An operation behavior verification method and system based on action sequence data

Abstract

The application discloses a kind of operation behavior verification method and system based on action sequence data, comprising: randomly generating target area for operation behavior verification, initializing the position of controllable object, and rendering the target area and controllable object to visual verification interface;In response to the action of behavior operator for the controllable object, the motion characteristics of the controllable object are simulated using a real physics engine to control the controllable object to move in the visual verification interface based on the motion characteristics;Action sequence data of the behavior operator is obtained and recorded in real time when the controllable object moves in the visual verification interface;When the controllable object moves to the target area, the extraction of multi-dimensional behavior characteristics is carried out based on the obtained action sequence data;The multi-dimensional behavior characteristics are analyzed to determine whether the behavior operator is a real user.

Description

Technical Field

[0001] This invention relates to the fields of network security and human-machine verification technology, and more specifically, to a method and system for verifying operational behavior based on action sequence data. Background Technology

[0002] CAPTCHA technology, as a core means of human-machine identity verification in cyberspace, is widely used in critical scenarios such as account login, form submission, and transaction protection. Its core function is to distinguish between real users and automated programs (bots), preventing malicious attacks and data abuse. However, with the rapid development of artificial intelligence and deep learning technologies, existing CAPTCHA systems have gradually exposed many technical defects and security vulnerabilities, specifically as follows: 1. Static verification mechanisms are easily cracked, leading to a continuous degradation in security. Currently, most mainstream CAPTCHA technologies rely on static images or graphics, such as traditional character-based CAPTCHAs, slider CAPTCHAs, and point-and-click CAPTCHAs. These methods increase the difficulty of recognition by introducing image distortion and noise interference, but their defense capabilities are extremely limited against deep learning-driven OCR (Optical Character Recognition) technology and adversarial attacks. Robots can accurately recognize image content and extract features through model training, rendering the protective barrier of static CAPTCHAs ineffective. This puts user accounts at risk of brute-force attacks, seriously threatening the security of the network environment.

[0003] 2. The operational logic is simplistic and lacks in-depth analysis of behavioral characteristics. Existing CAPTCHAs generally have simple structures and limited dimensions, typically employing only two discrete operations: "click" or "drag and drop." During verification, they focus solely on the correctness of the final result, completely ignoring the user's continuous behavioral characteristics throughout the process. Robots can complete simulated operations with high efficiency and standardized procedures using pre-set scripts or paths, requiring no human thought or reaction. This lack of behavioral profiling in the verification logic makes CAPTCHAs unable to effectively distinguish between human users and automated scripts, resulting in persistently high false positive and false negative rates.

[0004] 3. Detached from physical interaction attributes, failing to establish core human-machine differences. One of the essential characteristics that distinguishes humans from machines lies in their ability to perceive and control the laws of the real physical world. However, most existing CAPTCHA designs remain at the level of purely numerical or purely graphical virtual interaction, failing to incorporate a physics engine or physical parameters. This design causes the verification process to remain at the "cognitive level" rather than the "tactile level," failing to effectively verify the operator's ability to simulate and cope with real physical phenomena such as inertia, resistance, and collision feedback, leaving an opportunity for machine simulation.

[0005] 4. The value of directional key interaction is overlooked, and multi-dimensional features are not activated. Arrow keys (up, down, left, and right) are one of the most basic and universal interactive input devices on computer interfaces, with extremely wide applications. However, in the field of CAPTCHAs, the application of arrow keys is currently almost non-existent, and their rich operational features have not been fully explored. Arrow key operations involve multi-dimensional continuous data such as the time interval between key presses, changes in pressure, frequency of direction switching, and inertial control. If this dynamic data, which truly reflects human operating habits and physiological characteristics, could be introduced into CAPTCHA systems, it would greatly enrich the judgment dimensions and create a behavioral barrier that is difficult for machines to simulate.

[0006] Existing CAPTCHA technologies have shortcomings in terms of security, interaction depth, and human-computer differentiation verification. Therefore, there is an urgent need for a method to verify operational behavior based on action sequence data. Summary of the Invention

[0007] This invention proposes a method and system for verifying operational behavior based on action sequence data, in order to solve the problem of how to efficiently perform human-machine verification based on operational behavior.

[0008] To address the above problems, according to one aspect of the present invention, a method for verifying operational behavior based on action sequence data is provided, the method comprising: Randomly generate a target area for verifying operational behavior, initialize the position of the controllable object, and render the target area and the controllable object onto the visual verification interface; In response to the actions of the operator on the controllable object, a real physics engine is used to simulate the motion characteristics of the controllable object, so as to control the movement of the controllable object in the visual verification interface based on the motion characteristics; Acquire and record in real time the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface; When the controllable object moves into the target area, multi-dimensional behavioral features are extracted based on the acquired action sequence data; The multi-dimensional behavioral characteristics are analyzed to determine whether the operator of the behavior is a real user.

[0009] Preferably, the method further includes: By receiving the operator's input of the four directional keys (up, down, left, right) on the visual verification interface or on an external device, the operator can perform actions on the controllable object.

[0010] Preferably, the motion characteristics of the controllable object are simulated using a realistic physics engine, including: The moving speed of the controllable object is updated according to v(t+1)=v(t)×friction coefficient+acceleration; Update the position of the controllable object according to pos(t+1)=pos(t)+v(t+1); Boundary collision detection is performed, and the speed is zeroed when the controllable object touches the boundary of the verification area; Wherein, v(t+1) is the velocity of the controllable object at time t+1; v(t) is the velocity of the controllable object at time t; the friction coefficient is set according to the resistance of the real physical environment; pos(t+1) is the position of the controllable object at time t+1; and pos(t) is the position of the controllable object at time t.

[0011] Preferably, the action sequence data includes: the direction of each directional key press, the key press timestamp, the duration of each key press, and a complete record of the key press sequence.

[0012] Preferably, the method further includes: When the distance between the center of the controllable object and the center of the target area is less than a preset distance threshold, it is determined that the controllable object has moved into the target area.

[0013] Preferably, the extraction of multi-dimensional behavioral features based on the acquired action sequence data includes: The average and standard deviation of the time difference between all adjacent keys are calculated to determine the characteristics of the operation rhythm. Count the number of directional changes that differ between all adjacent key directions to determine the frequency characteristics of directional changes. Calculate the operation time for the controllable object to move into the target area to determine the operation time characteristics; Path efficiency is calculated based on the formula: shortest distance / actual distance traveled, in order to determine the degree of path optimization. The coefficient of variation of the duration of all key presses was calculated to determine the force control characteristics.

[0014] Preferably, the analysis of the multi-dimensional behavioral characteristics to determine whether the operator is a real user includes: If the average value is within the preset time interval and the standard deviation is greater than the preset standard deviation threshold, then the analysis result of the operation rhythm characteristics is deemed qualified. If the number of directional changes is greater than or equal to a preset threshold, then the analysis result of the operation time characteristics is deemed qualified. If the operation duration is within the preset operation time range, then the analysis result of the operation duration feature is determined to be qualified; If the path efficiency is less than the preset path efficiency threshold, then the analysis result of the path optimization degree feature is qualified. If the coefficient of variation is greater than the preset coefficient of variation threshold, the analysis result of the force control feature is determined to be qualified. The number of qualified items in the multi-dimensional behavioral feature analysis is determined. If the number of qualified items is greater than or equal to a preset threshold, the operator is determined to be a real user; otherwise, the operator is determined not to be a real user.

[0015] According to another aspect of the present invention, an operation behavior verification system based on action sequence data is provided, the system comprising: An initialization unit is used to randomly generate a target area for operation behavior verification, initialize the position of the controllable object, and render the target area and the controllable object to the visual verification interface. A control unit is used to respond to the actions of the operator on the controllable object by simulating the motion characteristics of the controllable object using a real physics engine, so as to control the movement of the controllable object in the visual verification interface based on the motion characteristics. Action sequence data acquisition unit, used to acquire and record in real time the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface; The feature extraction unit is used to extract multi-dimensional behavioral features based on the acquired action sequence data when the controllable object moves into the target area. The analysis unit is used to analyze the multi-dimensional behavioral characteristics to determine whether the operator of the behavior is a real user.

[0016] Preferably, the control unit is further configured to: By receiving the operator's input of the four directional keys (up, down, left, right) on the visual verification interface or on an external device, the operator can perform actions on the controllable object.

[0017] Preferably, the control unit employs a realistic physics engine to simulate the motion characteristics of the controllable object, including: The moving speed of the controllable object is updated according to v(t+1)=v(t)×friction coefficient+acceleration; Update the position of the controllable object according to pos(t+1)=pos(t)+v(t+1); Boundary collision detection is performed, and the speed is zeroed when the controllable object touches the boundary of the verification area; Wherein, v(t+1) is the velocity of the controllable object at time t+1; v(t) is the velocity of the controllable object at time t; the friction coefficient is set according to the resistance of the real physical environment; pos(t+1) is the position of the controllable object at time t+1; and pos(t) is the position of the controllable object at time t.

[0018] Preferably, the action sequence data includes: the direction of each directional key press, the key press timestamp, the duration of each key press, and a complete record of the key press sequence.

[0019] Preferably, the feature extraction unit is further configured to: When the distance between the center of the controllable object and the center of the target area is less than a preset distance threshold, it is determined that the controllable object has moved into the target area.

[0020] Preferably, the feature extraction unit extracts multi-dimensional behavioral features based on the acquired action sequence data, including: The average and standard deviation of the time difference between all adjacent keys are calculated to determine the characteristics of the operation rhythm. Count the number of directional changes that differ between all adjacent key directions to determine the frequency characteristics of directional changes. Calculate the operation time for the controllable object to move into the target area to determine the operation time characteristics; Path efficiency is calculated based on the formula: shortest distance / actual distance traveled, in order to determine the degree of path optimization. The coefficient of variation of the duration of all key presses was calculated to determine the force control characteristics.

[0021] Preferably, the analysis unit is used to analyze the multi-dimensional behavioral characteristics to determine whether the behavioral operator is a real user, including: If the average value is within the preset time interval and the standard deviation is greater than the preset standard deviation threshold, then the analysis result of the operation rhythm characteristics is deemed qualified. If the number of directional changes is greater than or equal to a preset threshold, then the analysis result of the operation time characteristics is deemed qualified. If the operation duration is within the preset operation time range, then the analysis result of the operation duration feature is determined to be qualified; If the path efficiency is less than the preset path efficiency threshold, then the analysis result of the path optimization degree feature is qualified. If the coefficient of variation is greater than the preset coefficient of variation threshold, the analysis result of the force control feature is determined to be qualified. The number of qualified items in the multi-dimensional behavioral feature analysis is determined. If the number of qualified items is greater than or equal to a preset threshold, the operator is determined to be a real user; otherwise, the operator is determined not to be a real user.

[0022] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the operation behavior verification methods based on action sequence data.

[0023] According to another aspect of the present invention, the present invention provides an electronic device, comprising: The aforementioned computer-readable storage medium; and One or more processors for executing a program in the computer-readable storage medium.

[0024] This invention provides a method and system for verifying operational behavior based on action sequence data, comprising: randomly generating a target area for operational behavior verification, initializing the position of a controllable object, and rendering the target area and the controllable object onto a visual verification interface; responding to the operator's actions on the controllable object, using a realistic physics engine to simulate the motion characteristics of the controllable object, and controlling the movement of the controllable object in the visual verification interface based on the motion characteristics; acquiring and recording in real time the action sequence data of the operator controlling the movement of the controllable object in the visual verification interface; when the controllable object moves into the target area, extracting multi-dimensional behavioral features based on the acquired action sequence data; and analyzing the multi-dimensional behavioral features to determine whether the operator is a real user. This invention introduces directional key control and a physics engine into the field of CAPTCHAs, pioneering a new paradigm for verification interaction. It innovatively utilizes real-world physical motion laws as a verification dimension, increasing the difficulty for machines to crack the code. A five-dimensional behavioral feature analysis algorithm is proposed, comprehensively judging from multiple angles including operation rhythm, direction change, duration, path, and intensity. A fusion judgment mechanism of "passing at least 4 / 5 dimensions" is designed to balance security and false positive rate. The verification process is gamified, significantly improving user experience and acceptance. A complete cross-platform solution is implemented, supporting seamless switching between PCs and mobile devices. It can be widely applied to scenarios requiring human-machine identity verification, such as website login, mobile application authentication, game anti-cheating, and anti-bot fraud. Attached Figure Description

[0025] Exemplary embodiments of the present invention can be more fully understood by referring to the following figures: Figure 1 This is a flowchart of an operation behavior verification method 100 based on action sequence data according to an embodiment of the present invention; Figure 2This is an example diagram of a gravity tilt verification code interface according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the motion curve of the physics engine according to an embodiment of the present invention; Figure 4 This is an example diagram of multi-dimensional behavioral feature analysis according to an embodiment of the present invention; Figure 5 This is a comparison diagram of human operation versus robot operation according to an embodiment of the present invention. Figure 6 This is a schematic diagram of the structure of an operation behavior verification system 600 based on action sequence data according to an embodiment of the present invention. Detailed Implementation

[0026] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0027] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0028] Figure 1 This is a flowchart of an operation behavior verification method 100 based on action sequence data according to an embodiment of the present invention. Figure 1As shown, the operation behavior verification method based on action sequence data provided by this invention introduces directional key control and a physics engine into the field of CAPTCHAs, creating a new verification interaction paradigm; it innovatively utilizes real physical motion laws as verification dimensions, increasing the difficulty for machines to crack it; it proposes a five-dimensional behavior feature analysis algorithm, comprehensively judging from multiple angles such as operation rhythm, direction change, duration, path, and intensity; it designs a fusion judgment mechanism of "passing at least 4 / 5 dimensions" to balance security and false judgment rate; it gamifies the verification process, significantly improving user experience and acceptance; it achieves a complete cross-platform solution, supporting seamless switching between PCs and mobile devices; it can be widely applied to scenarios requiring human-machine identity verification, such as website login, mobile application authentication, game anti-cheating, and anti-bot fraud. The operation behavior verification method 100 based on action sequence data provided by this invention starts from step 101. In step 101, a target area for operation behavior verification is randomly generated, the position of the controllable object is initialized, and the target area and the controllable object are rendered onto a visual verification interface.

[0029] In this invention, when operational behavior verification is required, a result such as... Figure 2 The visual verification interface shown features a controllable ball as the controllable object. The interface includes: a game area (a 600×600 pixel square area with a white background and grid lines); a controllable ball (a blue ball with a diameter of 35 pixels, supporting motion simulation via a physics engine); a target area (marked by a green dashed circle, with randomly generated position and size); and directional control buttons (including four arrow keys for up, down, left, and right, supporting mouse and keyboard operation).

[0030] In step 102, in response to the action of the operator on the controllable object, a real physics engine is used to simulate the motion characteristics of the controllable object, so as to control the movement of the controllable object in the visualization verification interface based on the motion characteristics.

[0031] Preferably, the method further includes: By receiving the operator's input of the four directional keys (up, down, left, right) on the visual verification interface or on an external device, the operator can perform actions on the controllable object.

[0032] Preferably, the motion characteristics of the controllable object are simulated using a realistic physics engine, including: The moving speed of the controllable object is updated according to v(t+1)=v(t)×friction coefficient+acceleration; Update the position of the controllable object according to pos(t+1)=pos(t)+v(t+1); Boundary collision detection is performed, and the speed is zeroed when the controllable object touches the boundary of the verification area; Wherein, v(t+1) is the velocity of the controllable object at time t+1; v(t) is the velocity of the controllable object at time t; the friction coefficient is set according to the resistance of the real physical environment; pos(t+1) is the position of the controllable object at time t+1; and pos(t) is the position of the controllable object at time t.

[0033] In this invention, when the operator receives an action on the controllable object via the four directional keys (up, down, left, right) on the visual verification interface or on an external device (e.g., a keyboard), a real physics engine is used to simulate the motion characteristics of the controllable ball, so as to control the movement of the controllable ball in the visual verification interface based on the motion characteristics.

[0034] like Figure 3 As shown, in this invention, a realistic physics engine is used to simulate the motion of a small ball. The operator applies force via the directional keys, and the ball moves according to physical laws, exhibiting acceleration, deceleration, inertia, and other motion characteristics. Simulating the ball's motion based on real physical laws increases the realism of the verification and the difficulty of cracking it. Specifically, the velocity update formula is: v(t+1) = v(t) × friction coefficient + acceleration; the position update formula is: pos(t+1) = pos(t) + v(t+1); boundary collision detection includes: the ball's velocity is reset to zero and it stops moving when it touches the boundary; the friction coefficient is set to 0.95 to simulate the resistance of a real physical environment.

[0035] Specifically, the process of simulating the ball's motion using a realistic physics engine includes: 1. Initialization settings: Build a virtual simulation scene, display the verification interface, define the ball model and scene boundary parameters, and initialize the initial velocity v(0) and initial position pos(0) of the ball. The initial velocity v(0) is 0 by default, and the initial position pos(0) can be preset according to the scene requirements. At the same time, set the core parameters of the physical simulation, including the friction coefficient, which is set to 0.95 to simulate the resistance in the real physical environment and realize the natural deceleration of the ball during its motion, which conforms to the law of inertia in reality.

[0036] 2. User Interaction Force: The system acquires user input via arrow keys and converts these inputs into forces applied to the ball. These forces are then converted into the ball's acceleration. The magnitude and direction of the acceleration are determined by the arrow keys pressed by the user (e.g., pressing the up arrow key results in upward acceleration, and pressing the right arrow key results in rightward acceleration), enabling the user to precisely control the ball's direction of motion and acceleration.

[0037] 3. Speed ​​Update: Based on real-time calculations using the physics engine, the ball's speed is periodically updated according to a preset speed update formula: v(t+1) = v(t) × friction coefficient + acceleration; where v(t) is the ball's speed at time t, v(t+1) is the ball's speed at time t+1, the friction coefficient is fixed at 0.95, and the acceleration is generated by the user's input via the directional keys. This formula enables dynamic adjustment of the ball's speed. When the user applies a force, the ball gains acceleration and moves faster; when the user stops applying the force, the ball gradually decelerates under the influence of the friction coefficient, simulating the inertia and drag effects in a real physical environment.

[0038] 4. Position Update: After completing the velocity update, the position of the ball is periodically updated according to a preset position update formula: pos(t+1) = pos(t) + v(t+1); where pos(t) is the position coordinate of the ball at time t, pos(t+1) is the position coordinate of the ball at time t+1, and v(t+1) is the updated velocity at time t+1. The position update and velocity update are performed synchronously to ensure the continuity and accuracy of the ball's trajectory, conforming to the laws of real physical motion.

[0039] 5. Boundary Collision Detection: Real-time detection of the ball's position pos(t+1) relative to the scene boundary. When the ball is detected to have touched the scene boundary, the ball's current velocity v(t+1) is immediately reset to zero, causing the ball to stop moving. This avoids issues such as the ball crossing the boundary or abnormal rebound, ensuring the rationality of the simulated scene and the stability of the motion state. When the user applies force again using the directional keys, the ball resumes its movement according to the above velocity and position update logic.

[0040] In step 103, the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface is acquired and recorded in real time.

[0041] Preferably, the action sequence data includes: the direction of each directional key press, the key press timestamp, the duration of each key press, and a complete record of the key press sequence.

[0042] In this invention, the operator controls the movement of a small ball using four directional keys (up, down, left, right). During the controllable ball's movement, the system records the action sequence data of each key press in real time, including: the direction of each key press, the key press timestamp, the key press duration, and the complete key press sequence. Specifically, the key press direction includes: UP / DOWN / LEFT / RIGHT; the key press timestamp is recorded with millisecond-level precision; the key press duration records the time difference between pressing and releasing the key; and the key press sequence record shows the complete key press sequence.

[0043] In step 104, when the controllable object moves into the target area, multi-dimensional behavioral features are extracted based on the acquired action sequence data.

[0044] Preferably, the method further includes: When the distance between the center of the controllable object and the center of the target area is less than a preset distance threshold, it is determined that the controllable object has moved into the target area.

[0045] Preferably, the extraction of multi-dimensional behavioral features based on the acquired action sequence data includes: The average and standard deviation of the time difference between all adjacent keys are calculated to determine the characteristics of the operation rhythm. Count the number of directional changes that differ between all adjacent key directions to determine the frequency characteristics of directional changes. Calculate the operation time for the controllable object to move into the target area to determine the operation time characteristics; Path efficiency is calculated based on the formula: shortest distance / actual distance traveled, in order to determine the degree of path optimization. The coefficient of variation of the duration of all key presses was calculated to determine the force control characteristics.

[0046] In step 105, the multi-dimensional behavioral characteristics are analyzed to determine whether the behavioral operator is a real user.

[0047] Preferably, the analysis of the multi-dimensional behavioral characteristics to determine whether the operator is a real user includes: If the average value is within the preset time interval and the standard deviation is greater than the preset standard deviation threshold, then the analysis result of the operation rhythm characteristics is deemed qualified. If the number of directional changes is greater than or equal to a preset threshold, then the analysis result of the operation time characteristics is deemed qualified. If the operation duration is within the preset operation time range, then the analysis result of the operation duration feature is determined to be qualified; If the path efficiency is less than the preset path efficiency threshold, then the analysis result of the path optimization degree feature is qualified. If the coefficient of variation is greater than the preset coefficient of variation threshold, the analysis result of the force control feature is determined to be qualified. The number of qualified items in the multi-dimensional behavioral feature analysis is determined. If the number of qualified items is greater than or equal to a preset threshold, the operator is determined to be a real user; otherwise, the operator is determined not to be a real user.

[0048] In the present invention, the distance between the center of the small ball and the center of the target area is detected in real time, it is determined whether the distance is less than a preset distance threshold, and when the distance is less than the preset distance threshold, it is determined that the small ball has moved into the target area. Then, multi-dimensional behavior features are extracted based on the action sequence data of the entire movement process of the small ball, so as to perform analysis based on the multi-dimensional behavior features and determine whether the operation is a real person operation, that is, determine whether the behavior operator is a real person.

[0049] Specifically, in the present invention, in-depth analysis of the operation behavior is performed from five dimensions to determine whether the operation is a real person operation. The five dimensions specifically include: 1. Operation rhythm feature: Statistics the time interval between adjacent key operations. The operation rhythm of a real person is irregular, with variations in speed; robots tend to operate at fixed intervals or too fast.

[0050] Calculation method: Calculate the average value and standard deviation of the time differences between all adjacent keys.

[0051] Judgment rule: If the average interval is 50-800ms and the standard deviation is greater than 30ms, the analysis result is qualified.

[0052] 2. Direction change frequency feature: Statistics the number of times the user changes direction. A real person will adjust the direction frequently when controlling the small ball; a robot may adopt a simplified straight path. Calculation method: Statistics the number of times the directions of adjacent keys are different. Judgment rule: If the number of direction switches is greater than or equal to 3, the analysis result is qualified.

[0053] 3. Operation duration feature: Statistics the total duration from the start to successfully entering the target area. A real person takes 3-30 seconds to complete; a robot may operate too fast or exceed the time limit. Judgment rule: If the total duration is more than 3 seconds and less than 30 seconds, the analysis result is qualified.

[0054] 4. Path optimization degree feature: Analyze whether the path selected by the user is the theoretical optimal path. A real person will not take a perfectly optimal path; a robot may use an algorithm to calculate the optimal solution. Calculation method: Path efficiency = shortest distance / actual moving distance. Judgment rule: If the path efficiency is less than 0.85 (non-perfect path), the analysis result is qualified.

[0055] 5. Force control feature: Analyze the variation of the duration of each key press. The key pressing force of a real person is inconsistent; the simulated click duration of a robot is highly consistent. Calculation method: Statistics the coefficient of variation of the duration of all key presses. Judgment rule: If the coefficient of variation is greater than 30%, the analysis result is qualified.

[0056] In the present invention, if at least 4 of the above 5 dimensions meet the conditions, it is determined as a real person operation; otherwise, it is determined as a robot operation. This multi-dimensional fusion judgment method makes the robot have to simulate a variety of complex human behavior features at the same time, and the difficulty of cracking increases exponentially.

[0057] like Figure 4 The diagram shown is an example of multi-dimensional behavioral feature analysis according to an embodiment of the present invention. The comprehensive judgment result is that the conditions of the five dimensions are met, so the verification is passed and it is determined to be a real person's operation, that is, the operator of the behavior is determined to be a real person.

[0058] The method of this invention can be applied to mobile devices and PCs. The difference between mobile devices and PCs is that on mobile devices, physical directional keys are replaced with touch buttons, the interface is adaptively scaled to the screen width, the button size is increased to 80×80 pixels for easy touch, and touch press and release event listening is supported. The behavior analysis algorithms of the two devices are consistent, and the verification process is exactly the same.

[0059] The operation behavior verification method based on action sequence data provided by this invention has the following beneficial effects: 1. Significantly improved security Through five-dimensional behavioral feature analysis, the robot must simultaneously simulate multiple complex features such as operation rhythm, direction changes, duration control, path planning, and force changes, making it extremely difficult to crack. This improves security by over 90% compared to traditional CAPTCHAs.

[0060] 2. Innovative use of physical interaction This is the first time that a physics engine has been introduced into the field of CAPTCHA, using real physical motion laws to increase the verification dimension, which is an area that current machine learning models can hardly simulate perfectly.

[0061] 3. Excellent user experience The operation is intuitive and fun, similar to a mini-game, allowing users to have an entertainment experience while completing the verification; the average completion time is 5-10 seconds, and user acceptance is high.

[0062] 4. Strong cross-platform compatibility Supports keyboard operation on PC and touch button operation on mobile devices; fully implemented based on HTML5 and JavaScript, requiring no plugins and can be quickly integrated into any web application.

[0063] 5. Excellent anti-cheating capabilities The randomness of the directional key sequence and the complexity of the physical simulation make cheat programs difficult to predict and crack, making them particularly suitable for game anti-cheat scenarios.

[0064] The method of this invention is innovatively verified based on directional key control, combined with a physics engine and behavioral feature analysis. It significantly improves security while maintaining a good user experience and fun, and can be widely applied to the following scenarios: 1. Website login verification replaces traditional graphic CAPTCHAs, providing a safer and more fun login experience.

[0065] 2. The game's anti-cheat system utilizes a physics engine and behavior analysis to effectively identify and block cheat programs.

[0066] 3. Mobile application authentication provides a smooth touch verification experience on mobile devices.

[0067] 4. Anti-bot fraud scenarios such as e-commerce platforms and ticketing systems that require preventing bots from operating in bulk.

[0068] 5. Form submission validation: Perform human-computer verification before users submit important forms.

[0069] The following specific examples illustrate the embodiments of the present invention. Comparative features of human operation vs. robot operation, such as Figure 6 As shown. The characteristics of human operation are: irregular operating rhythm, frequent changes in direction, pauses and thinking, non-optimal paths, fluctuating speed, and varying button pressure. The characteristics of robot operation are: fixed operating rhythm, few changes in direction, smooth and uninterrupted operation, optimized paths, constant speed, and consistent button pressure.

[0070] In one embodiment of the present invention, the standard verification process is as follows: Step 1: System initialization, the ball is located at (150, 450), and the target area is randomly generated at (450, 150); Step 2: The user observes the target location and begins the operation; Step 3: User key press sequence: ↑(300ms)→(150ms)↑(280ms)→(200ms)↑(180ms)→(220ms); Step 4: The ball moves according to the laws of physics and enters the target area after 8.2 seconds; Step 5: The system recorded a total of 15 key presses and 5 direction changes in the operation sequence; Step 6: Five-dimensional analysis: rhythm 156ms (pass), direction 5 times (pass), duration 8.2s (pass), efficiency 78% (pass), intensity 45% (pass); Step 7: 5 / 5 dimensions passed, confirming it's a real person operating the system, verification successful.

[0071] In another embodiment of the present invention, when the robot operates and successfully identifies, the feature analysis result is as follows: A fixed 20ms interval for operation (too fast) → fails; Direction switching only once (using algorithm to optimize the path) → Failed; Total duration 0.8 seconds (too fast) → Not approved; Path efficiency 95% (too perfect) → Not approved; The button press duration is exactly the same (no change) → Fail; In summary, since all five dimensions failed, the system determined it to be a bot and refused verification.

[0072] Figure 6 This is a schematic diagram of the structure of an operation behavior verification system 600 based on action sequence data according to an embodiment of the present invention. Figure 6 As shown, the operation behavior verification system 600 based on action sequence data provided in this embodiment of the invention includes: an initialization unit 601, a control unit 602, an action sequence data acquisition unit 603, a feature extraction unit 604, and an analysis unit 605.

[0073] Preferably, the initialization unit 601 is used to randomly generate a target area for operation behavior verification, initialize the position of the controllable object, and render the target area and the controllable object to the visual verification interface.

[0074] Preferably, the control unit 602 is used to simulate the motion characteristics of the controllable object using a real physics engine in response to the action of the operator on the controllable object, so as to control the movement of the controllable object in the visual verification interface based on the motion characteristics.

[0075] Preferably, the control unit 602 is further configured to: By receiving the operator's input of the four directional keys (up, down, left, right) on the visual verification interface or on an external device, the operator can perform actions on the controllable object.

[0076] Preferably, the control unit 602 uses a real physics engine to simulate the motion characteristics of the controllable object, including: The moving speed of the controllable object is updated according to v(t+1)=v(t)×friction coefficient+acceleration; Update the position of the controllable object according to pos(t+1)=pos(t)+v(t+1); Boundary collision detection is performed, and the speed is zeroed when the controllable object touches the boundary of the verification area; Wherein, v(t+1) is the velocity of the controllable object at time t+1; v(t) is the velocity of the controllable object at time t; the friction coefficient is set according to the resistance of the real physical environment; pos(t+1) is the position of the controllable object at time t+1; and pos(t) is the position of the controllable object at time t.

[0077] Preferably, the action sequence data acquisition unit 603 is used to acquire and record in real time the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface.

[0078] Preferably, the action sequence data includes: the direction of each directional key press, the key press timestamp, the duration of each key press, and a complete record of the key press sequence.

[0079] Preferably, the feature extraction unit 604 is used to extract multi-dimensional behavioral features based on the acquired action sequence data when the controllable object moves into the target area.

[0080] Preferably, the feature extraction unit 604 is further configured to: When the distance between the center of the controllable object and the center of the target area is less than a preset distance threshold, it is determined that the controllable object has moved into the target area.

[0081] Preferably, the feature extraction unit 604 extracts multi-dimensional behavioral features based on the acquired action sequence data, including: The average and standard deviation of the time difference between all adjacent keys are calculated to determine the characteristics of the operation rhythm. Count the number of directional changes that differ between all adjacent key directions to determine the frequency characteristics of directional changes. Calculate the operation time for the controllable object to move into the target area to determine the operation time characteristics; Path efficiency is calculated based on the formula: shortest distance / actual distance traveled, in order to determine the degree of path optimization. The coefficient of variation of the duration of all key presses was calculated to determine the force control characteristics.

[0082] Preferably, the analysis unit 605 is used to analyze the multi-dimensional behavioral characteristics to determine whether the behavioral operator is a real user.

[0083] Preferably, the analysis unit 605 is used to analyze the multi-dimensional behavioral characteristics to determine whether the behavioral operator is a real user, including: If the average value is within the preset time interval and the standard deviation is greater than the preset standard deviation threshold, then the analysis result of the operation rhythm characteristics is deemed qualified. If the number of directional changes is greater than or equal to a preset threshold, then the analysis result of the operation time characteristics is deemed qualified. If the operation duration is within the preset operation time range, then the analysis result of the operation duration feature is determined to be qualified; If the path efficiency is less than the preset path efficiency threshold, then the analysis result of the path optimization degree feature is qualified. If the coefficient of variation is greater than the preset coefficient of variation threshold, the analysis result of the force control feature is determined to be qualified. The number of qualified items in the multi-dimensional behavioral feature analysis is determined. If the number of qualified items is greater than or equal to a preset threshold, the operator is determined to be a real user; otherwise, the operator is determined not to be a real user.

[0084] The operation behavior verification system 600 based on action sequence data in an embodiment of the present invention corresponds to the operation behavior verification method 100 based on action sequence data in another embodiment of the present invention, and will not be described again here.

[0085] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the operation behavior verification methods based on action sequence data.

[0086] According to another aspect of the present invention, the present invention provides an electronic device, comprising: The aforementioned computer-readable storage medium; and One or more processors for executing a program in the computer-readable storage medium.

[0087] The present invention has been described with reference to a few embodiments. However, it will be apparent to those skilled in the art that other embodiments besides those disclosed above fall equivalently within the scope of the present invention.

[0088] Generally, all terms used in this invention are interpreted according to their ordinary meaning in the art, unless otherwise expressly defined herein. All references to “a / the / the [device, component, etc.]” ​​are openly interpreted as at least one instance of said device, component, etc., unless otherwise expressly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed, unless explicitly stated otherwise.

[0089] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0091] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for verifying operational behavior based on action sequence data, characterized in that, The method includes: Randomly generate a target area for verifying operational behavior, initialize the position of the controllable object, and render the target area and the controllable object onto the visual verification interface; In response to the actions of the operator on the controllable object, a real physics engine is used to simulate the motion characteristics of the controllable object, so as to control the movement of the controllable object in the visual verification interface based on the motion characteristics; Acquire and record in real time the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface; When the controllable object moves into the target area, multi-dimensional behavioral features are extracted based on the acquired action sequence data; The multi-dimensional behavioral characteristics are analyzed to determine whether the operator of the behavior is a real user.

2. The method according to claim 1, characterized in that, The method further includes: By receiving the operator's input of the four directional keys (up, down, left, right) on the visual verification interface or on an external device, the operator can perform actions on the controllable object.

3. The method according to claim 1, characterized in that, The motion characteristics of the controllable object are simulated using a realistic physics engine, including: The moving speed of the controllable object is updated according to v(t+1)=v(t)×friction coefficient+acceleration; Update the position of the controllable object according to pos(t+1)=pos(t)+v(t+1); Boundary collision detection is performed, and the speed is zeroed when the controllable object touches the boundary of the verification area; Wherein, v(t+1) is the velocity of the controllable object at time t+1; v(t) is the velocity of the controllable object at time t; the friction coefficient is set according to the resistance of the real physical environment; pos(t+1) is the position of the controllable object at time t+1; and pos(t) is the position of the controllable object at time t.

4. The method according to claim 1, characterized in that, The action sequence data includes: the direction of each directional key press, the key press timestamp, the duration of each key press, and a complete record of the key press sequence.

5. The method according to claim 1, characterized in that, The method further includes: When the distance between the center of the controllable object and the center of the target area is less than a preset distance threshold, it is determined that the controllable object has moved into the target area.

6. The method according to claim 1, characterized in that, Multi-dimensional behavioral features are extracted based on the acquired action sequence data, including: The average and standard deviation of the time difference between all adjacent keys are calculated to determine the characteristics of the operation rhythm. Count the number of directional changes that differ between all adjacent key directions to determine the frequency characteristics of directional changes. Calculate the operation time for the controllable object to move into the target area to determine the operation time characteristics; Path efficiency is calculated based on the formula: shortest distance / actual distance traveled, in order to determine the degree of path optimization. The coefficient of variation of the duration of all key presses was calculated to determine the force control characteristics.

7. The method according to claim 6, characterized in that, Analyzing the multi-dimensional behavioral characteristics to determine whether the operator of the behavior is a real user includes: If the average value is within the preset time interval and the standard deviation is greater than the preset standard deviation threshold, then the analysis result of the operation rhythm characteristics is deemed qualified. If the number of directional changes is greater than or equal to a preset threshold, then the analysis result of the operation time characteristics is deemed qualified. If the operation duration is within the preset operation time range, then the analysis result of the operation duration feature is determined to be qualified; If the path efficiency is less than the preset path efficiency threshold, then the analysis result of the path optimization degree feature is qualified. If the coefficient of variation is greater than the preset coefficient of variation threshold, the analysis result of the force control feature is determined to be qualified. The number of qualified items in the multi-dimensional behavioral feature analysis is determined. If the number of qualified items is greater than or equal to a preset threshold, the operator is determined to be a real user; otherwise, the operator is determined not to be a real user.

8. An operation behavior verification system based on action sequence data, characterized in that, The system includes: An initialization unit is used to randomly generate a target area for operation behavior verification, initialize the position of the controllable object, and render the target area and the controllable object to the visual verification interface. A control unit is used to respond to the actions of the operator on the controllable object by simulating the motion characteristics of the controllable object using a real physics engine, so as to control the movement of the controllable object in the visual verification interface based on the motion characteristics. Action sequence data acquisition unit, used to acquire and record in real time the action sequence data of the operator controlling the movement of the controllable object on the visual verification interface; The feature extraction unit is used to extract multi-dimensional behavioral features based on the acquired action sequence data when the controllable object moves into the target area. The analysis unit is used to analyze the multi-dimensional behavioral characteristics to determine whether the operator of the behavior is a real user.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-7.

10. An electronic device, characterized in that, include: The computer-readable storage medium as described in claim 9; as well as One or more processors for executing a program in the computer-readable storage medium.

Images