Privacy protection method, device, equipment and storage medium

Abstract

The present application discloses a privacy protection method, apparatus, device and storage medium, which relates to the field of privacy protection technology, including: obtaining the user's operation behavior data on the front-end interactive page to construct a user operation feature vector; performing a privacy risk assessment operation to obtain the privacy risk level of multiple page areas of the front-end interactive page; selecting a privacy protection policy that matches the privacy risk level from the privacy protection policy set to generate multiple policy tasks; predicting the subsequent interaction probability of the fertility page area other than the user's current interactive page area in multiple page areas and combining the corresponding privacy risk levels to sort the multiple policy tasks, generate a policy task execution sequence and execute them in sequence. The present application can integrate user behavior modeling to perform page area-level privacy risk assessment, realize automatic dynamic identification of risks and trusted execution of policies, and solve the problems of insufficient dynamic protection and lack of multi-dimensional privacy coverage in the existing technology.

Description

Technical Field

[0001] The present application relates to the field of privacy protection technology, and in particular to a privacy protection method, apparatus, device, and storage medium. Background Art

[0002] With the increasingly sophisticated development of Internet services and the increasing reliance on user data, user operational behavior data on front-end interactive pages (such as click events, page dwell time, scrolling behavior, input content, etc.) are increasingly becoming an important resource for supporting personalized recommendations, behavior prediction, and service optimization; however, these operational behaviors often carry users' privacy preferences, intentions, and even sensitive information, especially in areas such as identity authentication, financial information, and medical records. Users may be exposed to privacy risks without knowing it.

[0003] Although existing technologies have provided basic protection measures such as image blurring, third-party cookie blocking, and permission pop-up prompts, there are still technical problems that need to be solved: First, the protection logic mainly relies on static configuration rules, which is difficult to adapt to real-time changes in user behavior, resulting in insufficient privacy protection in dynamic scenarios; second, users need to actively configure privacy policies or perform manual authorization, which increases the operational burden and affects the user experience; third, existing technical means are mostly single-point protection, which is difficult to simultaneously cover multi-dimensional privacy risks such as page area identification, operation trajectory analysis, and behavior prediction. Summary of the Invention

[0004] This application provides a privacy protection method that can integrate user behavior modeling to conduct privacy risk assessment at the page area level, thereby accurately scheduling and executing tasks for privacy protection strategies, realizing automated dynamic identification of risks and trusted execution of strategies, so as to at least solve the problems of insufficient dynamic protection, strong configuration dependence, lack of multi-dimensional privacy coverage, and untraceable records in the existing technology.

[0005] This application provides a privacy protection method, including:

[0006] Obtain user operation behavior data on the front-end interactive page and construct user operation feature vectors based on the operation behavior data;

[0007] Based on the user operation feature vector, a privacy risk assessment operation is performed to obtain the privacy risk levels of multiple page areas of the front-end interactive page;

[0008] Based on the privacy risk levels of multiple page areas, privacy protection policies that match the privacy risk levels are selected from a pre-stored privacy protection policy set to generate multiple policy tasks;

[0009] Execute the corresponding policy task for the user's current interactive page area, and predict the subsequent interaction probability of the remaining page areas in multiple page areas except the user's current interactive page area based on the user's historical behavior data. Combined with the corresponding privacy risk level, multiple policy tasks are sorted to generate a policy task execution sequence;

[0010] Based on the execution sequence, the execution parameters of multiple strategy tasks are pre-cached and executed sequentially.

[0011] This application also provides a privacy protection device, including:

[0012] The operation record module is used to obtain the user's operation behavior data on the front-end interactive page and construct the user operation feature vector based on the operation behavior data;

[0013] The risk assessment module is used to perform a privacy risk assessment operation based on the user operation feature vector to obtain the privacy risk level of multiple page areas of the front-end interactive page;

[0014] A policy task generation module is used to select privacy protection policies that match the privacy risk levels of multiple page areas from a pre-stored privacy protection policy set based on the privacy risk levels of the multiple page areas, and generate multiple policy tasks;

[0015] The task scheduling module is used to execute the corresponding policy task for the user's current interactive page area, and predict the subsequent interaction probability of the remaining page areas in multiple page areas other than the user's current interactive page area based on the user's historical behavior data. It also sorts the multiple policy tasks according to the corresponding privacy risk level and generates a policy task execution sequence;

[0016] The task execution module is used to pre-cache the execution parameters of multiple strategy tasks based on the execution sequence and execute them in sequence.

[0017] The present application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of any one of the above-mentioned privacy protection methods when executing the computer program.

[0018] The present application also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned privacy protection methods are implemented.

[0019] Through this application, by collecting the user's operation behavior data in the front-end page and constructing the user operation feature vector, multi-dimensional modeling of the user's interaction intention is achieved, thereby improving the ability to characterize the page behavior risk, and effectively breaking through the limitation of the existing technology that cannot dynamically perceive user behavior; at the same time, due to the execution of risk assessment operations, the privacy risk level of different page areas can be accurately identified, thereby solving the existing problem of relying on static rules and lack of real-time judgment; furthermore, due to the combination of the user's current interaction area and the user's historical behavior data, the probability of subsequent interaction page areas is predicted, and the policy tasks can be sorted to achieve forward-looking and priority control of policy responses, thereby improving the protection efficiency in dynamic interaction scenarios, and, based on the sorting results, the policy tasks are pre-cached and managed, and high-priority policy tasks can be loaded in advance to ensure that the policy does not require manual configuration by the user, and can be ready in advance before the user is about to interact with the sensitive area, significantly reducing the response delay, and solving the problem of protection trigger lag in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] Figure 1 A schematic diagram of an application environment for a privacy protection architecture provided in an embodiment of the present application;

[0022] Figure 2 A schematic diagram of the operation of a privacy protection engine provided in an embodiment of the present application;

[0023] Figure 3 A schematic diagram of a task scheduler provided in an embodiment of the present application;

[0024] Figure 4 A user interaction diagram provided in an embodiment of the present application;

[0025] Figure 5 A privacy protection step diagram provided in an embodiment of the present application;

[0026] Figure 6 A flowchart of a privacy protection method provided in an embodiment of the present application;

[0027] Figure 7 A structural block diagram of a privacy protection device provided in an embodiment of the present application;

[0028] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0030] It should be noted that, in the description of this application, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. The terms "first," "second," etc., in this application are used to distinguish similar objects, and are not used to describe a particular order or sequence.

[0031] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0032] This application provides a privacy protection method that can be applied to Figure 1 In the application environment shown, client 102 communicates with server 104 via a network. Server 104 deploys a privacy protection architecture, which includes an event listener, a privacy protection engine, a resource monitor, a task scheduler, and a task recorder to protect the privacy of the front-end interactive interface currently operated by the user on client 102. Client 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. Server 104 can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0033] It is worth noting that if Figure 2 As shown, the event listener takes the user operation data on the client and sends it to the privacy protection engine. The privacy protection engine receives the user operation data sent by the event listener, performs intelligent analysis and risk assessment on the user operation data, and determines whether the privacy risk level of the front-end interactive page area where the user is currently operating is medium risk or high risk. If so, the corresponding privacy protection policy is matched according to the privacy risk level. Otherwise, the current privacy protection policy is maintained.

[0034] It is worth noting that if Figure 3As shown, the task scheduler receives the policy task, divides the policy task into multiple subtasks according to the required resources, and allocates them to the CPU, GPU and cloud respectively to achieve distributed task execution, and summarizes the results of the completed subtasks and returns them to the client.

[0035] It is worth noting that if Figure 4 As shown, the user operates the front-end interactive page on the client, and the event listener in the server monitors the user operation in real time and sends data to the privacy protection engine. The privacy protection engine performs policy matching and task generation, and sends the generated policy tasks to the task scheduler. The task scheduler receives the system status sent by the resource monitor and the policy tasks sent by the privacy protection engine, sorts, schedules and executes the policy tasks, and sends the execution results to the client. The corresponding privacy protection page is loaded on the client to achieve privacy protection of the front-end interactive page.

[0036] like Figure 5 As shown, an embodiment of the present application provides a privacy protection method, including:

[0037] Step 201: Obtain user operation behavior data on the front-end interactive page, and construct a user operation feature vector based on the operation behavior data;

[0038] Step 202: Perform a privacy risk assessment based on the user operation feature vector to obtain privacy risk levels for multiple page areas of the front-end interactive page.

[0039] Step 203 , based on the privacy risk levels of the multiple page regions, select privacy protection policies that match the privacy risk levels from a pre-stored privacy protection policy set to generate multiple policy tasks;

[0040] Step 204: Execute the corresponding policy task for the user's current interactive page area, and predict the subsequent interaction probability of the remaining page areas other than the user's current interactive page area based on the user's historical behavior data. Then, sort the multiple policy tasks according to their corresponding privacy risk levels to generate a policy task execution sequence.

[0041] Step 205 : Based on the execution sequence, pre-cache the execution parameters of multiple policy tasks and execute them in sequence.

[0042] Specifically, the present application provides a method for collecting user operation behavior data in the front-end page and constructing a user operation feature vector to achieve multi-dimensional modeling of user interaction intentions, thereby improving the ability to characterize page behavior risks and effectively breaking through the limitation of the existing technology that cannot dynamically perceive user behavior; at the same time, due to the execution of risk assessment operations, the privacy risk levels of different page areas can be accurately identified, thereby solving the existing problem of relying on static rules and lack of real-time judgment; furthermore, due to the combination of the user's current interaction area and the user's historical behavior data, the probability of subsequent interaction page areas is predicted, and the policy tasks can be sorted to achieve forward-looking and priority control of policy responses, thereby improving the protection efficiency in dynamic interaction scenarios, and pre-caching management of policy tasks based on the sorting results can load high-priority policy tasks in advance to ensure that the policy is ready in advance before the user is about to interact with the sensitive area, significantly reducing the response delay, and solving the problem of protection trigger lag in the existing technology.

[0043] In one embodiment, Figure 6 As shown, a privacy protection method provided by the present application is based on the following logic implementation: first, operation behavior data is acquired to obtain the user's operation data on the front-end interactive page; then, the operation behavior data is analyzed to realize the privacy risk assessment of the page area of ​​the front-end interactive page; furthermore, visual feedback is generated to the front-end interactive page based on the results of the privacy risk assessment to remind the user of the current risk status. At the same time, the privacy protection strategy is matched and the corresponding policy tasks are generated to perform targeted privacy protection on the page area; thereafter, the generated policy tasks are reordered so that the execution of the policy tasks can be more in line with user use; finally, the policy tasks are executed to realize privacy protection, wherein, when executing the policy tasks, the policy tasks to be executed are pre-cached to improve deployment efficiency, and a task completion record is generated for the executed policy tasks and written into the blockchain as a verifiable credential, and the user operations are monitored in real time to update the privacy risk level and the corresponding policy tasks at any time.

[0044] In one embodiment, obtaining the user's operation behavior data on the front-end interactive page and constructing the user operation feature vector based on the operation behavior data include:

[0045] Calling the performance interface provided by the front-end interactive page to collect user operation records in multiple page areas and obtain operation behavior data. The operation behavior data includes at least: the screen coordinate value sequence of the click event, the entry time and exit time of the page area, the scrolling speed and scrolling acceleration of the page scrolling event, and the input character sequence of the input content change event;

[0046] According to the screen coordinate value sequence of the click event, the standard deviation of the click position distribution is obtained;

[0047] According to the entry time and exit time of the page area, the average stay time of the page area is obtained;

[0048] According to the scrolling speed and scrolling acceleration of the page scrolling event, the scrolling acceleration integral value is obtained;

[0049] Calculate the weight of sensitive words in the input text based on the input character sequence of the input content change event;

[0050] Call the sensor of the device where the front-end interactive page is located to obtain the screen tilt angle and ambient light intensity and generate sensor auxiliary feature values;

[0051] The standard deviation of click position distribution, average dwell time in the page area, scroll acceleration integral value, weight value of sensitive words in the input text, and sensor-assisted feature value are concatenated to obtain the user operation feature vector.

[0052] Specifically, this embodiment defines behavioral indicators in dimensions such as the standard deviation of click position distribution, average page dwell time, scrolling acceleration integral value, input sensitive word weight, and sensor-assisted feature value, and combines them to construct a user operation feature vector. This helps to fine-grainedly characterize the user's behavioral characteristics in different page areas. Compared with the traditional method of only collecting simple interaction data, it achieves more comprehensive and accurate user behavior modeling, provides a stable data foundation for subsequent dynamic assessment of privacy risks, and enhances the adaptability and accuracy of protection strategies.

[0053] In a preferred embodiment, the standard deviation of the click position distribution is obtained based on the screen coordinate value sequence of the click event and is calculated based on the following formula:

[0054] ;

[0055] in, It represents the standard deviation of the click position distribution of user operations, which is used to measure the concentration of user click behavior. n represents the total number of click events, and x i Indicates the screen horizontal coordinate of the i-th click event, y i Indicates the screen vertical coordinate of the i-th click event, Indicates the average horizontal coordinate of the screen for all click events. Indicates the average vertical coordinate of the screen for all click events.

[0056] In a preferred embodiment, the scroll acceleration integral value is obtained according to the scroll speed and scroll acceleration of the page scroll event, and is obtained based on the following formula:

[0057] ;

[0058] Among them, S aIndicates the scrolling acceleration integral value of the user operation, which is used to reflect the continuity and rapidity of page browsing. k is the kth rolling acceleration, and Δt is the rolling event sampling interval.

[0059] In a preferred embodiment, the weight value of the sensitive word in the input text is calculated based on the input character sequence of the input content change event, including:

[0060] Based on the input character sequence, the predefined sensitive dictionary is matched. The sensitive dictionary records content including but not limited to ID card, mobile phone number, and address;

[0061] For each sensitive word in the dictionary w i Given weight If the input contains multiple sensitive words, its overall sensitivity weight value S sw Based on the following formula:

[0062] ;

[0063] Among them, S sw It represents the overall weight value of the sensitive words in the input text of the user operation, which is used to reflect the sensitivity of the text input by the user. M represents the set of sensitive words identified in the input text.

[0064] In one embodiment, a privacy risk assessment operation is performed based on the user operation feature vector to obtain privacy risk levels of multiple page regions, including:

[0065] Divide the user operation feature vector by page area to obtain multiple behavior segment vectors;

[0066] Construct a privacy risk assessment model, which includes a temporal feature extraction sub-model based on a long short-term memory network and a reinforcement scoring sub-model based on a reinforcement learning algorithm;

[0067] Through the temporal feature extraction sub-model, the time-dependent features in multiple behavior segment vectors are extracted and the intermediate hidden state sequence is output;

[0068] The intermediate hidden state sequence is input into the reinforcement scoring sub-model, and the behavior privacy risk score is output through the pre-trained state-action value function;

[0069] The behavioral privacy risk score value is compared with the preset privacy risk level range to determine the privacy risk level of the corresponding page area. Preferably, when the behavioral privacy risk score value is greater than or equal to 0.7, the privacy risk level of the corresponding page area is determined to be high risk; when the behavioral privacy risk score value is greater than or equal to 0.4 and less than 0.7, the privacy risk level of the corresponding page area is determined to be medium risk; when the behavioral privacy risk score value is less than 0.4, the privacy risk level of the corresponding page area is determined to be low risk.

[0070] In a preferred embodiment, the intermediate hidden state sequence is input into the reinforcement scoring sub-model, and the behavior privacy risk score is output through the pre-trained state-action value function, which is obtained based on the following formula:

[0071] ;

[0072] Among them, Q (s, a) represents taking action in state s The risk score of is the behavior privacy risk score corresponding to the behavior fragment vector under the current page area, α is the learning rate, r is the current reward value, γ is the discount factor, Indicates the future state All possible moves The maximum score.

[0073] Specifically, in this embodiment, the user operation feature vector is divided into multiple behavior segment vectors, and the time-dependent features are extracted using a long short-term memory network. The privacy risk score is output in combination with a reinforcement learning model, and then matched to the risk level interval. This enables the privacy risk assessment to not only have time series perception capabilities, but also adaptively optimize the judgment strategy, overcoming the problem that the static scoring model cannot adapt to behavioral changes. It can more accurately identify the potential privacy risks of users in different areas, and improve the dynamic and intelligent level of front-end risk identification.

[0074] In one embodiment, after performing a privacy risk assessment operation based on the user operation feature vector to obtain the privacy risk levels of multiple page regions, the method further includes:

[0075] Based on the privacy risk levels of multiple page areas, corresponding privacy protection visual feedback is generated to the front-end interactive page. Among them, high-risk page areas preferably use red particle icons, medium-risk page areas preferably use red particle icons, and low-risk page areas preferably use green particle icons.

[0076] In one embodiment, based on the privacy risk levels of multiple page regions, privacy protection policies that match the privacy risk levels are selected from a pre-stored privacy protection policy set to generate multiple policy tasks, including:

[0077] Set a privacy risk level to correspond to at least one or more privacy protection strategies, and generate a gradient strategy table. Preferably, when the privacy risk level is high, the corresponding privacy protection strategy at least includes enabling homomorphic encryption, displaying a virtual keyboard, and preventing screenshots. When the privacy risk level is medium, the corresponding privacy protection strategy at least includes using format-preserving encryption and displaying an input box watermark. When the privacy risk level is low, the corresponding privacy protection strategy at least includes performing data desensitization.

[0078] Obtain the privacy risk level of the page area, query the gradient policy table, and determine the privacy protection policy corresponding to the privacy risk level of the page area as the target policy;

[0079] The page area identifier of the combined page, the type identifier of the target policy, and the execution parameters are used to construct a corresponding policy task.

[0080] Specifically, this embodiment establishes a mapping relationship between privacy risk levels and multi-level protection strategies, namely a gradient strategy table, so that privacy protection strategies can be dynamically matched according to the risk levels of page areas. At the same time, the construction of strategy tasks not only considers the area identification and strategy type, but also includes specific execution parameters to ensure that each strategy has a clear execution goal and context adaptability, thereby improving the personalization of strategy configuration and the effectiveness of execution, and solving the shortcomings of the existing technology of single strategy response and lack of flexibility.

[0081] In one embodiment, predicting the subsequent interaction probability of the remaining page areas except the user's current interaction page area among multiple page areas based on the user's historical behavior data includes:

[0082] Extract user historical behavior data to obtain a time series record of user interaction events;

[0083] Construct a behavior path sequence based on the time series records of user interaction events;

[0084] Extract the page area identifier where the user interaction event is located in the behavior path sequence, obtain the interaction area node sequence and encode it in chronological order to construct a state set;

[0085] On the state set, the actual jump times between two states are counted in sequence. Nodes are set as states, edges are set as user jump behaviors, and edge weights are set as jump frequencies. A state jump frequency graph is constructed and a state transition probability matrix is ​​generated.

[0086] The user's current interactive page area is taken as input, and a multi-step jump simulation is performed on the state transition probability matrix to output the subsequent interaction probabilities of the remaining page areas respectively.

[0087] Specifically, this embodiment constructs a user interaction behavior path sequence and generates a state transition probability matrix based on the state jump frequency graph, which can make multi-step predictions on the user's subsequent interaction page areas, thereby knowing the high-probability interaction areas in advance; at the same time, this prediction capability enables privacy protection strategies to be prepared in advance before the user actually reaches the page, improving the preemptive nature of protection and resource scheduling efficiency, and effectively avoiding protection lags.

[0088] In one embodiment, multiple policy tasks are sorted according to their corresponding privacy risk levels to generate a policy task execution sequence, including:

[0089] According to the privacy risk level and subsequent interaction probability of the remaining page area, the ranking score of the remaining page area is obtained by weighted calculation;

[0090] According to the ranking scores of the remaining page areas, multiple corresponding policy tasks are arranged in descending order to obtain a policy task execution sequence.

[0091] Specifically, in this embodiment, the privacy risk level of the page area and the subsequent interaction probability are weightedly combined to construct a ranking score and sort the policy tasks. This helps to prioritize the execution of critical protection tasks with high risks and high interaction probabilities under limited system resources, and can achieve dynamic scheduling and priority management of policy resources, ensuring the timeliness and practicality of privacy protection, thereby improving the execution efficiency and risk mitigation capabilities of the overall privacy protection system.

[0092] In one embodiment, based on the execution sequence, execution parameters of multiple policy tasks are pre-cached and executed sequentially, including:

[0093] Set the number of cacheable tasks based on the current system resource status;

[0094] Taking the number of cacheable tasks as the selected number, select multiple policy tasks ranked first from the policy task execution sequence and write the corresponding multiple execution parameters into the front-end cache area;

[0095] According to the order of the policy task execution sequence, the cached policy tasks are scheduled and executed in sequence;

[0096] In response to the completion of the policy task execution, the execution parameters of the next policy task that has not been executed in the policy task execution sequence and has a higher ranking are obtained to replace the cache data of the completed task in the front-end cache area.

[0097] Specifically, in this embodiment, dynamic cache management is performed by setting the number of cacheable tasks and combining them with the policy task execution sequence. At the same time, an execution record is generated after the task is completed and stored through the blockchain, thereby ensuring the consistency, integrity and non-tamperability of the protection strategy during the execution process. It not only realizes the resource preloading and scheduling optimization before the execution of the policy task, but also establishes a trusted execution record chain, enhances the user's verifiability of the system's privacy protection behavior, and effectively solves the problem of unauditable policy execution process in the existing technology.

[0098] In a preferred embodiment, the number of cacheable tasks is set according to the current system resource status, including:

[0099] Obtain the available cache space of the current device system and the average execution parameter size of the policy task, and combine it with the preset maximum concurrent cache limit to calculate the number of cacheable tasks. Specifically, based on the following formula:

[0100] ;

[0101] Among them, S avail Indicates the available cache space in the current system, S avg is the average size of a single policy parameter package, K is the maximum number of parallel cache tasks, and is preferably set to 10~20.

[0102] Specifically, in this embodiment, by dynamically setting the number of cacheable tasks according to the current resource status of the system, the caching behavior of the policy tasks can be adaptively matched with the device operating environment. When resources are sufficient, the cache space can be fully utilized to load more high-priority tasks in advance, thereby improving the timeliness of policy response; and when resources are limited, the cache size is automatically reduced to avoid excessive memory occupation causing system freezes or functional abnormalities, thereby effectively improving the overall resource utilization efficiency and operational stability of the system, enhancing the adaptability of the privacy protection mechanism in various client environments, and overcoming the problems of policy loading delays or uneven system loads caused by the fixed number of cache tasks and rigid resource scheduling in the existing technology.

[0103] In one embodiment, after pre-caching the execution parameters of multiple policy tasks based on the execution sequence and executing them in sequence, the method further includes:

[0104] Build a layered blockchain architecture composed of consortium chains and public chains. In it, the consortium chain is set up to store basic metadata such as user identity tags, privacy preference settings, policy configurations, and has node interaction permission control. The public chain is set up to record hashed sensitive operation log anchor information and ensure the verifiability and non-tamperability of key behaviors. The on-chain data interface is set up to support the automatic hashing of task execution records.

[0105] In one embodiment, after constructing a layered blockchain architecture composed of a consortium chain and a public chain, the following steps are further included:

[0106] In response to the completion of the policy task execution, the task identifier, execution time and execution result information of the completed policy task are obtained and spliced ​​into a policy execution record data packet, and the policy execution record data packet is hashed with SHA-256 to generate a fixed-length digest value;

[0107] The summary value is used as the anchor value for sensitive policy operations and is combined with the task ID and user ID to form an anchor record.

[0108] The anchor record is written into the block through the consortium chain to form an on-chain evidence of the policy task. In response to the task being a highly sensitive operation, that is, the corresponding privacy risk level is high, the record is also anchored to the public chain to enhance public verification, forming a dual trust path;

[0109] Based on the state-parameter pair of task execution, a zero-knowledge proof (such as a zk-SNARK structure) is generated. The zero-knowledge proof, along with the task identifier and on-chain evidence, is stored in the smart contract for the auditor or user client to call for on-chain verification.

[0110] This embodiment achieves trusted evidence storage and verifiable protection of privacy policy task execution records by introducing a layered blockchain architecture composed of a consortium chain and a public chain. By recording user identity and policy metadata in the consortium chain and anchoring sensitive operation summaries to the public chain in high-risk policy operation scenarios, a complete "controllable interaction + tamper-proof + verifiable" trust mechanism is established. Furthermore, combined with a zero-knowledge proof protocol, it allows external auditors or user clients to verify the policy execution status without exposing plaintext data. This effectively addresses the technical challenges of existing privacy protection solutions, such as the lack of transparency, credibility, and auditability of the policy execution process, further enhancing the system's security, compliance, and user trust.

[0111] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method.

[0112] The embodiment of the present application also provides a privacy protection device, such as Figure 7 Shown, including:

[0113] The operation record module is used to obtain the user's operation behavior data on the front-end interactive page and construct the user operation feature vector based on the operation behavior data;

[0114] The risk assessment module is used to perform a privacy risk assessment operation based on the user operation feature vector to obtain the privacy risk level of multiple page areas of the front-end interactive page;

[0115] A policy task generation module is used to select privacy protection policies that match the privacy risk levels of multiple page areas from a pre-stored privacy protection policy set based on the privacy risk levels of the multiple page areas, and generate multiple policy tasks;

[0116] The task scheduling module is used to execute the corresponding policy task for the user's current interactive page area, and predict the subsequent interaction probability of the remaining page areas in multiple page areas other than the user's current interactive page area based on the user's historical behavior data. It also sorts the multiple policy tasks according to the corresponding privacy risk level and generates a policy task execution sequence;

[0117] The task execution module is used to pre-cache the execution parameters of multiple strategy tasks based on the execution sequence and execute them in sequence.

[0118] The operation recording module is also used to call the performance interface provided by the front-end interactive page, collect the user's operation records in multiple page areas, and obtain operation behavior data, which at least includes: the screen coordinate value sequence of the click event, the entry time and exit time of the page area, the scrolling speed and scrolling acceleration of the page scrolling event, and the input character sequence of the input content change event; obtain the standard deviation of the click position distribution based on the screen coordinate value sequence of the click event; obtain the average residence time of the page area based on the entry time and exit time of the page area; obtain the scrolling acceleration integral value based on the scrolling speed and scrolling acceleration of the page scrolling event; calculate the weight value of the sensitive word in the input text based on the input character sequence of the input content change event; call the sensor of the device where the front-end interactive page is located, obtain the screen tilt angle and ambient light intensity and generate a sensor auxiliary feature value; splice the standard deviation of the click position distribution, the average residence time of the page area, the scrolling acceleration integral value, the weight value of the sensitive word in the input text, and the sensor auxiliary feature value to obtain the user operation feature vector.

[0119] The risk assessment module is also used to divide the user operation feature vectors according to page areas to obtain multiple behavior fragment vectors; construct a privacy risk assessment model, which includes a temporal feature extraction sub-model constructed based on a long short-term memory network and a reinforcement scoring sub-model constructed based on a reinforcement learning algorithm; through the temporal feature extraction sub-model, the time-dependent features in multiple behavior fragment vectors are extracted, and an intermediate latent state sequence is output; the intermediate latent state sequence is input into the reinforcement scoring sub-model, and the behavior privacy risk score value is output through the pre-trained state-action value function; the behavior privacy risk score value is compared with the preset privacy risk level range to determine the privacy risk level of the corresponding page area.

[0120] The policy task generation module is also used to set a privacy risk level corresponding to at least one or more privacy protection policies and generate a gradient policy table; obtain the privacy risk level of the page area, query the gradient policy table, and determine the privacy protection policy corresponding to the privacy risk level of the page area as the target policy; combine the page area identifier of the page, the type identifier of the target policy and the execution parameters to construct the corresponding policy task.

[0121] The task scheduling module is also used to extract user historical behavior data to obtain a time-series record of user interaction events; construct a behavior path sequence based on the time-series record of user interaction events; extract the page area identifier where the user interaction event is located in the behavior path sequence, obtain an interaction area node sequence and encode it in chronological order to construct a state set; on the state set, count the actual number of jumps between two states in turn, set the node as the state, the edge as the user jump behavior, and the edge weight as the jump frequency, construct a state jump frequency graph and generate a state transition probability matrix; take the user's current interaction page area as input, perform a multi-step jump simulation on the state transition probability matrix, and output the subsequent interaction probability of multiple page areas respectively; according to the privacy risk level and subsequent interaction probability of the page area, perform weighted calculation to obtain the ranking score of the page area; according to the ranking scores of the multiple page areas, arrange the multiple corresponding policy tasks in descending order to obtain the policy task execution sequence.

[0122] The task execution module is also used to set the number of cacheable tasks according to the current system resource status; using the number of cacheable tasks as the selection number, select multiple policy tasks with the highest ranking from the policy task execution sequence and write the corresponding multiple execution parameters into the front-end cache area; schedule the execution of the cached policy tasks in sequence according to the order of the policy task execution sequence; in response to the completion of the policy task execution, obtain the execution parameters of the next policy task with the highest ranking that has not yet been executed in the policy task execution sequence to replace the cached data of the completed task in the front-end cache area.

[0123] For the description of the features in the embodiments corresponding to the privacy protection device, please refer to the relevant description of the embodiments corresponding to the privacy protection method, and will not be repeated here.

[0124] The embodiment of the present application also provides an electronic device, such as Figure 8 As shown, it includes a memory and a processor, the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any of the above privacy protection method embodiments.

[0125] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above-mentioned privacy protection method embodiments when running.

[0126] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0127] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any of the above-mentioned privacy protection method embodiments are implemented.

[0128] An embodiment of the present application also provides another computer program product, including a non-volatile computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps in any of the above-mentioned privacy protection method embodiments.

[0129] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0130] The above is a detailed introduction to the privacy protection method, device, equipment, storage medium and product provided by this application. This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to this application, and these improvements and modifications also fall within the scope of protection of this application.

Claims

1. A privacy protection method, characterized in that: include: Acquire the user's operation behavior data on the front-end interactive page, and construct a user operation feature vector based on the operation behavior data; Performing a privacy risk assessment operation based on the user operation feature vector to obtain privacy risk levels of multiple page areas of the front-end interactive page; Based on the privacy risk levels of the plurality of page regions, respectively selecting privacy protection policies that match the privacy risk levels from a pre-stored privacy protection policy set to generate a plurality of policy tasks; Executing the corresponding policy task on the user's current interaction page area, and predicting the subsequent interaction probability of the remaining page areas of the multiple page areas except the user's current interaction page area based on the user's historical behavior data, and sorting the multiple policy tasks in combination with the corresponding privacy risk levels to generate a policy task execution sequence, including: obtaining a ranking score for the remaining page areas through weighted calculation based on the privacy risk levels and the subsequent interaction probabilities of the remaining page areas; and arranging the multiple corresponding policy tasks in descending order based on the ranking scores of the remaining page areas to obtain the policy task execution sequence; Based on the execution sequence, execution parameters of a plurality of the policy tasks are pre-cached and executed in sequence.

2. The privacy protection method according to claim 1, characterized in that: The obtaining of the user's operation behavior data on the front-end interactive page and constructing a user operation feature vector based on the operation behavior data includes: Calling a performance interface provided by the front-end interactive page to collect user operation records in multiple page areas to obtain the operation behavior data, the operation behavior data including at least: a screen coordinate value sequence of a click event, an entry time and an exit time of a page area, a scrolling speed and scrolling acceleration of a page scrolling event, and an input character sequence of an input content change event; Obtaining a click position distribution standard deviation based on a sequence of screen coordinate values ​​of the click event; Obtaining an average stay time in the page area according to the entry time and exit time of the page area; Obtaining a scrolling acceleration integral value according to the scrolling speed and scrolling acceleration of the page scrolling event; Calculating weight values ​​of sensitive words in the input text according to the input character sequence of the input content change event; Call the sensor of the device where the front-end interactive page is located to obtain the screen tilt angle and ambient light intensity and generate sensor auxiliary feature values; The standard deviation of the click position distribution, the average dwell time of the page area, the scroll acceleration integral value, the weight value of the sensitive word in the input text, and the sensor auxiliary feature value are spliced ​​to obtain the user operation feature vector.

3. The privacy protection method according to claim 1, wherein: The privacy risk assessment operation is performed based on the user operation feature vector to obtain the privacy risk levels of multiple page areas, including: Dividing the user operation feature vector by page area to obtain multiple behavior segment vectors; Constructing a privacy risk assessment model, the privacy risk assessment model including a temporal feature extraction sub-model based on a long short-term memory network and a reinforcement scoring sub-model based on a reinforcement learning algorithm; Extracting time-dependent features from a plurality of behavior segment vectors through the temporal feature extraction sub-model and outputting an intermediate hidden state sequence; Inputting the intermediate hidden state sequence into the reinforcement scoring sub-model, and outputting the behavior privacy risk score value through the pre-trained state-action value function; The behavior privacy risk score value is compared with a preset privacy risk level range to determine the privacy risk level of the corresponding page area.

4. The privacy protection method according to claim 3, wherein: The step of selecting, based on the privacy risk levels of the plurality of page regions, privacy protection policies that match the privacy risk levels from a pre-stored privacy protection policy set to generate a plurality of policy tasks includes: Set a privacy risk level to correspond to at least one or more privacy protection strategies and generate a gradient strategy table; Obtaining the privacy risk level of the page area, querying the gradient strategy table, and determining a privacy protection strategy corresponding to the privacy risk level of the page area as a target strategy; The page area identifier of the page, the type identifier of the target policy and the execution parameters are combined to construct a corresponding policy task.

5. The privacy protection method according to claim 1, wherein: The predicting of subsequent interaction probabilities of the remaining page areas except the user's current interaction page area in the plurality of page areas based on the user's historical behavior data includes: Extracting the user's historical behavior data to obtain a time series record of user interaction events; Constructing a behavior path sequence based on the time series record of the user interaction events; Extracting the page area identifier where the user interaction event is located in the behavior path sequence, obtaining an interaction area node sequence and encoding it in chronological order to construct a state set; On the state set, the actual jump times between two states are counted in sequence, nodes are set as states, edges are set as user jump behaviors, and edge weights are set as jump frequencies. A state jump frequency graph is constructed and a state transition probability matrix is ​​generated. The user's current interaction page area is taken as input, a multi-step jump simulation is performed on the state transition probability matrix, and subsequent interaction probabilities of the remaining page areas are output respectively.

6. The privacy protection method according to claim 1, wherein: The step of pre-caching execution parameters of a plurality of the policy tasks based on the execution sequence and executing them in sequence includes: Set the number of cacheable tasks based on the current system resource status; Taking the number of cacheable tasks as the selected number, selecting a plurality of the policy tasks ranked first from the policy task execution sequence and writing the corresponding plurality of execution parameters into the front-end cache area; According to the order of the policy task execution sequence, the cached policy tasks are scheduled for execution in sequence; In response to the completion of the execution of the policy task, the execution parameters of the next policy task that has not been executed in the policy task execution sequence and has a higher ranking are obtained to replace the cache data of the completed task in the front-end cache area.

7. A privacy protection device, characterized in that: include: An operation recording module is used to obtain the user's operation behavior data on the front-end interactive page and construct a user operation feature vector based on the operation behavior data; a risk assessment module, configured to perform a privacy risk assessment operation based on the user operation feature vector to obtain privacy risk levels of multiple page areas of the front-end interactive page; a policy task generating module, configured to select, based on the privacy risk levels of the plurality of page regions, privacy protection policies that match the privacy risk levels from a pre-stored privacy protection policy set, and generate a plurality of policy tasks; a task scheduling module, configured to execute the corresponding policy task for the user's current interactive page area, predict the subsequent interaction probability of the remaining page areas other than the user's current interactive page area based on the user's historical behavior data, and sort the plurality of policy tasks in combination with the corresponding privacy risk levels to generate a policy task execution sequence, including: performing a weighted calculation based on the privacy risk levels and subsequent interaction probabilities of the remaining page areas to obtain a sorting score for the remaining page areas; Arrange the corresponding policy tasks in descending order according to the ranking scores of the remaining page areas to obtain the policy task execution sequence; The task execution module is used to pre-cache the execution parameters of multiple strategy tasks based on the execution sequence and execute them in sequence, generate corresponding task execution records and write them into the blockchain network.

8. An electronic device, characterized in that: include: memory for storing computer programs; A processor, configured to implement the steps of the privacy protection method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the privacy protection method according to any one of claims 1 to 6.

Images