risk threshold K1, the recharge operation is judged as a high risk behavior, the recharge operation is frozen and the user is notified;

[0079] When the risk score R is less than or equal to the low risk threshold K2, the recharge operation is completed directly;

[0080] When the risk score is less than the high-risk threshold K1 and greater than K2, the recharge operation is judged to be a medium-risk behavior, triggering multi-factor verification. The recharge operation is completed after successful verification, and the recharge operation is rejected after failed verification.

[0081] It should be noted that the K1 high-risk threshold indicates a threshold with a greater degree of behavioral abnormality and a higher risk; the K2 low-risk threshold indicates a threshold with a lower degree of behavioral abnormality and a lower risk; the settings of thresholds K1 and K2 can be dynamically adjusted according to specific scenarios. For example, the tolerance range for recharge behavior during peak holiday periods may be looser, while the threshold may be stricter in sensitive accounts.

[0082] It is understandable that the freezing operation is the ultimate protection measure for high-risk recharge behaviors, preventing abnormal operations from causing irreversible damage to the content of digital products, and at the same time strengthening the security control of accounts by notifying users.

[0083] It should be understood that multi-factor authentication can dynamically adjust the authentication strength according to the risk level. For example, ordinary medium-risk behaviors only require passwords and dynamic verification codes; when high-value top-ups or sensitive accounts are involved, biometric authentication such as fingerprint or facial recognition can be added to improve the security and applicability of the decision-making execution module.

[0084] For example, scenario 1: high-risk top-up behavior:

[0085] Input: A user's risk score R = 0.9, high risk threshold K1 = 0.8;

[0086] Processing: Since R≥K1, the system determines it as a high-risk behavior;

[0087] Freeze the recharge operation and notify the user that "your account has abnormal recharge behavior, please check the account activity or contact customer service";

[0088] Scenario 2: Medium-risk deposit behavior

[0089] Input: A user's risk score R = 0.5, high risk threshold K1 = 0.8, low risk threshold K2 = 0.3;

[0090] Processing: Since K2 < R < K1, the system determines it as a medium-risk behavior;

[0091] Trigger multi-factor authentication, including:

[0092] Step 1: Verify the user account password;

[0093] Step 2: Send a dynamic verification code (OTP) to the user's mobile phone and require the correct verification code to be entered;

[0094] Step 3: Require the user to complete the verification through fingerprint or facial recognition.

[0095] If the verification is successful, complete the recharge operation;

[0096] If the verification fails, reject the recharge operation and display "Recharge failed, abnormal behavior recorded" to the user.

[0097] In addition, the present invention also provides a digital product content protection device based on artificial intelligence. Please refer to Figure 2, a digital product content protection device based on artificial intelligence includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by at least one processor, and the instructions are executed by at least one processor so that at least one processor can execute the digital product content protection system based on artificial intelligence in the above-mentioned embodiment 1. A digital product content protection device based on artificial intelligence in the embodiment of the present invention may include but is not limited to mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. A digital product content protection device based on artificial intelligence is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. A digital product content protection device based on artificial intelligence may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of a digital product content protection device based on artificial intelligence are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow an artificial intelligence-based digital product content protection device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows an artificial intelligence-based digital product content protection device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have alternatively.

[0098] The present invention also provides a computer program product, including a computer program, which implements the steps of a digital product content protection system based on artificial intelligence when executed by a processor. The computer program product provided by the present invention can solve the technical problem of digital product content protection based on artificial intelligence. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as the beneficial effects of a digital product content protection system based on artificial intelligence provided by the above embodiment, which will not be repeated here.

[0099] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the system of the embodiments disclosed in the present invention are executed.

[0100] It should be understood that the various parts disclosed in the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0101] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A digital product content protection system based on artificial intelligence, characterized in that: The system includes: A data collection module is used to collect the user's digital product recharge behavior data, the recharge behavior data includes the recharge amount of the i-th recharge, the recharge time interval, and the recharge frequency in the preset time window; The collected recharge behavior data is segmented according to a preset time window to obtain discrete data of the recharge behavior data, and the discrete data is converted into time series data. The time series data includes a recharge amount series, a recharge time interval series, and a recharge frequency series; The feature extraction module is used to extract multi-dimensional behavior features from the obtained time series data, including: The CNN convolutional neural network is used to extract the features of the recharge amount sequence and obtain the distribution pattern vector F of the recharge amount sequence. A ; The recharge time interval sequence is feature extracted through the RNN recurrent neural network to obtain the change vector h of the recharge time interval sequence t ; The statistical feature extraction method is used to extract the features of the recharge frequency sequence, and the recharge frequency feature vector F of the recharge frequency sequence is obtained. F ; The FFT fast Fourier transform is used to analyze the recharge time interval sequence and obtain the periodic feature vector X k ; Comprehensive distribution pattern vector F A , change vector h t , recharge frequency feature vector F F and the periodic eigenvector X k , and obtain the multi-dimensional behavior feature vector [F A ,h t , F F , X k ]; The anomaly detection module is used to use the SVM support vector machine to detect the multi-dimensional behavior feature vector [F A ,h t , F F , X k ] to perform anomaly detection, the anomaly detection formula is: y=sgn(ω·x+b0) Among them, y is the abnormal detection result value, 1 represents normal, -1 represents abnormal, and x is the multidimensional behavior feature vector [F A ,h t , F F , X k ], ω is the weight coefficient, b0 is the bias term, and sgn is the symbolic activation function; The risk assessment module is used to perform risk scoring based on the detection results of the anomaly detection module: R=α·P+β·y Among them, R is the risk score, α and β are weight factors, and P is based on the periodic feature vector X in the feature extraction module. k Determine, represents the periodic feature score, y is the anomaly detection result value in the anomaly detection module; The decision execution module is used to determine whether the current digital product recharge operation is passed according to the obtained risk score R, and pre-set a high risk threshold K1 and a low risk threshold K2, and K1 is greater than K2: When the risk score is greater than or equal to the high risk threshold K1, the recharge operation is judged as a high risk behavior, the recharge operation is frozen and the user is notified; When the risk score R is less than or equal to the low risk threshold K2, the recharge operation is completed directly; When the risk score is less than the high-risk threshold K1 and greater than K2, the recharge operation is judged to be a medium-risk behavior, triggering multi-factor verification. The recharge operation is completed after successful verification, and the recharge operation is rejected after failed verification.

2. The artificial intelligence-based digital product content protection system according to claim 1, characterized in that: In the feature extraction module, the distribution pattern vector F of the recharge amount sequence is obtained A The extraction formula is: F A =ReLU(W A ·IN h +b) Among them, F A is the distribution pattern vector, ReLU is an activation function, W A is the recharge amount sequence, W h is the convolution kernel, which is used to capture the local amount change characteristics, and b is the bias term.

3. The artificial intelligence-based digital product content protection system according to claim 1, characterized in that: In the feature extraction module, the change vector h of the recharge time interval sequence is obtained t The extraction formula is: h t =tanh(x(n)·W Δ +W n ·h t-1 +p) Among them, h t is the change vector of the time interval, tanh is an activation function, x(n) is the recharge time interval sequence, W Δ is the time interval weight matrix, W n is the weight matrix of the hidden state, which is used to adjust the influence of the previous time step on the characteristics of the current time step, h t-1 is the hidden state of the previous time step, and p is the bias term.

4. The digital product content protection system based on artificial intelligence as claimed in claim 1, characterized in that: In the feature extraction module, the recharge frequency feature vector F F The mean μ of the recharge frequency sequence F and standard deviation σ F Composition, F F =[μ,σ F ].

5. The digital product content protection system based on artificial intelligence as claimed in claim 1, characterized in that: In the feature extraction module, the periodic feature vector X is obtained k The formula is: Where x(n) is the recharge time interval sequence, N is the length of the time interval sequence, and k represents the index of the discrete frequency. is a kernel function used to convert time domain data into frequency domain.

6. The artificial intelligence-based digital product content protection system according to claim 1, characterized in that: In the risk assessment module, P is the periodic feature score, based on the periodic feature vector X k OK, the associated formula is: Where P is the periodic feature score, |X k | is the modulus of the periodic eigenvector, ω k is the weight factor, k represents the index of the discrete frequency, and K is the maximum frequency index range.

7. The artificial intelligence-based digital product content protection system according to claim 1, characterized in that: In the decision-making execution module, multi-factor authentication includes first-factor account password authentication, second-factor dynamic verification code authentication, and third-factor biometric authentication.

Images