An operation data security protection method and system

Abstract

The application is suitable for the technical field of data security, and provides an operation data security protection method and system, including the following steps: receiving an operation data query instruction, calling an encrypted response data set containing original operation data; based on a screen visibility mechanism, performing real-time decryption and block loading on the response data set, performing complete rendering on the data in the screen visible area, and performing fuzzy placeholder rendering on the area outside the screen; identifying sensitive numerical value segments in the visible area, generating a numerical value group, and performing flashing carousel on each numerical value in the numerical value group at the corresponding sensitive numerical value segment; determining sensitive fields in the response data set, replacing the sensitive fields with fictitious data conforming to the original field semantic distribution and statistical characteristics; the replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility maintenance function. The operation personnel can normally perform crowd portrait analysis, trend analysis and other business operations, and the security target of being available but invisible is achieved.

Description

Technical Field

[0001] This invention relates to the field of data security technology, specifically to an operational data security protection method and system. Background Technology

[0002] With the rapid development of the e-commerce industry, e-commerce platforms have accumulated massive amounts of user operation data, including high-value and sensitive information such as user profiles, consumption behavior, transaction records, and delivery addresses. In existing data leakage prevention technology systems, mainstream solutions mainly focus on encryption at the network transmission layer (such as SSL / TLS protocols) and static encryption at the storage layer. However, these technologies cannot effectively solve the security problems during the data usage phase. When authorized operators perform normal data queries and analysis in the backend system, the data must be presented in plaintext form on the user's terminal for reading and operation. This necessary step creates a security vacuum. Traditional data masking techniques typically employ static masking, which involves processing sensitive fields only once when the data leaves the database, such as replacing the middle four digits of a phone number with asterisks or hiding the house number in an address. This static masking scheme has two inherent flaws: first, the masking rules are fixed, and once leaked, attackers can use the comparison of multiple versions of masked data or contextual semantics to perform consistency attacks and restore the original information; second, the masked data destroys the original statistical characteristics, making it impossible for operators to perform effective data analysis and mining, making it difficult to achieve a balance between security and usability. Therefore, there is a need to provide an operational data security protection method and system to solve the above problems. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an operational data security protection method and system to solve the problems existing in the above-mentioned background technology.

[0004] This invention is implemented as follows: a method for protecting operational data security, the method comprising the following steps:

[0005] Receive an operational data query command and retrieve an encrypted response dataset containing the original operational data;

[0006] Based on the screen visibility mechanism, the response dataset is decrypted and loaded in real time and in blocks. The data within the visible area of ​​the screen is fully rendered, while the area outside the screen is rendered with blurred placeholders.

[0007] Identify sensitive value segments within the visible area, generate value groups, and then flash and rotate each value in the value group at the corresponding sensitive value segment at a frequency greater than 60Hz. Stop flashing and rotating when mouse hover is detected at the sensitive value segment.

[0008] Sensitive fields in the response dataset are determined based on user permissions, and these sensitive fields are replaced with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields. The replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function.

[0009] Based on the user ID, timestamp, and session ID, an invisible diffusion watermark is generated and embedded into the background area.

[0010] As a further aspect of the present invention, the steps of real-time decryption and block loading of the response dataset based on the screen visibility mechanism specifically include:

[0011] Get the viewport size and scrollbar position of the browser window, and calculate the range of data row indices and data column indices that can be covered by the visible area of ​​the screen;

[0012] Based on the index range, the encrypted data block within the corresponding index range is extracted from the encrypted response dataset, decrypted, and the complete plaintext data within the visible area is obtained.

[0013] For data rows and columns outside the visible area of ​​the screen, render a preset blurry placeholder at the corresponding DOM node position;

[0014] Monitor the browser's scroll events in real time, and reset the screen's visible area when a change in the scroll bar position is detected.

[0015] As a further aspect of the present invention, the step of identifying sensitive numerical segments within a visible area and generating numerical groups specifically includes:

[0016] Identify the sensitive value segments that need to be flashed in a carousel based on preset sensitive value labels;

[0017] The original values ​​of the sensitive numerical range are determined, and a set of derived values ​​that are mathematically equivalent to the original values ​​are generated by calling the mathematical transformation function library. Each derived value has a different character representation.

[0018] As a further aspect of the present invention, the step of replacing sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields specifically includes:

[0019] Determine the domain constraints for each sensitive field, call the data generator and modify the original data in the sensitive fields based on their respective domain constraints, determine the fictitious data corresponding to each sensitive field, and obtain the fictitious dataset;

[0020] The function value is obtained by iteratively calculating the data utility preservation function by inputting a fictional dataset.

[0021] Determine if the function value of the current fiction dataset is lower than the function threshold. If so, adjust the random seed of the data generator and regenerate the fiction dataset until the function value corresponding to the fiction dataset is higher than the function threshold.

[0022] As a further aspect of the present invention, the formula for the data utility preservation function S is:

[0023]

[0024] in, This represents the original value of the sensitive field in the i-th record. This represents the fictitious data in the i-th row after replacement, and n represents the total number of rows with sensitive fields; X is the preset semantic feature extraction function; X is the feature matrix of the original dataset. For the feature matrix of a fictional dataset, Cov( ) is the function for calculating the covariance matrix; α represents the Frobenius norm of the matrix; α and β are preset weight coefficients, α+β=1.

[0025] As a further aspect of the present invention, the step of generating invisible diffusion watermark information based on user ID, timestamp, and session ID, and embedding the diffusion watermark information into the background area specifically includes:

[0026] The user ID, timestamp, and session ID are concatenated into the original watermark string. Error correction encoding and spread spectrum modulation are then applied to the original watermark string to generate a watermark signal sequence.

[0027] By iterating through each pixel in the background area, the watermark signal sequence is modulated into the blue channel or alpha transparency channel of the pixel at an amplitude lower than the human eye's contrast sensitivity threshold, forming a background noise watermark layer that is invisible to the human eye.

[0028] Another object of the present invention is to provide an operational data security protection system, the system comprising:

[0029] The operations data retrieval module is used to receive operations data query commands and retrieve an encrypted response dataset containing the original operations data.

[0030] The chunked loading and rendering module is used to decrypt and load the response dataset in real time based on the screen visibility mechanism, to fully render the data within the visible area of ​​the screen, and to render the data outside the screen using blurred placeholders.

[0031] The sensitive value carousel module is used to identify sensitive value segments within the visible area, generate value groups, and then flash and carousel each value in the value group at the corresponding sensitive value segment. The flashing and carousel frequency is greater than 60Hz. When mouse hover is detected at a sensitive value segment, the flashing and carousel stops.

[0032] The sensitive field replacement module is used to determine the sensitive fields in the response dataset based on user permissions, and replace the sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields; the replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function;

[0033] The watermark information embedding module is used to generate invisible diffusion watermark information based on user ID, timestamp, and session ID, and embed the diffusion watermark information into the background area.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] Employing a block loading and real-time decryption technology based on screen visibility, this method decrypts and renders only data blocks within the currently visible screen area, while rendering only blurred placeholders for areas outside the screen. This strictly limits the amount of plaintext data residing in memory to the visible range. Even if the terminal device is infected with malicious Trojans or suffers a memory dump attack, attackers can only obtain fragmented data pieces currently visible on the screen, and cannot obtain the complete query result set.

[0036] For core sensitive values, a high-frequency flickering carousel mechanism is introduced. By cyclically displaying multiple mathematically equivalent derived values ​​on the screen at a frequency exceeding 60Hz, the exact plaintext numbers cannot be captured by the naked eye or a camera in a single screenshot. When the user needs to view them in detail, the flickering stops by hovering the mouse over them, making it easier to read. This achieves a balance between security and usability.

[0037] By replacing the original sensitive fields with fictitious data that conforms to the original semantic distribution and statistical characteristics, and through iterative optimization using a data utility preservation function, the generated fictitious data is ensured to maintain a high degree of consistency with the original data at the macro-statistical analysis level. This allows operations personnel to perform normal business operations such as user profiling and trend analysis, while actually interacting with fictitious data that cannot be mapped to real individuals, achieving the security goal of being usable but invisible. Attached Figure Description

[0038] Figure 1 This is a flowchart of an operational data security protection method.

[0039] Figure 2 This is a flowchart illustrating real-time decryption and chunked loading in an operational data security protection method.

[0040] Figure 3 This is a flowchart illustrating the generation of numerical groups in an operational data security protection method.

[0041] Figure 4 This is a flowchart illustrating how sensitive fields are replaced with fictitious data in an operational data security protection method.

[0042] Figure 5 This is a flowchart illustrating how to embed diffusion watermark information into a background area in an operational data security protection method.

[0043] Figure 6 This is a schematic diagram of the structure of an operational data security protection system. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0045] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0046] like Figure 1 As shown in the figure, this embodiment of the invention provides a method for protecting operational data security, the method comprising the following steps:

[0047] S100 receives an operational data query command and retrieves an encrypted response dataset containing the original operational data.

[0048] S200 performs real-time decryption and chunk loading of the response dataset based on the screen visibility mechanism, fully renders the data within the visible area of ​​the screen, and renders the data outside the screen using blurred placeholders.

[0049] S300 identifies sensitive value segments within the visible area, generates value groups, and performs a blinking rotation of each value in the value group at the corresponding sensitive value segment with a blinking rotation frequency greater than 60Hz; when a mouse hover is detected at a sensitive value segment, the blinking rotation stops.

[0050] S400: Based on user permissions, determine the sensitive fields in the response dataset and replace them with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields; the replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function;

[0051] The S500 generates an invisible diffusion watermark based on the user ID, timestamp, and session ID, and embeds the diffusion watermark into the background area.

[0052] In this embodiment of the invention, when an operator performs a query operation on the backend interface, the frontend system sends the query command to the backend. The backend retrieves the corresponding raw operational data from the database based on the query conditions and encrypts it before returning it to the frontend, forming an encrypted response dataset. This ensures data security during transmission and prevents man-in-the-middle attacks or network sniffing. Upon receiving the encrypted data, the frontend does not decrypt all data at once. Instead, it decrypts and loads only the data blocks within the visible area of ​​the current screen for complete rendering. For areas outside the screen that have not yet been scrolled to, only blurred placeholders are rendered or the encrypted form is retained. This mechanism significantly reduces the amount of plaintext data residing in memory, preventing malicious programs from stealing the complete dataset through memory dumps. After rendering the visible area, highly sensitive numerical fields such as amounts and points are automatically identified. For these fields, a set of mathematically equivalent derived values ​​is generated and displayed in a loop at a frequency higher than the human visual persistence limit (greater than 60Hz). This high-frequency flickering makes it impossible for the naked eye or a camera to capture the exact plaintext numbers in a single screenshot. When a user needs to view the data more closely, hovering the mouse stops the flickering, facilitating reading and achieving a dynamic balance between security and usability. Furthermore, this embodiment of the invention determines sensitive fields in the response dataset based on user permissions. Different permissions correspond to different sensitive fields, and this mapping needs to be pre-set. For highly sensitive fields beyond the current user's viewing permissions (such as detailed addresses and phone numbers), the system does not return the original data. Instead, it generates a batch of fictitious data that is highly consistent with the original data semantically (e.g., the address remains in the same city) and statistically (e.g., the geographical proportion remains unchanged). The replacement process is not randomly generated but iteratively optimized using a utility function to ensure that the conclusions drawn from the generated fake data are essentially consistent with those drawn from the real data when used for statistical analysis. Finally, the unique identifier of the current user, the operation time, and session information are encoded into an invisible watermark and diffused and embedded into the background area of ​​the page using image processing technology. This ensures that any page images leaked through screenshots, photos, etc., carry the leaker's identity information, facilitating subsequent tracing.

[0053] like Figure 2 As shown, in a preferred embodiment of the present invention, the steps of real-time decryption and chunked loading of the response dataset based on the screen visibility mechanism specifically include:

[0054] S201, obtain the viewport size and scrollbar position of the browser window, and calculate the data row index range and data column index range that can be covered by the visible area of ​​the screen;

[0055] S202, extract the encrypted data block within the corresponding index range from the encrypted response dataset according to the index range, decrypt it, and obtain the complete plaintext data within the visible area;

[0056] S203 renders a preset blurry placeholder at the corresponding DOM node position for data rows and columns outside the visible area of ​​the screen;

[0057] S204 monitors the browser's scroll events in real time and resets the visible area of ​​the screen when a change in the scroll bar position is detected.

[0058] In this embodiment of the invention, the front-end rendering engine first obtains the height and width of the current window's visible area (viewport size) and the distance of the current scrollbar from the top and left (scrollbar position) through the browser API. Combining this with predefined row heights and column widths in the data table, it uses geometric calculations to accurately determine which rows (e.g., rows 10 to 25) and columns (e.g., columns 1 to 5) are visible on the current screen, generating a data index range for the visible area. Then, based on the calculated row and column index ranges, it locates the corresponding portion of the encrypted response dataset in local memory. Only these hit data blocks are decrypted using the decryption algorithm, restoring them to plaintext data, which exists only in the memory space corresponding to the current visible area. For data cells outside the current visible range, the rendering engine does not leave them blank but actively renders specific alternative content, which can be color blocks processed with Gaussian blur. This serves two purposes: firstly, visually informing the user that there is data below, and secondly, preventing accidental exposure of data in invisible areas due to rendering engine malfunctions or DOM structure leaks. In addition, once a scrolling event is detected, the current rendering state is immediately paused, and steps S201-S203 above are repeated. That is, the visible area index at the new position is recalculated, the encrypted data block of the newly entered area is decrypted and rendered, and the memory of the data that scrolls out of the area is reclaimed.

[0059] like Figure 3 As shown, in a preferred embodiment of the present invention, the step of identifying sensitive numerical segments within a visible area and generating numerical groups specifically includes:

[0060] S301, Identify the sensitive value segments that need to be flashed and rotated based on the preset sensitive value labels;

[0061] S302, determine the original value of the sensitive value segment, call the mathematical transformation function library, and generate a set of derived values ​​that are mathematically equivalent to the original value, with each derived value having a different character representation.

[0062] In this embodiment of the invention, fields belonging to sensitive numerical tags are predefined in the data dictionary or configuration file, such as order amount, member points, and cumulative consumption. After the visible area is rendered, the front-end rendering engine traverses each data cell in the current screen and matches its field names with the preset list of sensitive tags. Cells that match successfully are marked as target objects for the blinking carousel. For each marked sensitive cell, its underlying original value is obtained, for example, the original value is 12800.50. Subsequently, the system calls a built-in mathematical transformation function library, which contains various equivalent transformation algorithms, such as additive factorization (12799+1.50), multiplicative factorization (64×200.0078125), unit conversion (12.8005k), and approximate rounding (12800). For the same original value, the function library generates a set of derived values ​​with different character lengths, different operators, and different decimal places, ensuring that they are visually significantly different, but mathematically equivalent to the original value or equivalent within an allowable small error range.

[0063] like Figure 4 As shown, in a preferred embodiment of the present invention, the step of replacing sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields specifically includes:

[0064] S401, determine the domain constraints of each sensitive field, call the data generator and modify the original data in the sensitive fields based on their respective domain constraints, determine the fictitious data corresponding to each sensitive field, and obtain the fictitious dataset;

[0065] S402, Input the fictional dataset into the data utility preservation function and perform iterative calculations to obtain the function value;

[0066] S403, determine whether the function value of the current fiction dataset is lower than the function threshold. If so, adjust the random seed of the data generator and regenerate the fiction dataset until the function value corresponding to the fiction dataset is higher than the function threshold.

[0067] In the embodiments of the present invention, it is necessary to determine the data characteristics of each sensitive field and determine its domain constraints. For example, for the "receiving address" field, the domain constraints include: it must conform to the format of "province / city / district / street", the combination of province, city and district must actually exist, and the address code must be within a specific range. For the "mobile phone number" field, the domain constraints include: it must be 11 digits, and the first three digits must be valid operator number segments. Based on these constraints, a fake data generator (such as an address library random sampler, a mobile phone number generator) is called to generate a corresponding value that conforms to the constraints but is completely fictitious for each record in the original data set, forming an initial fictitious data set. For example, the original address "No. 78, College Road, B District, A City" will be replaced with a fake address "No. 42, Faculty Road, B District, A City" in the same city and district, and the original mobile phone number will be replaced with another compliant fake number with the same operator in the same place of registration. Then, the original data set and the fictitious data set generated in the previous step are used as inputs and substituted into the data utility preservation function S for calculation to obtain a function value. Here, an acceptable utility threshold is preset, and the calculated function value S is compared with the threshold T. If S < T, it means that the currently generated fictitious data set deviates too much from the original data semantically or statistically and cannot be used for substitution analysis. At this time, the random seed of the fake data generator will be adjusted (for example, using a different address library random strategy or fine-tuning the replacement algorithm) to generate a new batch of fictitious data, and the value of S will be calculated again. This process is iterated until a certain generated fictitious data set makes S ≥ T. At this time, the system considers that this data set meets the requirements in terms of utility and uses it as the final result for replacement.

[0068] In the embodiments of the present invention, the formula of the data utility preservation function S is:

[0069]

[0070] Wherein, represents the original value of the sensitive field of the i-th item, represents the corresponding fictitious data of the i-th item after replacement, and n represents the total number of sensitive fields. is a preset semantic feature extraction function, whose function is to extract key semantic labels from complex original values. For example, for an address string, extracts the "municipal administrative division code"; for an amount, extracts the "consumption level (high / medium / low)". calculates the absolute value of the difference in key semantics of a single record. The smaller this value is, the higher the semantic replacement accuracy of the single record. X is the feature matrix of the original data set. Each row of the matrix represents a field record, and each column represents a statistical feature (such as consumption amount, purchase frequency). is the feature matrix of the fictitious data set, Cov( ) is the covariance matrix calculation function. It reflects the correlation structure between various features in the original data (for example, the higher the consumption amount, the higher the membership level tends to be). This refers to the correlation structure in fictitious data. The Frobenius norm of a matrix can be understood as a scalar value that measures the difference between two matrices. This measure assesses the overall deviation of the fictional dataset from the original dataset in terms of statistical structure. A smaller value indicates that the fictional data is closer to reality in macroscopic statistics. α and β are preset weighting coefficients, α + β = 1, used to adjust the relative importance of individual semantic fidelity and overall statistical fidelity in the total utility evaluation. For example, α might be increased when performing individual user profile analysis, while β might be increased when performing macroscopic market trend analysis.

[0071] like Figure 5 As shown, in a preferred embodiment of the present invention, the step of generating invisible diffusion watermark information based on user ID, timestamp, and session ID, and embedding the diffusion watermark information into the background area specifically includes:

[0072] S501 concatenates the user ID, timestamp, and session ID into the original watermark string, performs error correction coding and spread spectrum modulation on the original watermark string, and generates a watermark signal sequence.

[0073] S502, traverse each pixel in the background area, and modulate the watermark signal sequence into the blue channel or alpha transparency channel of the pixel at an amplitude lower than the human eye contrast sensitivity threshold to form a background noise watermark layer invisible to the human eye.

[0074] In this embodiment of the invention, the unique identifier information of the current operation (user ID, timestamp accurate to milliseconds, session ID) is combined into a string (e.g., U123_20231027143022123_SESSC) according to a predetermined format. This string is then processed in two ways: first, error correction encoding, such as Reed-Solomon coding, adds redundant information so that the watermark can still be recovered after partial damage (e.g., cropping a screenshot); second, spread spectrum modulation expands the encoded data into a long pseudo-random sequence, improving the watermark's concealment and anti-interference ability, ultimately generating a watermark signal sequence that can be embedded. Then, the background layer of the page is drawn in the front-end rendering engine, traversing this background area pixel by pixel. For each pixel, based on the current bit value (0 or 1) in the watermark signal sequence, the blue channel value (B value) or alpha transparency channel value (transparency) of that pixel is adjusted very slightly. The degree of this adjustment is strictly controlled to a level imperceptible to the naked eye (usually no more than ±2 gray levels, or a change in transparency of less than 1%), but it is detectable to machine vision algorithms. After modulating each background pixel individually, the entire background layer becomes a noise layer carrying watermark information, which is overlaid on the foreground text and charts. Users cannot perceive anything unusual, but any screenshot will inevitably contain this unique identity fingerprint.

[0075] like Figure 6 As shown in the figure, this embodiment of the invention also provides an operational data security protection system, the system comprising:

[0076] The operation data retrieval module 100 is used to receive operation data query instructions and retrieve an encrypted response dataset containing the original operation data.

[0077] The chunked loading and rendering module 200 is used to decrypt and load the response dataset in real time based on the screen visibility mechanism, to fully render the data within the visible area of ​​the screen, and to render the data outside the screen using fuzzy placeholders.

[0078] The sensitive value carousel module 300 is used to identify sensitive value segments within the visible area, generate value groups, and perform a blinking carousel of each value in the value group at the corresponding sensitive value segment. The blinking carousel frequency is greater than 60Hz. When a mouse hover is detected at a sensitive value segment, the blinking carousel stops.

[0079] The sensitive field replacement module 400 is used to determine the sensitive fields in the response dataset based on user permissions, and replace the sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields; the replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function;

[0080] The watermark information embedding module 500 is used to generate invisible diffusion watermark information based on user ID, timestamp and session ID, and embed the diffusion watermark information into the background area.

[0081] In a preferred embodiment of the present invention, the block loading and rendering module 200 includes:

[0082] The data coverage determination unit is used to obtain the viewport size and scrollbar position of the browser window, and calculate the data row index range and data column index range that can be covered by the visible area of ​​the screen;

[0083] The visible area decryption unit is used to extract the encrypted data block within the corresponding index range from the encrypted response dataset according to the index range, decrypt it, and obtain the complete plaintext data within the visible area;

[0084] The blur rendering unit is used to render preset blur placeholders at the corresponding DOM node positions for data rows and columns outside the visible area of ​​the screen.

[0085] The visible area reset unit is used to listen for browser scroll events in real time and reset the screen visible area when a change in the scroll bar position is detected.

[0086] In a preferred embodiment of the present invention, the sensitive value carousel module 300 includes:

[0087] The sensitive value segment unit is used to identify the sensitive value segments that need to be flashed and rotated based on preset sensitive value labels;

[0088] The derived value determination unit is used to determine the original value of the sensitive value segment, call the mathematical transformation function library, and generate a set of derived values ​​that are mathematically equivalent to the original value. The character representation of each derived value is different.

[0089] In a preferred embodiment of the present invention, the sensitive field replacement module 400 includes:

[0090] The fictional data determination unit is used to determine the domain constraints of each sensitive field, call the data generator and modify the original data in the sensitive fields based on their respective domain constraints, determine the fictional data corresponding to each sensitive field, and obtain the fictional dataset.

[0091] The function value calculation unit is used to input a hypothetical dataset into a data utility preservation function and perform iterative calculations to obtain the function value.

[0092] The fictional data reset unit is used to determine whether the function value of the current fictional dataset is lower than the function threshold. If so, the random seed of the data generator is adjusted, and the fictional dataset is regenerated until the function value corresponding to the fictional dataset is higher than the function threshold.

[0093] The above description only details the preferred embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0094] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0095] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the disclosure in the specification and embodiments. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A method for protecting operational data security, characterized in that, The method includes the following steps: Receive an operational data query command and retrieve an encrypted response dataset containing the original operational data; Based on the screen visibility mechanism, the response dataset is decrypted and loaded in real time and in blocks. The data within the visible area of ​​the screen is fully rendered, while the area outside the screen is rendered with blurred placeholders. Identify sensitive value segments within the visible area, generate value groups, and then flash and rotate each value in the value group at the corresponding sensitive value segment at a frequency greater than 60Hz. Stop flashing and rotating when mouse hover is detected at the sensitive value segment. Sensitive fields in the response dataset are determined based on user permissions, and these sensitive fields are replaced with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields. The replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function. Based on the user ID, timestamp, and session ID, generate an invisible diffusion watermark and embed the diffusion watermark into the background area. The steps for real-time decryption and chunked loading of the response dataset based on screen visibility mechanisms specifically include: Get the viewport size and scrollbar position of the browser window, and calculate the range of data row indices and data column indices that can be covered by the visible area of ​​the screen; Based on the index range, the encrypted data block within the corresponding index range is extracted from the encrypted response dataset, decrypted, and the complete plaintext data within the visible area is obtained. For data rows and columns outside the visible area of ​​the screen, render a preset blurry placeholder at the corresponding DOM node position; Monitor the browser's scroll events in real time, and reset the screen's visible area when a change in the scroll bar position is detected. The steps of replacing sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields specifically include: Determine the domain constraints for each sensitive field, call the data generator and modify the original data in the sensitive fields based on their respective domain constraints, determine the fictitious data corresponding to each sensitive field, and obtain the fictitious dataset; The function value is obtained by iteratively calculating the data utility preservation function by inputting a fictional dataset. Determine if the function value of the current fiction dataset is lower than the function threshold. If so, adjust the random seed of the data generator and regenerate the fiction dataset until the function value corresponding to the fiction dataset is higher than the function threshold. The formula for the data utility preservation function S is: ;in, This represents the original value of the sensitive field in the i-th record. This represents the fictitious data in the i-th row after replacement, and n represents the total number of rows with sensitive fields; X is a predefined semantic feature extraction function; X is the feature matrix of the original dataset. For the feature matrix of a fictional dataset, Cov( ) is the function for calculating the covariance matrix; α represents the Frobenius norm of the matrix; α and β are preset weighting coefficients, α+β=1.

2. The operational data security protection method according to claim 1, characterized in that, The steps for identifying sensitive numerical segments within the visible area and generating numerical groups specifically include: Identify the sensitive value segments that need to be flashed in a carousel based on preset sensitive value labels; The original values ​​of the sensitive numerical range are determined, and a set of derived values ​​that are mathematically equivalent to the original values ​​are generated by calling the mathematical transformation function library. Each derived value has a different character representation.

3. The operational data security protection method according to claim 1, characterized in that, The steps of generating an invisible diffusion watermark based on the user ID, timestamp, and session ID, and embedding the diffusion watermark into the background area, specifically include: The user ID, timestamp, and session ID are concatenated into the original watermark string. Error correction encoding and spread spectrum modulation are then applied to the original watermark string to generate a watermark signal sequence. By iterating through each pixel in the background area, the watermark signal sequence is modulated into the blue channel or alpha transparency channel of the pixel at an amplitude lower than the human eye's contrast sensitivity threshold, forming a background noise watermark layer that is invisible to the human eye.

4. An operational data security protection system, characterized in that, The system includes: The operations data retrieval module is used to receive operations data query commands and retrieve an encrypted response dataset containing the original operations data. The chunked loading and rendering module is used to decrypt and load the response dataset in real time based on the screen visibility mechanism, to fully render the data within the visible area of ​​the screen, and to render the data outside the screen using blurred placeholders. The sensitive value carousel module is used to identify sensitive value segments within the visible area, generate value groups, and then flash and carousel each value in the value group at the corresponding sensitive value segment. The flashing and carousel frequency is greater than 60Hz. When mouse hover is detected at a sensitive value segment, the flashing and carousel stops. The sensitive field replacement module is used to determine the sensitive fields in the response dataset based on user permissions, and replace the sensitive fields with fictitious data that conforms to the semantic distribution and statistical characteristics of the original fields; the replacement process iteratively calculates the optimal replacement value by maximizing the value of the data utility preservation function; The watermark information embedding module is used to generate invisible diffusion watermark information based on user ID, timestamp, and session ID, and embed the diffusion watermark information into the background area. The block loading and rendering module includes: The data coverage determination unit is used to obtain the viewport size and scrollbar position of the browser window, and calculate the data row index range and data column index range that can be covered by the visible area of ​​the screen; The visible area decryption unit is used to extract the encrypted data block within the corresponding index range from the encrypted response dataset according to the index range, decrypt it, and obtain the complete plaintext data within the visible area; The blur rendering unit is used to render preset blur placeholders at the corresponding DOM node positions for data rows and columns outside the visible area of ​​the screen. The visible area reset unit is used to listen to the browser's scroll events in real time and reset the screen's visible area when a change in the scroll bar position is detected. The sensitive field replacement module includes: The fictional data determination unit is used to determine the domain constraints of each sensitive field, call the data generator and modify the original data in the sensitive fields based on their respective domain constraints, determine the fictional data corresponding to each sensitive field, and obtain the fictional dataset. The function value calculation unit is used to input a hypothetical dataset into a data utility preservation function and perform iterative calculations to obtain the function value. The fictional data reset unit is used to determine whether the function value of the current fictional dataset is lower than the function threshold. If so, the random seed of the data generator is adjusted and the fictional dataset is regenerated until the function value corresponding to the fictional dataset is higher than the function threshold. The formula for the data utility preservation function S is: ;in, This represents the original value of the sensitive field in the i-th record. This represents the fictitious data in the i-th row after replacement, and n represents the total number of rows with sensitive fields; X is a predefined semantic feature extraction function; X is the feature matrix of the original dataset. For the feature matrix of a fictional dataset, Cov( ) is the function for calculating the covariance matrix; α represents the Frobenius norm of the matrix; α and β are preset weighting coefficients, α+β=1.

5. The operational data security protection system according to claim 4, characterized in that, The sensitive value carousel module includes: The sensitive value segment unit is used to identify the sensitive value segments that need to be flashed and rotated based on preset sensitive value labels; The derived value determination unit is used to determine the original value of the sensitive value segment, call the mathematical transformation function library, and generate a set of derived values ​​that are mathematically equivalent to the original value. The character representation of each derived value is different.

Images