Methods and Systems for Verifying the Integrity of Approval Data in a Trusted Data Space

Abstract

This invention discloses a method and system for verifying the integrity of approval data in a trusted data space, relating to the field of data security technology. The method includes: assessing and determining an approval risk coefficient and an approval urgency index based on the data attribute information and approval-related characteristics of the data to be approved; calculating and obtaining an appropriate approval time limit based on the approval risk coefficient and approval urgency index; using a time limit shorter than the appropriate approval time limit as an efficiency constraint to optimize the initial approval strategy, selecting the combination of approval nodes and the combination of approval node verification depths corresponding to the minimum predicted approval error probability as the appropriate approval scheme; and performing integrity verification on the data to be approved within the trusted data space according to the appropriate approval scheme. This invention effectively improves the integrity, reliability, and processing efficiency of the entire approval data chain in a trusted data space.

Description

Technical Field

[0001] This invention relates to the field of data security technology, specifically to a method and system for verifying the integrity of approval data in a trusted data space. Background Technology

[0002] With the widespread application of trusted data spaces in various fields, the volume and complexity of data awaiting approval continue to rise, placing dual demands on the efficiency and accuracy of data integrity verification.

[0003] However, existing technologies mostly adopt static and fixed approval strategies, which make it difficult to match verification resources with dynamic risks and to achieve a balance between approval efficiency and verification quality. This leads to an imbalance in the allocation of verification resources. Over-verification can easily cause resource waste and inefficiency, while insufficient verification will affect the accuracy of approval data and fail to meet the needs of efficient and accurate verification.

[0004] Therefore, there is an urgent need for a new method that can intelligently balance approval efficiency and accuracy and achieve dynamic optimization of verification resources. Summary of the Invention

[0005] This invention provides a method and system for verifying the integrity of approval data in a trusted data space, aiming to solve the technical problem that existing technologies cannot meet the requirements for efficient and accurate verification.

[0006] In view of the above problems, the present invention provides a method and system for verifying the integrity of approval data in a trusted data space.

[0007] In a first aspect, the present invention provides a method for verifying the integrity of approval data in a trusted data space, including:

[0008] Based on the data attribute information and approval-related characteristics of the data to be approved, the approval risk coefficient and approval urgency index are determined.

[0009] The appropriate approval time limit is calculated based on the aforementioned approval risk coefficient and approval urgency index.

[0010] Using the time limit less than the adaptive approval time limit as an efficiency constraint, the initial approval strategy is optimized, and the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability are selected as the adaptive approval scheme.

[0011] Within the trusted data space, the integrity of the data to be approved is verified according to the adapted approval scheme.

[0012] Secondly, this invention provides a system for verifying the integrity of approval data in a trusted data space, comprising:

[0013] The risk urgency assessment module is used to assess and determine the approval risk coefficient and approval urgency index based on the data attribute information and approval-related characteristics of the data to be approved.

[0014] The adaptation time limit calculation module is used to calculate and obtain the adaptation approval time limit based on the approval risk coefficient and the approval urgency index.

[0015] The approval strategy optimization module is used to optimize the initial approval strategy by taking the time limit less than the adapted approval time limit as an efficiency constraint, and select the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability as the adapted approval scheme.

[0016] The data integrity verification module is used to perform integrity verification on the data to be approved in accordance with the adapted approval scheme within the trusted data space.

[0017] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0018] This invention provides a method and system for verifying the integrity of approval data in a trusted data space. Dynamic risk assessment based on data attributes and correlation characteristics provides a precise basis for resource allocation. An adaptation time limit is calculated according to the degree of risk and urgency, setting constraints that balance security and efficiency. The approval strategy is optimized using this time limit as a constraint, intelligently selecting the optimal combination of nodes and verification depth to achieve optimal resource allocation. The adaptation scheme is executed within the trusted space to ensure the reliability of the verification process. This invention optimizes the approval process and data verification depth while ensuring approval efficiency, achieving optimal allocation of verification resources, thereby effectively improving approval quality and achieving efficient and accurate approval data verification. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the method for verifying the integrity of approval data in a trusted data space provided in an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of the approval data integrity verification system in the trusted data space provided in an embodiment of the present invention;

[0022] The components represented by each number in the attached diagram are explained below:

[0023] Risk emergency assessment module 11, adaptation time limit calculation module 12, approval strategy optimization module 13, data integrity verification module 14. Detailed Implementation

[0024] This invention provides a method and system for verifying the integrity of approval data in a trusted data space, which addresses the technical problem that existing technologies cannot meet the requirements for efficient and accurate verification.

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0027] Example 1, as Figure 1 As shown, this invention provides a method for verifying the integrity of approval data in a trusted data space, the method comprising:

[0028] S100: Based on the data attribute information and approval-related characteristics of the data to be approved, determine the approval risk coefficient and approval urgency index.

[0029] In this embodiment of the invention, the approval risk coefficient and approval urgency index are determined based on the data attribute information and approval-related characteristics of the data to be approved. In existing approval scenarios, risk and urgency assessments often rely on manual judgment or single indicators, resulting in significant subjective bias and poor adaptability. When encountering situations with complex data attributes and diverse related business constraints, traditional assessment methods cannot accurately match the actual scenario requirements, easily leading to unreasonable subsequent approval timelines and ineffective solution optimization. Therefore, it is necessary to use a historical data-driven intelligent model, combined with multi-dimensional data attributes and approval-related characteristics, to achieve accurate quantitative assessment of the risk coefficient and approval urgency index.

[0030] Step S100 in the method provided in this embodiment of the invention includes:

[0031] Based on historical approval data verification records, sample data attribute information set and sample approval related feature set are collected, and the historical approval risk coefficient and historical approval urgency index corresponding to different sample data attribute information and sample approval related features are labeled to obtain sample approval risk coefficient set and sample approval urgency index set.

[0032] Using the sample data attribute information set and sample approval related feature set as input, and the sample approval risk coefficient set and sample approval urgency index set as supervision, a generative adversarial network is trained until both the generator and discriminator converge, thus constructing an approval feature recognition engine.

[0033] Using the aforementioned approval feature recognition engine, the approval risk coefficient and approval urgency index are determined based on the data attribute information and approval association features of the data to be approved.

[0034] The data attribute information includes at least the data source credibility, data sensitivity level, data complexity, data volume, and historical error rate of similar data. The approval association features include at least the priority of related business and the supervision time limit of related business. The approval urgency index is positively correlated with the priority of related business and negatively correlated with the supervision time limit of related business.

[0035] The approval risk coefficient is positively correlated with the data sensitivity level, data complexity, data volume, and historical error rate of similar data, and negatively correlated with the credibility of the data source.

[0036] First, based on historical approval data verification records, sample data attribute information sets and sample approval related feature sets are collected. Then, historical approval risk coefficients and historical approval urgency indices are labeled for different scenarios of sample data attribute information and sample approval related features, obtaining sample approval risk coefficient sets and sample approval urgency index sets. The sample data attribute information set refers to the set extracted from historical approval data verification records that describes the characteristics of the data itself, including data source credibility, data sensitivity level, data complexity, data volume, and historical error rate of similar data. Data source credibility refers to the qualification and compliance score of the data provider; data sensitivity level refers to the degree of loss caused by data leakage or errors, divided into three levels; data complexity refers to the complexity of data structure and logic, divided into high, medium, and low; data volume refers to the size of data storage; and historical error rate of similar data refers to the proportion of integrity problems that occurred in past verifications of similar approval data. The sample approval related feature set refers to the set of constraint features related to approval business extracted from historical approval data, including related business priority and related business supervision time limit. Related business priority refers to the degree of impact of the approval result on subsequent business, categorized as high, medium, and low; related business supervision time limit refers to the maximum time required by the regulatory authority to complete the approval. Sample approval risk coefficient set / urgency index set refers to the quantitative values ​​marked by experts on historical approval scenarios, ranging from 0 to 1, with higher values ​​indicating higher risk / urgency.

[0037] For example, taking the online multi-party rights confirmation and approval process in the automotive parts supply chain as a scenario, and focusing on the rights confirmation and approval of core component drawings for the braking system: 1000 sets of historical verification records for rights confirmation and approval from the past three years were collected from the automotive parts supply chain, including engine drawings, transmission assembly parameters, and braking system drawings. Taking one set of rights confirmation and approval as an example, the following data attributes were extracted: Data source credibility: supplier qualification score 88 points; Data sensitivity level: Level 1, involving core safety components; Data complexity: Medium, containing 3 sets of assembly diagrams; Data size: 45MB; Historical error rate of similar data: 1.5%. The following sample approval correlation features were extracted: Corresponding business priority: High; Corresponding business supervision time limit: 18 days. Experts labeled the historical approval risk coefficient (0.75) and historical approval urgency index (0.82) based on the historical verification results of this approval. This operation was repeated to form a set of sample data attribute information, a set of sample approval correlation features, a set of sample approval risk coefficients, and a set of sample approval urgency indices corresponding to 1000 samples.

[0038] Secondly, using the sample data attribute information set and sample approval-related feature set as input, and the sample approval risk coefficient set and sample approval urgency index set as supervision, a generative adversarial network (GAN) is trained until both the generator and discriminator converge, thus constructing an approval feature recognition engine. The GAN is a machine learning model composed of two sub-networks: a generator and a discriminator. Prediction accuracy is improved through adversarial training between the generator and discriminator. The generator learns the sample patterns and outputs prediction results, while the discriminator distinguishes the prediction results from the actual sample labels. This process iterates until the performance of the generator and discriminator converges. The approval feature recognition engine refers to an intelligent model that, after training with the GAN, can output accurate risk coefficients and urgency indices from the input attribute information and related features of the data to be approved.

[0039] For example, the sample data attribute information set and sample approval-related feature set of the aforementioned 1000 samples are used as model input, and the sample approval risk coefficient set and sample approval urgency index set are used as real supervision labels to train a generative adversarial network. During training, the generator predicts the risk coefficient and urgency index based on the input features. For example, for a certain gearbox drawing approval sample, the generator outputs a risk coefficient of 0.68 and an urgency index of 0.75. The discriminator compares this predicted value with the real labeled values ​​of 0.70 and 0.78, and provides feedback on the error, driving the generator to adjust its parameters. After 600 rounds of continuous iterative training, the generator's prediction error is less than 3%, and the discriminator's discrimination accuracy stabilizes above 95%, both reaching convergence. At this point, the construction of the approval feature recognition engine is complete.

[0040] Furthermore, the approval feature recognition engine is used to assess and determine the approval risk coefficient and approval urgency index based on the data attribute information and approval correlation characteristics of the data to be approved. The data attribute information and approval correlation characteristics of the data to be approved refer to the inherent characteristics and business correlation constraints of the specific data currently requiring confirmation of rights approval. For example, a multi-party online confirmation of rights approval request submitted by a supplier for hydraulic pipeline drawings of a new car's brake system is the data to be approved. First, the attribute information of this data is extracted: data source credibility: qualification score 92 points; data sensitivity level: level one, involving brake safety; data complexity: medium, containing 2 sets of pipeline layout diagrams; data size: 35MB; historical error rate of similar data: 0.8%. Approval correlation characteristics are extracted: related business priority: medium, corresponding to the mass production of a new car facelift, not a first-launch model; related business supervision time limit: 25 days. These characteristics are input into the constructed approval feature recognition engine. Based on learned historical patterns, the engine outputs an approval risk coefficient of 0.62 and an approval urgency index of 0.65 for the data to be approved, completing the assessment.

[0041] In this embodiment of the invention, a closed-loop process of historical sample collection and annotation, generative adversarial network training, and target data feature evaluation is employed. This process leverages comprehensive coverage of data attribute information and approval-related features, combined with the strong fitting ability of generative adversarial networks to complex scenario patterns, to avoid subjective biases in manual evaluation. This makes the evaluation results of approval risk coefficients and approval urgency indices more closely aligned with actual scenario needs, resulting in higher accuracy and reliability. Simultaneously, the data's inherent characteristics and related business constraints are transformed into quantifiable evaluation indicators, providing objective and scenario-appropriate core inputs for subsequent accurate calculation of approval time limits and scientific optimization of approval schemes. This lays a solid foundation for the efficient advancement of the entire approval data integrity verification process.

[0042] S200: Calculate and obtain the appropriate approval time limit based on the approval risk coefficient and approval urgency index.

[0043] In this embodiment of the invention, an appropriate approval time limit is calculated based on the approval risk coefficient and the approval urgency index. The approval time limit is a core constraint for balancing the review time required for risk control with the timeliness required for business urgency. Setting a time limit solely based on the risk coefficient ignores the time requirements of the business, while setting a time limit solely based on the urgency index amplifies the risk. Existing technologies often use fixed approval time limits, which cannot adapt to the differences in risk and urgency of different approval items. In scenarios such as multi-party authorization approvals in the automotive parts supply chain, problems such as insufficient review of high-risk items or redundant time consumption for low-urgency items easily arise. Therefore, it is necessary to dynamically calculate the appropriate approval time limit based on the risk coefficient and urgency index of S100.

[0044] Step S200 in the method provided in this embodiment of the invention includes:

[0045] The ratio of the aforementioned approval risk coefficient to the average historical approval risk coefficient of similar data within the historical time zone is used as the risk correction coefficient.

[0046] The product of the risk correction coefficient and the initial approval risk weight is used as the adaptive approval risk weight, wherein the initial approval risk weight is 0.5, and the adaptive approval risk weight is not less than 0.3 and not greater than 0.7.

[0047] The difference between 1 and the adaptation approval risk weight is taken as the adaptation approval emergency weight;

[0048] Based on the adaptive approval risk weight and the adaptive approval urgency weight, the adaptive time limit correction coefficient is obtained by weighting according to the first time limit correction coefficient and the second time limit correction coefficient. The first time limit correction coefficient is the risk correction coefficient, and the second time limit correction coefficient is the ratio of the approval urgency index to the average historical approval urgency index of the same type of data in the historical time zone.

[0049] The product of the adaptation time limit correction coefficient and the initial approval time limit is used as the adaptation approval time limit.

[0050] First, the ratio of the stated approval risk coefficient to the average historical approval risk coefficient of similar data within the historical time zone is used as the risk correction coefficient. The historical time zone refers to the past period for statistically analyzing the characteristics of similar data, such as the past 6 months. The average historical approval risk coefficient of similar data refers to the average risk coefficient of matters with the same approval business and data attributes as the current one within the historical time zone. Risk correction coefficient = Approval risk coefficient / Average historical approval risk coefficient. The risk correction coefficient reflects the degree of deviation of the current approval risk from the average risk of similar data. For example, if the average historical approval risk coefficient for the approval of automotive brake component drawings over the past 6 months is 0.5, and the current approval risk coefficient is 0.62, the risk correction coefficient = 0.62 / 0.5 = 1.24, indicating that the current approval risk is higher than the average level of similar data.

[0051] Secondly, the product of the risk correction coefficient and the initial approval risk weight is used as the adapted approval risk weight, where the initial approval risk weight is 0.5, and the adapted approval risk weight is not less than 0.3 and not greater than 0.7. The initial approval risk weight refers to the preset basic weight of the risk dimension, which is fixed at 0.5. The adapted approval risk weight = risk correction coefficient × initial approval risk weight. The adapted approval risk weight is the risk dimension weight adjusted based on the degree of risk deviation, and is limited to between 0.3 and 0.7 to avoid dimension imbalance. For example, with a risk correction coefficient of 1.24 and an initial approval risk weight of 0.5, the adapted approval risk weight = 1.24 × 0.5 = 0.62, which is within the range of 0.3-0.7, meaning the current adapted approval risk weight is 0.62.

[0052] Then, the difference between 1 and the adaptive approval risk weight is taken as the adaptive approval urgency weight. The adaptive approval urgency weight is the weight corresponding to the urgency index, and the adaptive approval urgency weight = 1 - the adaptive approval risk weight, ensuring that the sum of the risk and urgency dimension weights is 1. For example, if the adaptive approval risk weight is 0.62, the adaptive approval urgency weight = 1 - 0.62 = 0.38.

[0053] Further, based on the adapted approval risk weight and adapted approval urgency weight, an adapted time limit correction coefficient is obtained by weighting the first time limit correction coefficient and the second time limit correction coefficient. The first time limit correction coefficient is the risk correction coefficient, and the second time limit correction coefficient is the ratio of the approval urgency index to the average historical approval urgency index of similar data within the historical time zone. The first time limit correction coefficient, i.e., the risk correction coefficient, reflects the impact of risk on the time limit; the higher the risk, the larger the first time limit correction coefficient. The second time limit correction coefficient = current approval urgency index / average historical urgency index of similar data. The second time limit correction coefficient reflects the impact of urgency on the time limit; the higher the urgency, the larger the second time limit correction coefficient. The adapted time limit correction coefficient is a comprehensive coefficient weighted by the risk and urgency dimensions, used to adjust the initial time limit. Adapted time limit correction coefficient = adapted approval risk weight × first time limit correction coefficient + adapted approval urgency weight × second time limit correction coefficient.

[0054] For example, the current risk correction coefficient is 1.24, and the risk weight for the adaptation approval is 0.62; the average urgency index for the approval of automotive brake component drawings over the past 6 months is 0.7, and the current urgency index is 0.65. The second time limit correction coefficient = 0.65 / 0.7 ≈ 0.93. Weighted summation, the adaptation time limit correction coefficient = 0.62 × 1.24 + 0.38 × 0.93 ≈ 0.769 + 0.353 = 1.122.

[0055] Finally, the product of the adaptation time limit correction factor and the initial approval time limit is taken as the adaptation approval time limit. The initial approval time limit refers to the normal approval time for similar data within the historical time zone. The adaptation approval time limit refers to the final approval time limit after dynamic adjustment. Adaptive approval time limit = adaptation time limit correction factor × initial approval time limit. For example, the initial approval time limit for the approval of automotive brake component drawings in the past 6 months was 20 days. The adaptation time limit correction factor is 1.122, and the adaptation approval time limit = 20 × 1.122 ≈ 22 days. This time limit is longer than the initial time limit because the risk is higher than the average, but it is not excessively compressed because the urgency is lower than the average, thus balancing risk and business needs.

[0056] In this embodiment of the invention, a balance between risk control needs and urgent business needs is achieved through risk correction, weight adjustment, and weighted calculation: ensuring that high-risk matters receive sufficient review time, with the time limit extended accordingly as the risk coefficient increases; while also adapting to the efficiency demands of low-urgency matters, with the time limit shortened accordingly as the urgency index increases; at the same time, by limiting the weight range, the time limit is avoided from being excessively affected by a single dimension, making the adapted approval time limit more in line with the actual needs of the scenario, and providing a reasonable time constraint for subsequent approval strategy optimization.

[0057] S300: Using the time limit less than the adaptive approval time limit as an efficiency constraint, optimize the initial approval strategy and select the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability as the adaptive approval scheme.

[0058] In this embodiment of the invention, the time limit for approval is less than the adaptive approval time limit as an efficiency constraint. The initial approval strategy is optimized by selecting the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability as the adaptive approval scheme. Initial approval strategies often use fixed node combinations and uniform verification depths, without dynamically adapting to actual approval time limit constraints and error probability risks. For example, in the automotive parts supply chain rights confirmation approval scenario, fixed strategies either time out due to excessive verification depth or have a high error rate due to insufficient nodes / insufficient verification, thus affecting approval quality. Therefore, it is necessary to use the adaptive approval time limit as a constraint, enumerate and filter combinations of nodes and verification depths, and finally select a low-error and compliant scheme.

[0059] Step S300 in the method provided in this embodiment of the invention includes:

[0060] An initial approval strategy for obtaining similar data of the data to be approved is provided, wherein the initial approval strategy includes several approval nodes and several approval node verification depth thresholds, and the approval node verification depth is the number of channel calls of the pre-built data anomaly identification engine.

[0061] Based on the aforementioned approval nodes and approval node verification depth thresholds, and according to a preset node verification depth step size, random combinations of approval nodes and verification depths are enumerated to generate several initial approval schemes. Each initial approval scheme includes an approval node combination and an approval node verification depth combination, and the approval node combination includes at least one approval node.

[0062] Using the time limit shorter than the adaptive approval time limit as an efficiency constraint, the initial approval schemes are screened to obtain multiple qualified approval schemes;

[0063] For each of the multiple qualified approval schemes, the probability of approval error is predicted, and multiple predicted approval error probabilities are output. The qualified approval scheme with the lowest predicted approval error probability is selected as the adapted approval scheme.

[0064] First, obtain the initial approval strategy for similar data of the data to be approved. The initial approval strategy includes several approval nodes and several approval node verification depth thresholds. The approval node verification depth is the number of channel calls of the pre-built data anomaly identification engine.

[0065] The method for constructing the data anomaly detection engine includes:

[0066] Based on the historical approval data verification records of similar data, a sample approval dataset was collected, and historical abnormal data corresponding to different sample approval data were collected to obtain a sample abnormal dataset.

[0067] The sample approval dataset and the sample anomaly dataset are divided into several parts to obtain several sample training sets;

[0068] Deep learning models are trained to convergence using the aforementioned sample training sets, generating several data anomaly identification channels, and then integrated to construct a data anomaly identification engine according to the mean fusion strategy.

[0069] First, based on historical approval data verification records of similar data, a sample approval dataset is collected. Then, historical anomaly data corresponding to different sample approval datasets is collected to obtain a sample anomaly dataset. The sample approval dataset refers to the collection of historical approval data of the same type. The sample anomaly dataset refers to the data set in the historical approval data that has integrity issues. For example, 1000 sets of data from the past 6 months of automotive brake component drawing rights confirmation approvals are collected as the sample approval dataset. From this, 20 sets of anomaly data records are extracted, such as inconsistent drawing versions and incorrect technical parameters, and these are used as the sample anomaly dataset.

[0070] Secondly, the sample approval dataset and the sample anomaly dataset are divided into several equal parts to obtain several sample training sets. Dividing the dataset into multiple equal parts is used to train multiple independent recognition channels to avoid overfitting of a single model. For example, the 1000 sample approval dataset and the 20 sample anomaly dataset are divided into 5 equal parts to obtain 5 sample training sets, each containing 200 sets of approval data and 4 sets of anomaly data. Each sample training set is used to train an independent recognition channel.

[0071] Subsequently, deep learning models are trained to convergence using the aforementioned sample training sets, generating several data anomaly detection channels. These channels are then integrated and used to construct a data anomaly detection engine based on a mean fusion strategy. Each data anomaly detection channel refers to a deep learning model, such as a convolutional neural network model, trained using a single sample training set, capable of identifying data integrity anomalies. The mean fusion strategy involves averaging the detection results of multiple data anomaly detection channels on the same data to obtain the final detection result, thereby improving the reliability of the detection. For example, a convolutional neural network (CNN) suitable for feature extraction of semi-structured data such as drawings is selected as the deep learning model. Five independent CNN branches are constructed for each of the five sample training sets, with each branch corresponding to one sample training set. Each branch model adopts a unified structure of input layer, double convolutional pooling layer, fully connected layer, and output layer. The loss function is binary cross-entropy, and the optimizer is Adam (learning rate 0.001). The training is iteratively trained with 32 data points per batch. When the anomaly recognition accuracy of each branch on the validation set is stable above 98%, the loss value drops below 0.02 and does not decrease for five consecutive rounds, convergence is determined. The five converged branches are used as data anomaly recognition channels. Finally, the average anomaly probability of the five data anomaly recognition channels is taken as the final recognition result of the engine through the mean fusion strategy, thus completing the construction of the data anomaly recognition engine.

[0072] Furthermore, an initial approval strategy for similar data to the data to be approved is obtained. This initial approval strategy includes several approval nodes and several approval node verification depth thresholds. The approval node verification depth is the number of channel calls made by a pre-built data anomaly detection engine. The initial approval strategy is a standard approval configuration for similar data, including approval nodes and their respective verification depth thresholds. An approval node refers to a review stage. The approval node verification depth threshold refers to the upper limit of engine channel calls. For example, the initial approval strategy for obtaining the ownership approval of automotive brake component drawings includes three approval nodes: supplier qualification review, drawing format compliance, and technical parameter matching with automaker standards; the verification depth threshold for each node is 2, meaning each node can call a maximum of two engine channels.

[0073] Secondly, based on the aforementioned approval nodes and approval node verification depth thresholds, random combinations of approval nodes and verification depths are enumerated according to a preset node verification depth step size to generate several initial approval schemes. Each initial approval scheme includes an approval node combination and an approval node verification depth combination, and the approval node combination includes at least one approval node. The preset node verification depth step size refers to the adjustment interval of the verification depth; here, the step size is set to 1, meaning the verification depth increases by values ​​of 1 and 2. An initial approval scheme refers to the configuration of different approval node combinations with corresponding verification depths, and each scheme contains at least one approval node. For example, based on 3 approval nodes and a verification depth threshold of 2, different combinations are enumerated with a step size of 1: Scheme 1: The approval node combination is the supplier qualification review stage, with a verification depth of 1, which means calling 1 data anomaly identification engine channel; Scheme 2: The approval node combination is the supplier qualification review stage and the drawing format compliance review stage, with verification depths of 2 and 1 respectively; Scheme 3: The approval node combination is the supplier qualification review stage, the drawing format compliance review stage, and the technical parameter matching with the car manufacturer's standards review stage, with verification depths of 2, 2, and 2 respectively; and so on, generating a total of 20 different initial approval schemes.

[0074] Furthermore, using the time limit less than the adaptive approval time limit as an efficiency constraint, the initial approval schemes are screened to obtain multiple qualified approval schemes.

[0075] Specifically, the timeframe shorter than the adaptive approval limit is used as an efficiency constraint to filter the initial approval schemes and obtain multiple qualified approval schemes, including:

[0076] Based on the historical approval data verification records of similar data, several historical approval duration sets corresponding to the several initial approval schemes are statistically analyzed, and the average values ​​are calculated to obtain several historical approval duration averages.

[0077] An initial approval scheme whose average historical approval time is less than the appropriate approval time limit is selected as a qualified approval scheme, resulting in multiple qualified approval schemes.

[0078] First, based on historical approval data verification records of similar data, several historical approval duration sets corresponding to the several initial approval schemes are statistically analyzed, and their average values ​​are calculated to obtain several historical approval duration averages. A historical approval duration set refers to the collection of time records for each approval in the past execution process of a certain initial approval scheme. The historical approval duration average is the average execution time of a scheme obtained by averaging the historical approval duration sets of a certain initial approval scheme. For example, considering 20 initial approval schemes, their historical approval times are calculated as follows: Scheme 1 has a historical approval time set of [14 days, 15 days, 16 days], with an average historical approval time of (14+15+16) / 3 = 15 days; Scheme 2 has a historical approval time set of [17 days, 18 days, 19 days], with an average historical approval time of (17+18+19) / 3 = 18 days; Scheme 3 has a historical approval time set of [24 days, 25 days, 26 days], with an average historical approval time of (24+25+26) / 3 = 25 days. This process is repeated to obtain a total of 20 average historical approval times.

[0079] Secondly, initial approval schemes with an average historical approval time shorter than the suitable approval time limit are selected as qualified approval schemes, resulting in multiple qualified approval schemes. A qualified approval scheme refers to an initial approval scheme with an average historical approval time shorter than the suitable approval time limit. For example, if the current suitable approval time limit is 22 days, and the average historical approval time of 12 schemes, such as Scheme 1 (15 days) and Scheme 2 (18 days), is less than 22 days, then 12 qualified approval schemes are obtained.

[0080] Finally, the approval error probability is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output. The qualified approval scheme with the lowest predicted approval error probability is selected as the adapted approval scheme.

[0081] Specifically, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output, including:

[0082] Based on historical approval data verification records of similar data, a sample approval scheme set is collected, and the proportion of historical approval error events corresponding to different sample approval schemes is counted as the sample approval error probability to obtain the sample approval error probability set.

[0083] Using the sample approval scheme set and sample approval error probability set as training data, a deep learning model is trained until convergence, and an approval error prediction plugin is generated.

[0084] Using the aforementioned approval error prediction plugin, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output.

[0085] First, based on historical approval data verification records of similar data, a sample approval scheme set is collected. The proportion of historical approval error events corresponding to different sample approval schemes is then calculated as the sample approval error probability, resulting in a sample approval error probability set. The sample approval scheme set refers to the collection of approval schemes that have actually been used in the past. The sample approval error probability refers to the proportion of times a data integrity error occurred during the past execution of a particular sample approval scheme out of the total number of executions. For example, 500 past schemes for the approval of automotive brake component drawings within the past 6 months are collected as the sample approval scheme set. The error probability of each scheme is calculated: if a scheme was executed 100 times, with 2 data integrity errors, its error probability = 2 / 100 = 2%. This process is repeated to obtain 500 sample approval error probabilities, which are then combined to obtain the sample approval error probability set.

[0086] Secondly, using the sample approval scheme set and sample approval error probability set as training data, a deep learning model is trained until convergence, generating an approval error prediction plugin. The approval error prediction plugin uses the sample approval scheme set as input and the sample approval error probability set as labels to train a deep learning model, such as a decision tree model, capable of predicting the error probability of new approval schemes. For example, targeting the structured feature attributes of approval schemes, a gradient boosting decision tree (XGBoost) is selected as the training model. The sample approval scheme is transformed into an 8-dimensional structured feature vector containing the number of approval nodes, verification depth, and node combination type. The sample approval error probability is used as the output label, and the mean squared error is used as the loss function. Core parameters such as a learning rate of 0.01 and a maximum decision tree depth of 5 are set. The dataset is split into a training set and a validation set in an 8:2 ratio for iterative training. When the mean squared error of the validation set stabilizes below 0.0001 and the mean absolute error is ≤0.1% and no longer decreases after 10 consecutive training rounds, the model is considered converged. Finally, the converged model is encapsulated as an approval error prediction plugin, which can take approval scheme features as input and output predicted error probability values.

[0087] Furthermore, using the aforementioned approval error prediction plugin, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output. For example, when 12 qualified approval schemes are input into the approval error prediction plugin, the predicted error probabilities for each scheme are obtained: the predicted error probability for scheme 1 is 3.2%; the predicted error probability for scheme 2 is 1.8%; and the predicted error probabilities for the remaining schemes are all not less than 1.8%, resulting in a total of 12 predicted approval error probabilities.

[0088] Finally, the qualified approval scheme with the lowest predicted approval error probability is selected as the adapted approval scheme. For example, Scheme 2, which has the lowest predicted error probability, is selected as the adapted approval scheme. The approval node combination of Scheme 2 is: supplier qualification review stage and drawing format compliance review stage, with verification depths of 2 and 1, respectively.

[0089] In this embodiment of the invention, a multi-channel data anomaly identification engine is constructed, providing reliable accuracy support for adjusting the verification depth and solving the problem of limited recognition capability of a single model. By enumerating various combinations of approval nodes and verification depth, the limitations of fixed approval strategies are broken, improving the adaptability of the solution to different scenarios. Screening based on the appropriate approval time limit ensures that the approval scheme meets efficiency requirements and avoids the risk of timeouts. An approval error prediction plugin is used to predict the probability of qualified schemes and select the optimal one, minimizing the risk of errors in approval data integrity. Ultimately, a dynamic balance between approval efficiency and verification accuracy is achieved, forming a low-risk, high-efficiency approval scheme adapted to actual business needs, laying a solid foundation for the accurate execution of approval data integrity verification in the subsequent trusted data space.

[0090] S400: Within the trusted data space, perform integrity verification on the data to be approved according to the adapted approval scheme.

[0091] In this embodiment of the invention, within a trusted data space, the data to be approved is subjected to integrity verification according to the adapted approval scheme. The trusted data space is a virtual data environment with secure isolation, encrypted transmission, and traceability characteristics, used to ensure the security and trustworthiness of the approval data and the verification process. Integrity verification refers to verifying the data content and associated information according to the requirements of the adapted approval scheme to confirm that there has been no tampering, omission, or logical error.

[0092] For example, firstly, the hydraulic piping drawings and related approval data for a new car's braking system submitted by the supplier are uploaded to a trusted data space. Symmetric encryption technology is used during data transmission to ensure security. Secondly, verification is initiated according to the approval node combination of the adapted approval scheme [supplier qualification review stage, drawing format compliance review stage, with verification depths of 2 and 1 respectively]. The first step is the supplier qualification review stage: two channels of the data anomaly detection engine are invoked to verify the supplier qualification files stored in the trusted data space. Channel 1 verifies whether the hash value of the qualification file matches the historical hash value stored on the blockchain; Channel 2 verifies whether the validity period and approval authority of the qualification file meet the supply chain rights confirmation requirements. The engine outputs the verification results using a mean fusion strategy, with an anomaly probability of 0.01, indicating a pass. The second step is the drawing format compliance review stage: one channel of the data anomaly detection engine is invoked to verify the hydraulic piping drawings of the braking system, focusing on whether the file format of the drawings conforms to the CAD standards specified by the automaker, whether the piping layout parameters match the automaker's basic design specifications, and whether the version identifier is the latest submission. The channel outputs an anomaly probability of 0.02, indicating a pass. After both approval nodes pass verification, a verification report is generated in the trusted data space, recording the verification node, verification depth, channel call results, and verification time (total time taken 16 days, less than the 22-day adaptation period). The report is then uploaded to the blockchain for evidence storage, completing the integrity verification.

[0093] In this embodiment of the invention, in the secure environment of the trusted data space, the verification is precisely performed according to the adapted approval scheme. By using the specified node combination and verification depth, data integrity issues are accurately identified. Furthermore, the encryption and traceability features of the trusted data space ensure the credibility and immutability of the verification process. At the same time, the verification time is controlled within the adapted approval time limit, ultimately achieving secure, efficient, and accurate verification of approval data integrity, and improving the integrity of the entire approval data chain in the trusted data space.

[0094] Through the specific implementation methods described above, the embodiments of the present invention achieve the following technical effects:

[0095] This invention provides a method and system for verifying the integrity of approval data in a trusted data space. Through intelligent evaluation of multi-dimensional data attributes and approval-related characteristics, it accurately quantifies the risk and urgency of approvals, providing a reliable basis for subsequent processes. Based on this evaluation, it dynamically calculates and adapts approval time limits, achieving a balance between risk control and urgent business needs. By enumerating combinations of approval nodes and verification depths, filtering time limit constraints, and predicting error probabilities, it optimizes and forms a low-risk, high-efficiency adapted approval scheme. Finally, the verification is executed in the secure environment of the trusted data space. Leveraging multi-channel anomaly identification and trusted evidence storage characteristics, it ensures the security and immutability of the verification process, comprehensively breaking the limitations of traditional fixed approval strategies, achieving dynamic adaptation between approval efficiency and verification accuracy, and ultimately effectively improving the integrity, reliability, and processing efficiency of the entire approval data chain in the trusted data space.

[0096] Example 2, as Figure 2 As shown, this invention provides an approval data integrity verification system in a trusted data space, the system comprising:

[0097] Risk and urgency assessment module 11 is used to assess and determine the approval risk coefficient and approval urgency index based on the data attribute information and approval association characteristics of the data to be approved;

[0098] The adaptation time limit calculation module 12 is used to calculate and obtain the adaptation approval time limit based on the approval risk coefficient and the approval urgency index.

[0099] The approval strategy optimization module 13 is used to optimize the initial approval strategy by taking the time limit less than the adapted approval time limit as an efficiency constraint, and select the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability as the adapted approval scheme.

[0100] The data integrity verification module 14 is used to perform integrity verification on the data to be approved in accordance with the adapted approval scheme within the trusted data space.

[0101] In one embodiment, the risk emergency assessment module 11 is further configured to:

[0102] Based on historical approval data verification records, sample data attribute information set and sample approval related feature set are collected, and the historical approval risk coefficient and historical approval urgency index corresponding to different sample data attribute information and sample approval related features are labeled to obtain sample approval risk coefficient set and sample approval urgency index set.

[0103] Using the sample data attribute information set and sample approval related feature set as input, and the sample approval risk coefficient set and sample approval urgency index set as supervision, a generative adversarial network is trained until both the generator and discriminator converge, thus constructing an approval feature recognition engine.

[0104] Using the aforementioned approval feature recognition engine, the approval risk coefficient and approval urgency index are determined based on the data attribute information and approval association features of the data to be approved.

[0105] The data attribute information includes at least the data source credibility, data sensitivity level, data complexity, data volume, and historical error rate of similar data. The approval association features include at least the priority of related business and the supervision time limit of related business. The approval urgency index is positively correlated with the priority of related business and negatively correlated with the supervision time limit of related business.

[0106] The approval risk coefficient is positively correlated with the data sensitivity level, data complexity, data volume, and historical error rate of similar data, and negatively correlated with the credibility of the data source.

[0107] In one embodiment, the adaptation time limit calculation module 12 is further configured to:

[0108] The ratio of the aforementioned approval risk coefficient to the average historical approval risk coefficient of similar data within the historical time zone is used as the risk correction coefficient.

[0109] The product of the risk correction coefficient and the initial approval risk weight is used as the adaptive approval risk weight, wherein the initial approval risk weight is 0.5, and the adaptive approval risk weight is not less than 0.3 and not greater than 0.7.

[0110] The difference between 1 and the adaptation approval risk weight is taken as the adaptation approval emergency weight;

[0111] Based on the adaptive approval risk weight and the adaptive approval urgency weight, the adaptive time limit correction coefficient is obtained by weighting according to the first time limit correction coefficient and the second time limit correction coefficient. The first time limit correction coefficient is the risk correction coefficient, and the second time limit correction coefficient is the ratio of the approval urgency index to the average historical approval urgency index of the same type of data in the historical time zone.

[0112] The product of the adaptation time limit correction coefficient and the initial approval time limit is used as the adaptation approval time limit.

[0113] In one embodiment, the approval strategy optimization module 13 is further configured to:

[0114] An initial approval strategy for obtaining similar data of the data to be approved is provided, wherein the initial approval strategy includes several approval nodes and several approval node verification depth thresholds, and the approval node verification depth is the number of channel calls of the pre-built data anomaly identification engine.

[0115] Based on the aforementioned approval nodes and approval node verification depth thresholds, and according to a preset node verification depth step size, random combinations of approval nodes and verification depths are enumerated to generate several initial approval schemes. Each initial approval scheme includes an approval node combination and an approval node verification depth combination, and the approval node combination includes at least one approval node.

[0116] Using the time limit shorter than the adaptive approval time limit as an efficiency constraint, the initial approval schemes are screened to obtain multiple qualified approval schemes;

[0117] For each of the multiple qualified approval schemes, the probability of approval error is predicted, and multiple predicted approval error probabilities are output. The qualified approval scheme with the lowest predicted approval error probability is selected as the adapted approval scheme.

[0118] The method for constructing the data anomaly detection engine includes:

[0119] Based on the historical approval data verification records of similar data, a sample approval dataset was collected, and historical abnormal data corresponding to different sample approval data were collected to obtain a sample abnormal dataset.

[0120] The sample approval dataset and the sample anomaly dataset are divided into several parts to obtain several sample training sets;

[0121] Deep learning models are trained to convergence using the aforementioned sample training sets, generating several data anomaly identification channels, and then integrated to construct a data anomaly identification engine according to the mean fusion strategy.

[0122] Specifically, the timeframe shorter than the adaptive approval limit is used as an efficiency constraint to filter the initial approval schemes and obtain multiple qualified approval schemes, including:

[0123] Based on the historical approval data verification records of similar data, several historical approval duration sets corresponding to the several initial approval schemes are statistically analyzed, and the average values ​​are calculated to obtain several historical approval duration averages.

[0124] An initial approval scheme whose average historical approval time is less than the appropriate approval time limit is selected as a qualified approval scheme, resulting in multiple qualified approval schemes.

[0125] Specifically, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output, including:

[0126] Based on historical approval data verification records of similar data, a sample approval scheme set is collected, and the proportion of historical approval error events corresponding to different sample approval schemes is counted as the sample approval error probability to obtain the sample approval error probability set.

[0127] Using the sample approval scheme set and sample approval error probability set as training data, a deep learning model is trained until convergence, and an approval error prediction plugin is generated.

[0128] Using the aforementioned approval error prediction plugin, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output.

[0129] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0131] This specification and accompanying drawings are merely illustrative examples of the invention and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its scope. Therefore, if such modifications and modifications fall within the scope of the invention and its equivalents, the invention is intended to include these modifications and modifications.

Claims

1. A method for verifying the integrity of approval data in a trusted data space, characterized in that, The methods include: Based on the data attribute information and approval association characteristics of the data to be approved, the approval risk coefficient and approval urgency index are determined. The approval association characteristics include the priority of related business and the supervision time limit of related business. The appropriate approval time limit is calculated based on the aforementioned approval risk coefficient and approval urgency index. Using the time limit less than the adaptive approval time limit as an efficiency constraint, the initial approval strategy is optimized, and the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability are selected as the adaptive approval scheme. Within the trusted data space, the integrity of the data to be approved is verified according to the adapted approval scheme. Based on the data attribute information and approval-related characteristics of the data to be approved, the approval risk coefficient and approval urgency index are assessed and determined, including: Based on historical approval data verification records, sample data attribute information set and sample approval related feature set are collected, and the historical approval risk coefficient and historical approval urgency index corresponding to different sample data attribute information and sample approval related features are labeled to obtain sample approval risk coefficient set and sample approval urgency index set. Using the sample data attribute information set and sample approval related feature set as input, and the sample approval risk coefficient set and sample approval urgency index set as supervision, a generative adversarial network is trained until both the generator and discriminator converge, thus constructing an approval feature recognition engine. Using the aforementioned approval feature recognition engine, the approval risk coefficient and approval urgency index are determined based on the data attribute information and approval correlation features of the data to be approved. Using the time limit less than the adaptive approval time limit as an efficiency constraint, the initial approval strategy is optimized. The combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability are selected as the adaptive approval scheme, including: An initial approval strategy for obtaining similar data of the data to be approved is provided, wherein the initial approval strategy includes multiple approval nodes and multiple approval node verification depth thresholds, and the approval node verification depth is the number of channel calls of the pre-built data anomaly identification engine. Based on the multiple approval nodes and multiple approval node verification depth thresholds, according to the preset node verification depth step size, random combinations of approval nodes and verification depths are enumerated to generate multiple initial approval schemes. Each initial approval scheme includes an approval node combination and an approval node verification depth combination, and the approval node combination includes at least one approval node. Using the time limit shorter than the adaptive approval time limit as an efficiency constraint, the multiple initial approval schemes are screened to obtain multiple qualified approval schemes; For each of the multiple qualified approval schemes, the probability of approval error is predicted, multiple predicted approval error probabilities are output, and the qualified approval scheme corresponding to the minimum predicted approval error probability is selected as the adapted approval scheme. The method for constructing the data anomaly detection engine includes: Based on the historical approval data verification records of similar data, a sample approval dataset was collected, and historical abnormal data corresponding to different sample approval data was collected to obtain a sample abnormal dataset. The sample approval dataset and the sample anomaly dataset are divided into multiple parts to obtain multiple sample training sets; Deep learning models are trained to convergence using the multiple sample training sets to generate multiple data anomaly identification channels, and a data anomaly identification engine is constructed by integrating them according to the mean fusion strategy.

2. The method for verifying the integrity of approval data in a trusted data space according to claim 1, characterized in that, The data attribute information includes at least the data source credibility, data sensitivity level, data complexity, data volume, and historical error rate of similar data. The approval association features include at least the priority of related business and the supervision time limit of related business. The approval urgency index is positively correlated with the priority of related business and negatively correlated with the supervision time limit of related business.

3. The method for verifying the integrity of approval data in a trusted data space according to claim 2, characterized in that, The approval risk coefficient is positively correlated with the data sensitivity level, data complexity, data volume, and historical error rate of similar data, and negatively correlated with the credibility of the data source.

4. The method for verifying the integrity of approval data in a trusted data space according to claim 1, characterized in that, The appropriate approval time limit is calculated based on the aforementioned approval risk coefficient and approval urgency index, including: The ratio of the aforementioned approval risk coefficient to the average historical approval risk coefficient of similar data within the historical time zone is used as the risk correction coefficient. The product of the risk correction coefficient and the initial approval risk weight is used as the adaptive approval risk weight, wherein the initial approval risk weight is 0.5, and the adaptive approval risk weight is not less than 0.3 and not greater than 0.

7. The difference between 1 and the adaptation approval risk weight is taken as the adaptation approval emergency weight; Based on the adaptive approval risk weight and the adaptive approval urgency weight, the adaptive time limit correction coefficient is obtained by weighting according to the first time limit correction coefficient and the second time limit correction coefficient. The first time limit correction coefficient is the risk correction coefficient, and the second time limit correction coefficient is the ratio of the approval urgency index to the average historical approval urgency index of the same type of data in the historical time zone. The product of the adaptation time limit correction coefficient and the initial approval time limit is used as the adaptation approval time limit.

5. The method for verifying the integrity of approval data in a trusted data space according to claim 1, characterized in that, Using the timeframe shorter than the specified adaptation approval time limit as an efficiency constraint, the multiple initial approval schemes are screened to obtain multiple qualified approval schemes, including: Based on the historical approval data verification records of similar data, the historical approval duration sets corresponding to the multiple initial approval schemes are statistically analyzed, and the average values ​​are calculated to obtain the average historical approval durations. An initial approval scheme whose average historical approval time is less than the appropriate approval time limit is selected as a qualified approval scheme, resulting in multiple qualified approval schemes.

6. The method for verifying the integrity of approval data in a trusted data space according to claim 1, characterized in that, For each of the multiple qualified approval schemes, the probability of approval error is predicted, and multiple predicted approval error probabilities are output, including: Based on historical approval data verification records of similar data, a sample approval scheme set is collected, and the proportion of historical approval error events corresponding to different sample approval schemes is counted as the sample approval error probability to obtain the sample approval error probability set. Using the sample approval scheme set and sample approval error probability set as training data, a deep learning model is trained until convergence, and an approval error prediction plugin is generated. Using the aforementioned approval error prediction plugin, the probability of approval errors is predicted for each of the multiple qualified approval schemes, and multiple predicted approval error probabilities are output.

7. A system for verifying the integrity of approval data in a trusted data space, characterized in that, The system is used to implement the method for verifying the integrity of approval data in a trusted data space as described in any one of claims 1-6, the system comprising: The risk urgency assessment module is used to assess and determine the approval risk coefficient and approval urgency index based on the data attribute information and approval-related characteristics of the data to be approved. The adaptation time limit calculation module is used to calculate and obtain the adaptation approval time limit based on the approval risk coefficient and the approval urgency index. The approval strategy optimization module is used to optimize the initial approval strategy by taking the time limit less than the adapted approval time limit as an efficiency constraint, and select the combination of approval nodes and the combination of approval node verification depth corresponding to the minimum predicted approval error probability as the adapted approval scheme. The data integrity verification module is used to perform integrity verification on the data to be approved in accordance with the adapted approval scheme within the trusted data space.

Images