Blockchain-based enterprise trusted data access method and system

Abstract

The application relates to the technical field of data security, and discloses an enterprise trusted data access method and system based on a block chain, which comprises the following steps: constructing a sandbox activity coordinate system to identify potential bifurcation risks and generating a risk assessment report; extracting an affected sandbox list in the risk assessment report, creating a three-dimensional space coordinate system of a dependency relationship, generating a sandbox dependency fingerprint, drawing a dependency trajectory path, and exporting a dependency influence matrix; constructing a multi-sector time sequence coordination disc according to the dependency influence matrix to obtain a time sequence correctness guarantee state; and executing an influence range adaptive isolation mechanism by using the time sequence correctness guarantee state and the dependency influence matrix; the application can effectively avoid sandbox state bifurcation and guarantee the trustworthiness and stability of enterprise data access.

Description

Technical Field

[0001] This invention relates to the field of data security technology, and more specifically, to a blockchain-based method and system for accessing trusted enterprise data. Background Technology

[0002] In distributed environments where enterprise data interaction is becoming increasingly frequent, the trustworthiness and stability of data access have become core requirements for ensuring business continuity. Especially when multiple entities operate on data simultaneously, how to prevent the spread of risks caused by inconsistencies in state and ensure the correctness of data sequence are critical issues that urgently need to be addressed in the field of data security.

[0003] In the prior art, Chinese patent application CN119918065A discloses a data risk management system and method based on a large model. This system acquires and preprocesses data, uses a large model to analyze the data to construct a data relationship graph, identifies potential hidden relationships, performs multimodal fusion analysis combining semantic association and temporal dependency, fits a risk assessment function, constructs a dynamic threshold function, and finally provides risk response suggestions, aiming to improve the accuracy of risk identification. Chinese patent application CN120298117A discloses a method and system for identifying transaction risks. This method initializes a dynamic transaction graph by acquiring multi-source heterogeneous data, updates the graph in real time based on multi-source event streams, extracts the topological risk features of affected enterprise nodes and outputs a risk score, and identifies transaction risks by combining cross-chain verification results, with a focus on improving the accuracy of invoice risk identification.

[0004] However, while the aforementioned existing technologies can improve the accuracy and systematic nature of risk identification, none of them provide an effective solution to the state fork paradox caused by "savepoint" recovery operations in distributed sandbox environments. In scenarios with concurrent operations from multiple data providers, if one party recovers to a historical savepoint while the modifications made by other parties have already been consumed downstream, existing technologies cannot ensure a consistent understanding of the state change among all relevant parties through consensus mechanisms, nor can they accurately calculate the scope of the recovery operation to avoid chain reactions under circular dependencies. Furthermore, their lack of end-to-end traceability of state changes makes it difficult to support multi-level rollbacks of operations, directly leading to data inconsistency, causing temporal disorder, and ultimately affecting the stable operation of the entire data ecosystem. Summary of the Invention

[0005] This invention is applicable to distributed sandbox environments for cross-domain data collaboration among multiple enterprises, such as real-time data sharing between upstream and downstream enterprises in the supply chain and trusted exchange of government data across institutions. In these scenarios, multiple entities simultaneously perform read / write, savepoint creation, and recovery operations on data, requiring strict assurance of consistency and resilience during data access. To overcome the aforementioned deficiencies of existing technologies, this invention provides a blockchain-based enterprise trusted data access method and system. By constructing a sandbox activity coordinate system to warn of potential fork risks, generating a dependency-influence matrix to map relationships between sandboxes, coordinating timing, and executing adaptive isolation, it effectively solves the state fork problem during savepoint recovery in distributed sandbox environments. It ensures consensus on data state changes, predictability of the scope of impact, and rollbackability of operations, significantly improving the trustworthiness of enterprise data access and system stability.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] Blockchain-based methods for enterprise trusted data access include:

[0008] Construct a sandbox activity coordinate system, set a fork warning area in the sandbox activity coordinate system, identify potential fork risks through the fork warning area and generate a fork risk assessment report;

[0009] Extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependencies, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependencies, and derive the dependency impact matrix from the dependency trajectory paths.

[0010] Construct a multi-sector timing coordination disk based on the dependency and influence matrix, and obtain the timing correctness guarantee state based on the multi-sector timing coordination disk;

[0011] By leveraging the temporal correctness guarantee of the state and dependency influence matrix, an adaptive isolation mechanism for the scope of influence is implemented.

[0012] Furthermore, the sandbox activity coordinate system takes the current moment as the origin, the horizontal axis X as the time axis, the positive direction points to future time, the negative direction points to historical time, and the vertical axis Y is the data change frequency deviation axis, the positive direction indicates that the data change frequency increases relative to the baseline value, and the negative direction indicates that the data change frequency decreases relative to the baseline value.

[0013] Furthermore, the bifurcation warning area is a square, with the origin of the sandbox activity coordinate system as the geometric center, and the coordinates of the four vertices are (-M' / 2,-M' / 2), (-M' / 2,M' / 2), (M' / 2,M' / 2), and (M' / 2,-M' / 2), respectively, where M' is the side length of the square bifurcation warning area.

[0014] Furthermore, the method for identifying potential fork risks through fork warning areas includes:

[0015] Continuously monitor the sandbox's data read frequency, data write frequency, savepoint creation interval, and sandbox-dependent interaction strength to form a sandbox operation feature vector sequence;

[0016] A three-layer conditional judgment for bifurcation warning detection is performed on the sandbox operation feature vector sequence to identify potential bifurcation risks; the first layer of the three-layer conditional judgment is the judgment within the region, the second layer is the judgment of increasing trend, and the third layer is the judgment of threshold exceeding limit.

[0017] Furthermore, the method for generating the sandbox-dependent fingerprint includes:

[0018] Collect the input and output datasets for each sandbox at time t;

[0019] Calculate the feature hash values ​​of the input and output datasets to generate the input feature hash value and the output feature hash value;

[0020] The input feature hash value, the output feature hash value, and time t are fused together using a hash combination function to generate a unique sandbox dependency fingerprint.

[0021] Furthermore, the method for drawing the trajectory path includes:

[0022] By leveraging sandbox dependency fingerprints to identify data consumption relationships between sandboxes, the data provider sandbox and the data consumer sandbox can be determined.

[0023] Based on the dependency relationship three-dimensional spatial coordinate system, obtain the three-dimensional spatial coordinate positions of the data provider sandbox and the data consumer sandbox;

[0024] Directed edge connections are created between the three-dimensional spatial coordinates of the data provider sandbox and the data consumer sandbox to form a dependent trajectory path.

[0025] Furthermore, the method for determining the data provider sandbox and the data consumer sandbox includes:

[0026] For any two sandboxes i' and j', by comparing the input feature hash value of sandbox j' with the output feature hash value of sandbox i', if the first 48 bits of the two are the same, it is determined that sandbox j' has consumed the data of sandbox i'. Sandbox i' is the data provider sandbox, and sandbox j' is the data consumer sandbox. Here, i' and j' are the index variables of the sandboxes, and i' ≠ j'.

[0027] Furthermore, the method for deriving the dependency influence matrix from the dependent trajectory path includes:

[0028] The geometric feature parameters of the dependent trajectory path are analyzed, and based on the geometric feature parameters, a dynamic threshold criterion is applied to identify three types of dependency relationships: strong dependency, deep dependency, and circular dependency.

[0029] By integrating the three types of identified dependencies, a dependency influence matrix is ​​constructed to describe the mutual influence between sandboxes. The matrix element values ​​in the dependency influence matrix represent the influence coefficients.

[0030] Furthermore, the method for identifying the three types of dependency relationships—strong dependency, deep dependency, and circular dependency—using the dynamic threshold criterion includes:

[0031] Calculate the slope of each directed edge in the dependent trajectory path. When the absolute value of the slope exceeds the set slope threshold, it is marked as a strong dependency type.

[0032] Calculate the length of the dependency trajectory path from the source sandbox to the target sandbox. When the number of hops through the sandboxes traversed by the dependency trajectory path exceeds the set hop threshold, it is marked as a deep dependency type.

[0033] Detect whether the dependency path forms a closed loop. When it starts from the starting sandbox and eventually returns to the starting sandbox, it is marked as a circular dependency type.

[0034] A blockchain-based enterprise trusted data access system is used to implement the aforementioned blockchain-based enterprise trusted data access method. The system includes:

[0035] Fork Warning Module: Used to construct a sandbox activity coordinate system, set a fork warning area in the sandbox activity coordinate system, identify potential fork risks through the fork warning area and generate a fork risk assessment report;

[0036] Dependency Mapping Module: Used to extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependency relationships, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependency relationships, and derive the dependency impact matrix from the dependency trajectory paths;

[0037] Timing coordination module: used to construct a multi-sector timing coordination disk based on the dependency influence matrix, and to obtain the timing correctness guarantee state based on the multi-sector timing coordination disk;

[0038] Adaptive Isolation Module: Utilizes timing correctness to ensure state and dependency influence matrix, and executes an adaptive isolation mechanism for the scope of influence.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] This invention enables early identification and assessment of potential forking risks by constructing a sandbox activity coordinate system, providing a precise basis for subsequent risk management. Utilizing a three-dimensional coordinate system of dependencies, sandbox dependency fingerprints, and dependency trajectory paths, this invention clearly maps dependencies between sandboxes and quantifies them into a dependency influence matrix, allowing for precise analysis of interactions between sandboxes. Based on the dependency influence matrix, this invention constructs a multi-sector temporal coordination disk, ensuring the temporal correctness of savepoint operations and avoiding state forks caused by temporal chaos. Combining the temporal correctness guarantee of the state and the adaptive isolation mechanism of the dependency influence matrix execution's impact range, this invention can precisely control the scope of risk diffusion and avoid chain reactions. Through these solutions, this invention effectively solves the potential state forking problem during sandbox savepoint recovery in a distributed environment, ensuring data consistency, improving the system's resilience and business continuity under complex dependencies, and providing comprehensive and reliable technical support for enterprise trusted data access. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0042] Figure 1 A flowchart illustrating a blockchain-based enterprise trusted data access method provided in this embodiment of the invention;

[0043] Figure 2 A schematic diagram illustrating the principle of identifying potential fork risks provided in an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram illustrating the principle of the first-level judgment provided in an embodiment of the present invention;

[0045] Figure 4 A schematic diagram illustrating the principle of failing the first-level judgment in an embodiment of the present invention;

[0046] Figure 5 This is a flowchart of a method for drawing dependent trajectory paths provided in an embodiment of the present invention;

[0047] Figure 6 This is a functional block diagram of the blockchain-based enterprise trusted data access system of the present invention. Detailed Implementation

[0048] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1:

[0050] Please see Figure 1 As shown, this embodiment provides a blockchain-based method for accessing trusted enterprise data, including:

[0051] Step S10: Construct a sandbox activity coordinate system, set a fork warning area in the sandbox activity coordinate system, identify potential fork risks through the fork warning area and generate a fork risk assessment report;

[0052] Further, step S10 includes:

[0053] Step S11: Establish a sandbox activity coordinate system with the current time as the origin;

[0054] Step S11 establishes a two-dimensional Cartesian coordinate system for sandbox activity with the current time T as the origin. The horizontal axis X is the time axis, with the positive direction pointing to future time and the negative direction pointing to historical time, in minutes, used to quantify the temporal changes in sandbox operation. The vertical axis Y is the data change frequency deviation axis, with the positive direction indicating an increase in data change frequency relative to the baseline value and the negative direction indicating a decrease in data change frequency relative to the baseline value, in operations per minute, used to quantify the degree of deviation between the sandbox data interaction activity and the baseline state. Here, the abstract relationship between time and frequency deviation is transformed into a concrete geometric space. Traditional techniques for describing sandbox states often rely on discrete numerical records, lacking a unified spatial reference framework, making it difficult to intuitively compare and analyze states at different points in time. The establishment of the sandbox activity coordinate system allows the sandbox's operational status at any given time to be uniquely identified by coordinates (x, y). Here, x represents the time difference between the current time and the target time; for example, 2 minutes after the current time is denoted as x = 2, and 3 minutes before the current time is denoted as x = -3. y represents the difference between the number of data changes per minute at that time and a baseline value. The baseline value is determined by statistically analyzing the average data change frequency of the sandbox under the same business scenario over the past 72 hours. For example, if the baseline value is 8 changes per minute and the current actual data change frequency is 10 changes per minute, then y = 2; if the current actual data change frequency is 3 changes per minute, then y = -5. This transformation provides a spatial basis for setting subsequent warning areas. Without this step, subsequent bifurcation warning areas would lack a unified coordinate baseline and would be impossible to define accurately, leading to incomparability of state data from different sandboxes and consequently, confusion in risk assessment standards. The sandbox activity coordinate system provides a clear boundary definition dimension for square bifurcation warning areas, enabling mathematical precision in judgments within and outside the area, rather than relying on fuzzy empirical thresholds.

[0055] Step S12: Set a square forking warning area with the origin as the geometric center in the sandbox activity coordinate system;

[0056] In the established sandbox activity coordinate system, a square bifurcation warning area with a side length of M' is set with the origin of the coordinate system as the geometric center. The coordinates of its four vertices are (-M' / 2, -M' / 2), (-M' / 2, M' / 2), (M' / 2, M' / 2), and (M' / 2, -M' / 2), respectively. The value of the side length M' is determined by the historical average data change frequency Favg. Specifically, M' equals Favg multiplied by 2. Favg is obtained by statistically analyzing the average number of data changes per minute of this type of sandbox in the past preset period (such as 90 days). For example, if the historical average data change frequency is 8 times per minute, then M' = 16. The boundary of the square area is x = -8 to x = 8 and y = -8 to y = 8. The square was chosen over a circle or other shapes because traditional circular boundary determination requires calculating the distance from a point to the center, involving square root operations, which increases the computational complexity of the computer program. In contrast, square boundary determination only requires a simple inequality comparison, such as whether x is within the range [-M' / 2, M' / 2], making it more suitable for efficient computation in real-time monitoring scenarios. The upper boundary of the square (y = M' / 2) corresponds to a high-frequency change time window, the lower boundary (y = -M' / 2) corresponds to a low-frequency change time window, the left boundary (x = -M' / 2) limits the frequency range of historical time, and the right boundary (x = M' / 2) limits the frequency prediction of future time. This clear boundary division allows for precise identification of whether the sandbox state has entered a risk range.

[0057] Step S13: Continuously monitor the sandbox's data reading frequency, data writing frequency, save point creation interval, and sandbox-dependent interaction strength to form a sandbox operation feature vector sequence;

[0058] The system continuously monitors the sandbox's data read frequency (Fr), data write frequency (Fw), savepoint creation interval (Ts), and sandbox interaction strength (Id) to form a sandbox operation feature vector sequence. The data read frequency (Fr) represents the number of data read operations per minute, the data write frequency (Fw) represents the number of data write operations per minute, the savepoint creation interval (Ts) is the time difference between two adjacent savepoint creations, and the sandbox interaction strength (Id) represents the amount of data exchanged with other sandboxes per unit time, such as the number of data blocks exchanged per minute. The collection interval is set to 30 seconds, meaning the values ​​of the above four parameters are recorded every 30 seconds. For example, the sandbox operation feature vector at a certain moment can be represented as (Fr = 5 times / minute, Fw = 6 times / minute, Ts = 10 minutes, Id = 4 blocks / minute). These four parameters were chosen because traditional techniques only monitor a single read / write frequency, which cannot comprehensively reflect the sandbox's operational status. Ts reflects the activity level of the savepoint mechanism, and Id reflects the strength of the sandbox's connection with the outside world. Abnormal changes in these two parameters often precede forking and are important early warning indicators. The 30-second data collection interval is determined based on the average cycle of sandbox data changes. If the interval is too long, it may miss critical state changes; if the interval is too short, it will generate a large amount of redundant data, increasing the storage and computing burden.

[0059] Step S14: Perform a three-layer conditional judgment fork warning detection on the sandbox operation feature vectors in the sandbox operation feature vector sequence to identify potential fork risks;

[0060] Please see Figure 2 As shown, step S14 further includes:

[0061] Step S141: The first layer of the three-layer condition judgment is the judgment within the region, the second layer is the judgment of increasing trend, and the third layer is the judgment of threshold exceeding limit;

[0062] Step S142: Map the data reading frequency Fr and data writing frequency Fw in the sandbox operation feature vector to the sandbox activity coordinate system to form feature points. Determine whether the feature points fall into the square bifurcation warning area to complete the judgment in the first layer of the area.

[0063] Step S143: For the sandbox operation feature vector judged through the first layer region, calculate the ratio Fr / Fw of data reading frequency to data writing frequency within n1 consecutive sampling periods, detect whether the ratio Fr / Fw shows an increasing trend, and complete the second layer trend increasing judgment.

[0064] Step S144: For the sandbox operation feature vectors judged by the first two layers, check whether the creation interval of the save point is less than the preset time threshold and whether the intensity of the interaction with the sandbox exceeds K times the historical average, complete the third layer threshold over-limit judgment, and identify potential fork risks.

[0065] Specifically, step S141 defines three layers of conditional judgment: judgment within the region, judgment of increasing trend, and judgment of exceeding the threshold. Step S142 is the first layer of judgment within the region. The feature point (x, y) is generated by mapping Fr and Fw. The x-coordinate of the feature point is the time difference between the sampling time and the current time T. The current time is x = 0, future time is a positive value, and historical time is a negative value. The y-coordinate of the feature point is the difference between (Fr + Fw) / 2 and the baseline value. To determine whether the feature point falls within the square bifurcation warning region, we check whether the x-coordinate satisfies -M' / 2 ≤ x ≤ M' / 2 and whether the y-coordinate satisfies -M' / 2 ≤ y ≤ M' / 2. If both conditions are met, the feature point is determined to fall within the region, completing the first layer of judgment within the region. Figure 3-4 As shown, feature point (x1, y1) falls within the square bifurcation warning area and passes the first-level judgment; feature point (x2, y2) does not fall within the square bifurcation warning area and fails the first-level judgment. This step ensures that the analysis object is a sandbox that has entered the risk area, excludes sandboxes that are obviously in a safe state, and reduces the amount of computation required for subsequent judgments.

[0066] Step S143 is the second-layer trend increase judgment. For the sandbox operation feature vector that passed the first-layer judgment, the ratio of Fr to Fw, Fr / Fw, is calculated over n1 consecutive sampling periods. If the ratio of the next period is greater than that of the previous period, it is determined to show an increasing trend. The value of n1 is determined based on the statistical analysis of the feature periods before the historical fork, and is usually 3. For example, the ratios Fr / Fw of the three sampling periods are 0.8, 0.9, and 1.1, which meet the increasing trend. Step S144 is the third-layer threshold exceedance judgment. For the sandbox operation feature vector that passed the first two layers of judgment, it is checked whether Ts is less than the preset time threshold Tmin and whether Id is greater than K times the historical average. K is a multiplier used to quantify the abnormal degree of the sandbox interaction intensity. The time threshold is set based on the sandbox data consistency requirements. If the save point interval is too short, it will increase the system burden. If it is too long, it may lead to difficulties in recovery after the fork. For example, the time threshold can be set to 5 minutes. The K value is determined by the fluctuation range of normal interaction intensity in historical data. Generally, an interaction exceeding three times the historical average indicates abnormally high activity; therefore, K=3 can be set. The three-layer conditional judgment works synergistically: the first layer ensures the risk range, the second layer identifies trend changes, and the third layer verifies the degree of anomaly. Compared to traditional single-threshold judgments, this significantly improves the accuracy of risk identification. The absence of any layer may lead to false positives or false negatives. False positives, for example, might be misjudging a sandbox entry into a risky area if only the first layer is used; false negatives, for example, might be ignoring the true risk of abnormal interaction intensity if the third layer is missing.

[0067] Step S15: Based on the identified potential fork risks, generate a fork risk assessment report that includes the time window of potential fork points and a list of affected sandboxes.

[0068] Specifically, the fork risk assessment report includes the potential fork point time window, a list of affected sandboxes and their dependency depth, and a suggested adjustment plan for savepoint creation strategy. The potential fork point time window is predicted by analyzing the changing trends of Ts and Id in the sandbox operation feature vector. For example, if Ts gradually shortens and Id gradually increases, the time range in which a fork may occur can be predicted, such as the next 10 to 15 minutes. The dependency depth is calculated based on the hierarchical distance between the current sandbox and the target sandbox by counting the number of intermediate sandboxes (i.e., hops) in the statistical data transmission path. Specifically, it is "1 divided by (hop count + 1)", where the hop count is the total number of intermediate sandboxes in the data transmission path from the current sandbox to the target sandbox. Its value ranges from 0 to 1, where 0 indicates no dependency and 1 indicates direct dependency, i.e., direct data transmission. The larger the value, the more direct the dependency. For example, sandbox A directly depends on sandbox B, with no intermediate sandboxes between them, resulting in a hop count of 0. Therefore, the dependency depth of sandbox A on sandbox B is 1 / (0+1) = 1. Sandbox B directly depends on sandbox C, and similarly, the dependency depth of sandbox B on sandbox C is 1. Sandbox A indirectly depends on sandbox C through sandbox B, with one intermediate sandbox (sandbox B) in the data transmission path, resulting in a hop count of 1. Therefore, the dependency depth of sandbox A on sandbox C is 1 / (1+1) = 0.5. In this case, the list of associated sandboxes includes A, B, and C, and the dependency depths of sandbox A on each sandbox in the list are: 0 for A itself (no dependency), 1 for B, and 0.5 for C. The list of affected sandboxes lists the sandboxes that are directly or indirectly associated. A suggested adjustment to the savepoint creation strategy is, for example, to adjust the savepoint interval from the original 10 minutes to 5 minutes based on the risk level, and to create savepoints encrypted within the predicted time window.

[0069] Step S10, through the synergistic effect of constructing a sandbox activity coordinate system, setting a fork warning area, collecting sandbox operation feature vectors, performing three-layer condition judgments, and generating a fork risk assessment report, achieves proactive prediction of sandbox fork risks to identify risks in advance and avoid irreversible losses. By spatializing the relationship between time and frequency, the risk states of different sandboxes can be compared and analyzed under the same coordinate system, achieving relative quantification of risks and solving the problem that risks can only be qualitatively described in traditional technologies. The combination of the square fork warning area and the three-layer condition judgments reduces computational complexity while ensuring judgment accuracy, making real-time monitoring possible. Traditional complex models often cannot meet real-time requirements due to excessive computation. The collaborative analysis of multiple parameters in the sandbox operation feature vectors can not only identify fork risks but also predict the severity of risks in advance through the correlation between parameters, such as the synchronicity of Ts shortening and Id increasing, providing a basis for risk level classification for subsequent steps. This is something that single-parameter monitoring cannot achieve. The above steps together promote the transformation of the entire solution from passive processing to proactive prevention.

[0070] Step S20: Extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependency relationships, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependency relationships, and derive the dependency impact matrix from the dependency trajectory paths.

[0071] Further, step S20 includes:

[0072] Step S21: Based on the list of affected sandboxes, create a three-dimensional spatial coordinate system for dependencies with data flow direction as the X-axis, dependency depth as the Y-axis, and time dimension as the Z-axis.

[0073] Based on the list of affected sandboxes in the fork risk assessment report, a three-dimensional spatial coordinate system for dependencies is created, with data flow direction as the X-axis, dependency depth as the Y-axis, and time as the Z-axis. The list of affected sandboxes includes a unique sandbox identifier, its associated business module, direct association records with other sandboxes, and the time of its most recent status update. This information is extracted from the metadata registry of the sandbox management system to ensure coverage of all sandboxes potentially affected by the fork risk. Data flow direction refers to the direction of data transmission between sandboxes. An X-axis coordinate value of -1 indicates an upstream sandbox providing data to the current sandbox, 0 indicates the current sandbox itself, and 1 indicates a downstream sandbox receiving data from the current sandbox. For example, if sandbox A transmits data to sandbox B, then sandbox A's X-axis coordinate is -1, sandbox B's is 0, and sandbox C's is 1 if it receives data from sandbox B. Dependency depth refers to the hierarchical distance between the current sandbox and the target sandbox. The Y-axis coordinate value ranges from 0 to 1, where 0 indicates no dependency and 1 indicates direct dependency (direct data transmission). The Z-axis, with the current time as its origin and units in minutes, represents future time in the positive direction and historical time in the negative direction. It is used to record the time points when dependencies occur. For example, if sandbox A transmitted data to sandbox B 5 minutes before the current time, the Z-axis coordinate of this dependency would be -5. The creation of the three-dimensional coordinate system for dependencies is based on the list of affected sandboxes: First, the unique identifiers of all sandboxes in the list of affected sandboxes are extracted as the core nodes of the coordinate system. Then, the X-axis direction is determined based on the direct association records of each sandbox. The dependency depth of the Y-axis is calculated using the hop count, and the Z-axis coordinate is determined by combining the state update time. Ultimately, each sandbox has a unique coordinate in three-dimensional space. In traditional techniques, two-dimensional dependency graphs can only reflect the direction and intensity of data flow, but cannot reflect the dynamic changes in the time dimension. However, this three-dimensional coordinate system, by introducing the Z-axis, can simultaneously describe the directionality (X-axis), tightness (Y-axis), and timeliness (Z-axis) of the dependency relationship, providing a unified framework for the subsequent spatial localization of dependency fingerprints and time series analysis of trajectory paths. Without this step, the dependency fingerprint in step S22 will not be able to achieve dynamic tracking due to the lack of spatial coordinate reference, and the trajectory path in step S23 will also not be able to be accurately drawn due to the lack of dimensional reference, causing the description of the dependency relationship to remain at the static level and unable to reflect the changes during runtime.

[0074] Step S22: Generate a unique sandbox dependency fingerprint for each sandbox;

[0075] Specifically, a unique sandbox dependency fingerprint is generated for each sandbox. The process includes: Step S221, collecting the input dataset Din and output dataset Dout for each sandbox at time t; the input dataset Din contains all data blocks and their checksums received by the sandbox from the upstream sandbox in the past minute, and the output dataset Dout contains all data blocks and their checksums sent to the downstream sandbox during the same period. The collection period is synchronized with the 30-second sampling period in step S13 to ensure data timeliness. Step S222, calculating the feature hash values ​​of the input dataset and the output dataset to generate the input feature hash value and the output feature hash value; hashing Din and Dout using the SHA-256 algorithm is performed respectively. Because SHA-256 has 128-bit collision resistance, it can ensure that different datasets generate different hash values. The input feature hash value Hin and the output feature hash value Hout are both 64-bit hexadecimal strings. For example, if Din contains data blocks "abc" and "def", then Hin is the hash result calculated after concatenating "abcdef". Step S223: The input feature hash value, output feature hash value, and time t are fused using a hash combination function to generate a unique sandbox dependency fingerprint. The hash combination function uses a concatenation method of "the first 32 bits of Hin + the hexadecimal representation of time t + the last 32 bits of Hout", where time t is represented as a 32-bit Unix timestamp (in seconds), converted to an 8-digit hexadecimal string (padded with leading zeros if less than 8 digits). The reason for this fingerprint design is that traditional dependency identifiers only rely on the sandbox ID and cannot distinguish the state changes of the same sandbox at different times. In contrast, the sandbox dependency fingerprint contains both input and output data features and time information, which can uniquely identify the dependency state of the sandbox at a certain time. This step works in conjunction with step S21. The temporal information in the sandbox dependency fingerprint corresponds to the Z-axis coordinate of the three-dimensional spatial coordinate system of the dependency relationship. The input dataset Din is the data block and check value received by the current sandbox from the upstream sandbox (X-axis coordinate value is -1). Its corresponding input feature hash value Hin is directly associated with the upstream data flow in the negative X-axis direction. The output dataset Dout is the data block and check value sent by the current sandbox to the downstream sandbox (X-axis coordinate value is 1). Its corresponding output feature hash value Hout is directly associated with the downstream data flow in the positive X-axis direction. This makes the fingerprint not only a status identifier but also a dynamic node in three-dimensional space. If step S22 is missing, step S23 will not be able to identify the data consumption relationship between sandboxes through fingerprint matching, resulting in the dependency trajectory path being unable to be drawn due to the lack of node identifiers.

[0076] Step S23: Based on the sandbox dependency fingerprint and the three-dimensional spatial coordinate system of dependency relationship, draw the dependency trajectory path when there is a data consumption relationship between sandboxes;

[0077] Please see Figure 5 As shown, step S23 further includes:

[0078] Step S231: Use sandbox dependency fingerprinting to identify the data consumption relationship between sandboxes and determine the data provider sandbox and the data consumer sandbox;

[0079] Step S232: Based on the three-dimensional spatial coordinate system of the dependency relationship, obtain the three-dimensional spatial coordinate positions of the data provider sandbox and the data consumer sandbox;

[0080] Step S233: Create directed edge connections between the three-dimensional spatial coordinates of the data provider sandbox and the data consumer sandbox to form a dependent trajectory path.

[0081] The method of using sandbox-dependent fingerprint identification to determine data consumption relationships includes: for any two sandboxes i' and j', comparing the input feature hash value of sandbox j' with the output feature hash value of sandbox i', if the first 48 bits of the two are the same, then it is determined that sandbox j' has consumed data from sandbox i'. Sandbox i' is the data provider sandbox, and sandbox j' is the data consumer sandbox, where i' and j' are index variables of the sandboxes, and i' ≠ j'. The reason for choosing the first 48 bits is that the first 48 bits (24 bytes) of the SHA-256 hash already have an extremely low collision probability, while reducing the amount of comparison computation, balancing accuracy and efficiency.

[0082] The 3D coordinates of the data provider sandbox and the data consumer sandbox are obtained based on a 3D spatial coordinate system: the coordinates of the data provider sandbox (X... pro Y pro Z pro In ), X pro =-1, relative to the data consumer sandbox, Y pro Z represents the depth of dependence of the data provider sandbox on the data consumer sandbox. pro Z represents the time when the data provider sandbox sends data, such as 2 minutes before the current time. pro =-2; Coordinates of the data consumer sandbox (X con Y con Z con In ), X con =0, Y con Z represents the depth of dependence of the data consumer sandbox on the data provider sandbox. con Z represents the time when the data consumer sandbox receives data, such as 1 minute before the current time. con =-1.

[0083] Create a directed edge connecting the three-dimensional spatial coordinates of the data provider sandbox and the data consumer sandbox, with the starting point of the directed edge being (X). pro Y pro Z pro ), the endpoint is (X) con Y con Z con The thickness of the edges is determined by the amount of data transmitted; for example, the edge width is 5 pixels when transmitting 100MB of data and 2 pixels when transmitting 50MB. The color intensity reflects the dependency strength. In traditional technologies, dependencies are recorded through text lists, making it difficult to intuitively identify critical paths. However, the dependency trajectory path of this invention, through its geometric shape in three-dimensional space, allows direct observation of the time difference (Z) of data flow. con -Z pro =1 minute) reflects changes in transmission delay and dependency strength (Y) pro =1 to Y con =1) This indicates that the dependency has not decayed, and the differences in color and thickness make strong dependencies and high-frequency transmission paths more prominent. The three-dimensional spatial coordinate system of dependency relationships provides a spatial carrier for dependency trajectory paths, and the sandbox dependency fingerprint ensures the accuracy of dependency trajectory paths. The combination of the three transforms dynamic dependencies from abstract descriptions into concrete geometric figures. If step S23 is missing, the dynamic threshold judgment in step S24 will be unable to be performed due to the lack of quantifiable geometric features, resulting in the loss of a basis for dependency classification.

[0084] Steps S21-S23, through the synergistic effect of constructing a 3D coordinate system, generating sandbox dependency fingerprints, and drawing dependency trajectory paths, achieve a dynamic, three-dimensional, and quantitative description of sandbox dependencies, solving the problem that static dependency graphs cannot reflect dynamic changes. The time dimension (Z-axis) of the 3D spatial coordinate system of dependencies allows the impact of historical dependencies on the current situation to be traced. For example, by analyzing the trajectory path in the negative Z-axis direction, the potential impact of a data transmission error in a sandbox two hours ago on the current dependency can be discovered, which is impossible with a 2D graph. The uniqueness of the sandbox dependency fingerprint allows changes in dependencies to be accurately tracked. When the input dataset of a sandbox undergoes subtle changes, the difference in Hin will be directly reflected in the fingerprint, making it easy to locate the time point of data tampering or transmission errors. The geometric features of the trajectory path, such as the slope of the edges and whether the path is closed, can automatically expose hidden dependency problems. For example, if a trajectory path forms a closed loop in the negative Z-axis direction in 3D space, starting from sandbox A and returning to sandbox A through historical time, it indicates the existence of undiscovered historical cyclic dependencies, which traditional text records cannot identify.

[0085] Step S24: parse the geometric feature parameters of the dependent trajectory path, and based on the geometric feature parameters, apply the dynamic threshold criterion to identify three types of dependency relationships: strong dependency, deep dependency, and circular dependency.

[0086] Further, step S24 includes:

[0087] Step S241: Calculate the slope of each directed edge in the dependent trajectory path. When the absolute value of the slope exceeds the set slope threshold, it is marked as a strong dependency type.

[0088] Step S242: Calculate the length of the dependency trajectory path from the source sandbox to the target sandbox. When the number of hops through the sandboxes on the path exceeds the set hop threshold, it is marked as a deep dependency type.

[0089] Step S243: Detect whether the dependency path forms a closed loop. When it starts from the starting sandbox and eventually returns to the starting sandbox, it is marked as a circular dependency type.

[0090] Specifically, the slope represents the rate of change of dependency strength per unit time, obtained as the ratio of (end point Y coordinate value - start point Y coordinate value) to (end point Z coordinate value - start point Z coordinate value). When the absolute value of the slope exceeds a set slope threshold, the dependency relationship between sandboxes is marked as a strong dependency. The slope threshold is determined based on statistical analysis of historical data: the slope values ​​of all dependent edges over the past 90 days are collected, their standard deviation σ is calculated, and the slope threshold is set to 3 times the standard deviation. This value covers 99.7% of the normal slope fluctuation range, and only significantly abnormal slope changes are judged as strong dependencies. The source sandbox refers to the initial sandbox for data transmission, and the target sandbox refers to the final receiving sandbox. The length of the dependency path is measured by the number of hops the path traverses. The sandbox hop count is the number of intermediate sandboxes the data passes through from the source sandbox to the target sandbox plus one (including hops from the source sandbox to the first intermediate sandbox). For example, if the path is source sandbox A → intermediate sandbox B → target sandbox C, with two hops (A to B, B to C), the sandbox hop count is 2. When the sandbox hop count exceeds a set hop count threshold, it is marked as a deep dependency. The hop count threshold L is determined based on the system complexity. For example, for a simple system with fewer than 100 sandboxes, L = 3; for a medium-sized system with 100-500 sandboxes, L = 5; and for a complex system with more than 500 sandboxes, L = 8. This division is based on the average length statistics of normal dependency chains in systems of different sizes, ensuring that only excessively long paths are marked as deep dependencies.

[0091] The method for detecting whether a dependency path forms a closed loop is as follows: By traversing all dependency paths, the starting and ending sandboxes of each path are recorded. When the starting and ending sandboxes of a path are the same, it is determined that a closed loop has formed and is marked as a circular dependency type. For example, sandbox A → sandbox B → sandbox C → sandbox A, where both the starting and ending sandboxes are sandbox A, is a circular dependency. Traditional techniques rely on manually defined fixed rules for classifying dependencies, which cannot adapt to dynamic system changes. This dynamic threshold criterion, by combining geometric features and statistical analysis, allows the classification results to adaptively adjust according to the system's operating state. Without this step, the dependency influence matrix in step S25 would lack classification labels and would be unable to quantify the differences in influence between different types of dependencies, resulting in the matrix element values ​​being unfounded.

[0092] Step S25: Integrate the three types of identified dependencies and construct a dependency influence matrix that describes the mutual influence between sandboxes. The matrix element values ​​in the dependency influence matrix represent the influence coefficients.

[0093] Further, step S25 includes:

[0094] Step S251: Based on the identified strong dependency type, deep dependency type, and circular dependency type, assign different basic influence weights respectively;

[0095] Step S252: Calculate the direct dependency impact value between sandboxes, the indirect dependency impact value through intermediate sandboxes, and the time decay coefficient over time.

[0096] Step S253: Construct a dependency influence matrix by integrating the basic influence weight, direct dependency influence value, indirect dependency influence value, and time decay coefficient. Matrix element M ij This represents the influence coefficient of sandbox i on sandbox j, where i and j are index variables of the sandbox.

[0097] The Analytic Hierarchy Process (AHP) was used to assign basic influence weights based on the identified dependency types. Multiple experts were invited to compare the influence of the three types of dependencies pairwise, construct a judgment matrix, and calculate the weight values. The results were confirmed to be reasonable after a consistency test. For example, the basic weight for strong dependency type was 0.6, for deep dependency type it was 0.3, and for circular dependency type it was 0.4.

[0098] Calculate the direct dependency impact value, indirect dependency impact value, and time decay coefficient: The direct dependency impact value is calculated based on the average data transmission frequency from sandbox i to sandbox j, using the formula "Direct dependency impact value = Average transmission frequency / System maximum transmission frequency". For example, if the average transmission frequency is 20 times / minute and the system maximum is 40 times / minute, then the direct dependency impact value = 0.5. The indirect dependency impact value is calculated based on the number of intermediate sandboxes, using the formula "Indirect dependency impact value = Direct impact value × 0.5". kThe indirect impact is calculated only when there is at least one intermediate sandbox, where k is the number of intermediate sandboxes. k ≥ 1, meaning indirect impact is calculated only when there is at least one intermediate sandbox. When k = 0, it is a direct dependency, and the indirect impact value is 0. For example, if sandbox i affects sandbox j through one intermediate sandbox, the direct impact value is 0.5, then the indirect dependency impact value = 0.5 × 0.5 = 0.25. The time decay coefficient is calculated based on the occurrence time of the dependency relationship, using the formula: "Time decay coefficient = e -t' / τ ", where t' is the time difference between the current time and the time when the dependency occurs, in minutes, and τ is the decay constant. For example, if τ is 60 minutes and t' = 30 minutes, then the time decay coefficient = e -30 / 60 ≈0.606.

[0099] The influence coefficient is calculated using the formula: "Influence Coefficient = Basic Influence Weight × (Direct Dependency Influence Value + Indirect Dependency Influence Value) × Time Decay Coefficient". The rows and columns of the dependency influence matrix correspond to all sandboxes in the system. For example, in a system containing sandboxes A, B, and C, the matrix is ​​3×3, with M... AB This represents the influence coefficient of sandbox A on sandbox B. Traditional techniques rely on influence but lack quantitative models, making it difficult to accurately assess the scope of influence. In contrast, the influence matrix, through comprehensive calculation of multiple factors, allows the influence coefficient to be directly used for mathematical analysis.

[0100] Steps S24-S25, through the synergy of dynamic threshold classification and influence matrix quantification, transform dependency relationships from qualitative description to quantitative analysis, overcoming the limitations of static dependency graphs. The slope judgment of strong dependencies, combined with the time dimension (Z-axis), can identify the potential risk of a "sudden increase in dependency strength within a short period." For example, if the slope of a dependency edge rises from 0.2 to 0.9 within 5 minutes, exceeding the set slope threshold of 0.6, it indicates possible abnormal data transmission, a dynamic change that traditional static analysis cannot capture. The combination of hop count statistics for deep dependencies and exponential decay of influence coefficients makes the indirect influence weaken non-linearly as the path lengthens, better reflecting the attenuation law of data transmission in actual systems and avoiding overestimation or underestimation caused by traditional linear decay. The closed-loop detection and special weight assignment of cyclic dependencies enable the matrix to reflect the characteristic that "cyclic dependencies amplify the scope of influence." For example, in a sandbox A→B→C→A cyclic dependency, the influence coefficient will have a cumulative effect due to the closed-loop characteristic, which cannot be achieved by single dependency analysis. This provides a high-precision quantitative foundation for subsequent steps, making time-series coordination and influence isolation more targeted.

[0101] Step S20 constructs a dynamic dependency mapping mechanism through the entire process of dependency 3D modeling, fingerprint identification, trajectory drawing, dynamic classification, and matrix quantization, which solves the problem that static dependency graphs cannot reflect runtime changes. The combination of the time dimension (Z-axis) of the three-dimensional spatial coordinate system of dependencies with the timestamps of sandbox dependency fingerprints enables the tracing of historical dependencies. For example, by querying the dependency trajectory path in the negative direction of the Z-axis, the cumulative impact of a data transmission error in a sandbox 3 hours ago on the current dependency chain can be found, which is impossible with two-dimensional graphs. The combination of the geometric features of dependency trajectory paths with the mathematical calculation of dynamic thresholds makes dependency classification spatially interpretable. For example, strong dependencies with abnormal slopes are represented as "steeply rising trajectory lines" in three-dimensional space, which is easy to identify intuitively. The quantitative results of the dependency impact matrix are combined with the fork risk assessment report, which enables the dependency impact of high-risk sandboxes to be accurately calculated. For example, the impact coefficients of high-risk sandbox A marked in the report on downstream sandboxes B and C can be directly queried through the matrix, providing a priority basis for risk disposal. Step S20 promotes the transformation of dependency management from passive response to proactive prediction, laying a dynamic adaptation foundation for the temporal coordination and risk isolation of the entire solution.

[0102] Step S30: Construct a multi-sector timing coordination disk based on the dependency influence matrix, and obtain the timing correctness guarantee state based on the multi-sector timing coordination disk;

[0103] Step S30 constructs a multi-sector timing coordination disk based on the dependency influence matrix. Through timing coordinate allocation, conflict locking, verification chain creation and compensation mechanisms, the timing correctness guarantee state is obtained, which solves the problem of operational disorder caused by the lack of global timing coordination when multiple data providers concurrently create save points in traditional technologies.

[0104] Further, step S30 includes:

[0105] Step S31: Design a multi-sector timing coordination disk based on the dependency influence matrix; each sector represents a time slice;

[0106] The multi-sector timing coordination disk divides 24 hours into W equal sectors, centered on the current time. The sector angle θ* is 360 degrees divided by W, and the radius R represents the time span; for example, R=1 corresponds to 1 hour. The value of W is determined based on the statistical analysis of historical savepoint operation frequency: when the system generates an average of 12 savepoint operations per hour, W=288, meaning each sector corresponds to 5 minutes (24×60 / 5=288), ensuring that each sector can accommodate a single operation without overlap. If the operation frequency decreases, W decreases; for example, with 6 operations per hour, W=144, and the sector corresponds to 10 minutes. Sectors are numbered clockwise, from 1 to W starting from the current time, with the numbering increasing in the same direction as the future time flow. Sectors corresponding to sandbox operations with higher impact coefficients in the dependency matrix are marked as "high-priority areas" and used preferentially in subsequent allocations. Traditional linear timelines are difficult to visualize the spatial relationships of concurrent operations, while the disk structure makes time cyclical, making the conflicts between adjacent time slices visible and facilitating the quick location of overlapping operations; without this step, the time coordinates of step S32 would lose their spatial carrier, resulting in a lack of global perspective in time management.

[0107] Step S32: Based on the multi-sector timing coordination disk and the dependency influence matrix, prioritize the save point operation and assign it a timing coordinate in polar coordinate form;

[0108] Further, step S32 includes:

[0109] Step S321: Save point operations are divided into emergency recovery operations, regular save operations, and test save operations. Based on the save point operation type, corresponding basic priority values ​​are assigned, and the basic priority values ​​are used as the radial coordinate r of the time series coordinates in polar coordinate form.

[0110] Step S322: Map the expected execution time of the save point operation to the corresponding sector of the multi-sector timing coordination disk, and calculate the sector angle as the angle coordinate θ of the timing coordinate in polar coordinate form;

[0111] Step S323, adjust the basic priority value according to the dependency influence matrix: when the influence coefficient M of sandbox i on sandbox j ij When the impact exceeds the preset threshold, increase the base priority value of the save point operation in sandbox i.

[0112] The save point operation refers to the operation of creating a state snapshot during the running of the sandbox, including emergency recovery operations, such as immediately restoring the state due to the risk of fork, regular save operations, such as periodic backups, and test save operations, such as temporary snapshots during function verification. Step S321 assigns a basic priority value as the radial coordinate r for different operation types: the basic priority value of the emergency recovery operation is the largest, for example, r = 1, because it directly relates to data consistency and requires the highest priority; the basic priority value of the regular save operation is the second, such as r = 0.5, which is required for daily maintenance; the basic priority value of the test save operation is the smallest, such as r = 0.1, with the lowest priority to avoid occupying key resources. The basic priority value is determined by the analytic hierarchy process. Invite multiple operation and maintenance experts to make pairwise comparisons on the operation urgency, construct a judgment matrix to calculate the weights, and confirm reasonableness through consistency test. Step S322 maps the expected execution time to the corresponding sector to obtain the angular coordinate θ: If the expected execution time is 10 minutes after the current moment, and the disk W = 288, then each sector is 5 minutes, the current moment is sector 1, 10 minutes corresponds to 2 sector intervals, so the corresponding sector number is 3, θ = 3×(360 / 288) = 3.75 degrees. At the same time, adjust the basic priority value according to the dependency impact matrix: When the influence coefficient M ij of sandbox i on sandbox j is greater than the preset influence threshold, increase the basic priority value of the save point operation of sandbox i. The setting basis of the influence threshold is: statistically analyze the distribution characteristics of the influence coefficients in the dependency impact matrix in the past 90 days, and take the critical value that can distinguish high-impact and general-impact operations, such as the 75% quantile, to ensure that high-impact operations can obtain timing resources first. For example, when the influence coefficient M ij of sandbox i on sandbox j > 0.6, the basic priority value r of the save point operation of sandbox i increases by 0.2 (the upper limit is 1) to ensure that high-impact operations are executed first. The traditional priority is only based on the operation type and ignores the influence of the dependency relationship. This step makes the priority more in line with the actual association of the system through matrix adjustment; if this step is missing, the conflict detection in step S33 will not be able to quantify the conflict degree due to the lack of space-time coordinates.

[0113] Step S33, based on the timing coordinates, check for operation conflicts in adjacent sectors, apply the priority arbitration rule to perform timing locking, and generate a conflict-free operation sequence;

[0114] The determination condition for operation conflicts in adjacent sectors is: there exists the timing coordinates (r', θ') of other save point operations, satisfying the absolute value of the angle difference |θ - θ'| < E and there is a dependency relationship between the two sandboxes in the dependency impact matrix, that is, M ij > 0 or M ji>0. E represents the number of adjacent sector checks, determined based on the average operation execution time. For example, if the average operation lasts 3 minutes, W = 288 (5 minutes per sector), then E = 1, ensuring no conflicts within 5 minutes before and after; if it lasts 8 minutes, E = 2, covering 10 minutes before and after. When a conflict is detected, for two conflicting savepoint operations, the savepoint operation with higher priority maintains its original coordinates, while the savepoint operation with lower priority is delayed to the nearest conflict-free sector. For example, if operation A (r = 0.7, θ = 3.75 degrees) conflicts with operation B (r = 0.5, θ = 5 degrees), and operation A's priority r = 0.7 is higher than operation B's priority r = 0.5, then operation B is delayed to the next sector. Delay calculation is performed by traversing sectors θ+E, θ+2E... until the first conflict-free location is found. Traditional conflict resolution relies on manual intervention, while step S33 automates the process through rule-based arbitration, reducing latency. Without this step, concurrent operations will result in inconsistent data states due to conflicts, affecting the accuracy of subsequent savepoint recovery.

[0115] Step S34: For each savepoint operation in the conflict-free operation sequence, generate a timing consistency verification chain;

[0116] Each node in the temporal consistency verification chain contains a verification record, which includes a sandbox identifier, temporal coordinates, and a preceding hash value. The sandbox identifier is the unique ID of the sandbox, and the preceding hash value is the SHA-256 hash of the previous record. The verification records are concatenated in chronological order of execution, with the head representing the system's initial state (r=0, θ=0) and the tail representing the latest save point operation record. After a new operation is completed, the hash of the current record is calculated and added to the tail of the chain, while the overall chain hash is updated to ensure that records in the chain are immutable. If a record is modified, its hash and all subsequent hashes will change, and anomalies can be detected by comparing the overall chain hashes. Traditional logs are easily tampered with and difficult to trace, while this chain achieves temporal auditability through hash linking; without this step, the temporal disorder detection in step S35 would lose its original data basis.

[0117] Step S35: Traverse the timing consistency verification chain to detect timing errors, derive the timing correction vector based on the dependency influence matrix, distribute correction instructions through the blockchain network to complete timing compensation, and obtain the timing correctness guarantee state.

[0118] Timing disorder means that the completion time of a subsequent operation is earlier than that of a previous operation. For example, in the timing consistency verification chain, the completion time of the 5th record is t5, the 6th is t6, and t6 < t5. During detection, traverse the records in the timing consistency verification chain, compare the completion timestamps of adjacent records, and mark the disordered positions. The timing correction vector contains a list of operations to be adjusted and the adjustment amplitude: the list of operations contains disordered and affected operations, and the adjustment amplitude is calculated as "the time difference between the disordered operation and the correct timing × the influence coefficient weight", where the influence coefficient weight is the ratio of the inter-sandbox influence coefficient to the maximum influence coefficient, ensuring that the adjustment amplitude changes dynamically with the influence intensity. For example, if the operation disorder in sandbox A causes a time difference of 3 minutes, and the influence coefficient M AB = 0.4 between sandbox B and A, and the maximum influence coefficient of the system is 1, then the adjustment amplitude of sandbox B = 3 minutes × 0.4 = 1.2 minutes; ensuring that the weaker the influence of the sandbox, the smaller the adjustment amplitude, avoiding new timing conflicts caused by overcorrection.

[0119] The timing correction vector is broadcast through the blockchain network. After the relevant sandboxes receive it, they adjust the execution order and recovery points to ensure timing consistency. Traditional timing errors require manual troubleshooting, while this mechanism automatically corrects through vectors, improving efficiency; without this step, timing disorders will accumulate, resulting in the inavailability of the save point state and affecting the isolation accuracy of S40. The timing correctness guarantee state refers to the stable state after the coordination, locking, verification, and compensation of steps S31 - S35, where the timing relationships of all save point operations conform to the logical sequence and are consistent with the dependency influence matrix, manifested as no disordered records in the timing consistency verification chain, providing a timing benchmark for the influence scope isolation of step S40.

[0120] Step S30 realizes the global coordination of timing operations through the collaboration of disk spatialization management, dynamic coordinate allocation, automatic conflict locking, on-chain verification, and vector compensation, solving the problem of operation disorders caused by the lack of global timing coordination when multiple data providers concurrently create save points in traditional technologies. The spatialization of the disk structure transforms the timing conflict from an abstract time difference into a concrete angular difference, and the conflict source can be quickly located through geometric calculations. Compared with the traditional linear time that requires comparing timestamps one by one, the efficiency is significantly improved; the combination of the verification chain and the dependency influence matrix enables the timing legality of each operation to be double-verified through "hash verification + influence coefficient tracing". For example, when an operation is tampered with, not only is the hash abnormal, but its matching degree with the associated sandbox's influence coefficient also decreases, facilitating the positioning of the tampering point. Step S30 promotes the transformation of timing management from passive processing to active prevention, providing an accurate timing benchmark for the isolation mechanism of S40 and ensuring the accuracy of the risk diffusion range calculation.

[0121] Step S40 utilizes the timing correctness guarantee state and the dependency influence matrix to execute the influence scope adaptive isolation mechanism.

[0122] Step S40 addresses the issues of over-isolation or under-isolation caused by the fixed scope of traditional static isolation by dynamically defining isolation areas, implementing differentiated strategies, and monitoring recovery in real time, thereby ensuring precise control of the spread of fork risk.

[0123] Further, step S40 includes:

[0124] Step S41: Based on the influence coefficients in the dependency influence matrix and the temporal correctness guarantee status, construct an influence diffusion simulation model and delineate the core isolation region and the edge isolation region;

[0125] Specifically, sandboxes with influence coefficients greater than the upper limit of a preset isolation threshold range in the dependency influence matrix are designated as core nodes. Combining the temporal correlation characteristics of points with temporal discrepancies, a Monte Carlo simulation is used to deduce the risk diffusion path, incorporating sandboxes covered by the path into the core isolation region. The marginal isolation region includes sandboxes outside the core region whose influence coefficients are within the isolation threshold range and have an indirect dependency relationship with the core region. The upper limit of the isolation threshold range is determined based on the historical influence coefficient distribution: influence coefficient data from all sandboxes over the past 6 months are collected, and their 90th percentile is calculated as the upper limit of the isolation threshold range. For example, if the statistical result is 0.6, then sandboxes with influence coefficients greater than 0.6 are considered core nodes. The lower limit of the isolation threshold range is determined by collecting influence coefficient data between sandboxes in scenarios without bifurcation risks and analyzing their distribution. A critical value that can filter out the vast majority of low-impact dependencies is selected as the lower limit to distinguish between low-impact and medium-impact dependencies and avoid including irrelevant sandboxes, such as 0.3. The isolation threshold range is set to 0.3-0.6 as the criterion for determining the marginal region. The Monte Carlo simulation process for deducing the risk diffusion path is as follows: starting from the core node, based on M in the dependency influence matrix... ij The magnitude of M determines the diffusion probability. ij When M = 0.8, the diffusion probability is 80%. ij When the impact factor is 0.4, the risk rate is 40%. Each simulation randomly generates 1000 possible diffusion paths. After 100 simulations, sandboxes with a path coverage frequency exceeding 70% are included in the core isolation area. The delineation of the edge isolation area must simultaneously meet two conditions: the impact coefficient must be between 0.3 and 0.6, and there must be an indirect dependency on the core isolation area through at least one intermediate sandbox, with the number of intermediate sandboxes ≤ 2 to avoid an excessively large range. The temporal correlation characteristics of the time-series disorder points refer to sandboxes that have interacted with the core node within the past hour; these sandboxes are prioritized for inclusion in the simulation due to time priority. Traditional isolation methods rely on fixed rules to define the range, failing to reflect real-time dependency strength and temporal correlation. This model, however, combines statistical simulation with dynamic thresholds to make the isolation area more closely resemble the actual risk propagation patterns. Without this step, the isolation range will either be too large, leading to business interruption, or too small to prevent risk propagation, thus losing the core meaning of isolation.

[0126] Step S42: Implement differentiated isolation strategies for core isolation areas and edge isolation areas;

[0127] The core isolation zone employs a "read-write dual-lock" mechanism: data input / output operations within the sandbox are paused, savepoint creation is frozen, and blockchain notarization is triggered. Every 5 minutes, a SHA-256 hash is calculated for the current isolation state, and the hash value is written to a block on the consortium blockchain, ensuring the isolation process is traceable and tamper-proof. Pausing data input / output operations within the sandbox is achieved by intercepting operating system I / O system calls, and freezing savepoint creation disables the savepoint generation interface. The current isolation state includes the sandbox identifier, isolation start time, and sandbox operation feature vector.

[0128] Dynamic rate limiting is implemented in edge isolation areas: data interaction bandwidth is adjusted based on the time decay coefficient in the dependency impact matrix. The calculation formula is "current bandwidth = baseline bandwidth × time decay coefficient", where the baseline bandwidth is the average bandwidth of the sandbox under normal conditions. Minimal synchronous communication is maintained, such as one status heartbeat per second, to maintain timing consistency with other sandboxes and avoid timing discrepancies caused by complete isolation. The differentiated strategy is based on the fact that core areas have rapid risk propagation and require mandatory blocking; edge areas have weaker risks and can balance isolation and business continuity through rate limiting. Without this step, a unified isolation strategy would result in incomplete isolation in core areas or excessive restriction in edge areas, impacting system availability.

[0129] Step S43: Monitor the changes in the sandbox operation feature vector within the isolation area in real time, and assess the isolation effect in conjunction with the timing correctness assurance status.

[0130] Real-time monitoring of sandbox operation feature vector changes within the isolation area is conducted, with the monitoring frequency consistent with the 30-second sampling period in step S13. The safe regression range of the sandbox operation feature vector refers to the feature points mapped by Fr and Fw in the sandbox activity coordinate system falling outside the square fork warning area of ​​step S12, and Ts≥Tmin, Id≤K times the historical average. If there are no anomalies for three consecutive time slices (90 seconds), the sandbox operation feature vector stabilizes within the safe range, ensuring that the risk has been substantially reduced. The order of de-isolation follows "edge areas first, impact coefficient decreasing": first de-isolate the sandboxes with the lowest impact coefficients in the edge isolation areas, then gradually de-isolate the core isolation areas, with a 30-second interval between each de-isolation to observe whether new risks are triggered. After de-isolation, a timing consistency verification chain is used for verification: the deviation between the operation timing coordinates of the sandbox after de-isolation and the previous records in the chain is calculated. If the deviation is ≤1 sector angle, timing coordination is determined. The final isolation-recovery closed-loop record includes the isolation start time, scope, strategy, de-isolation time, and verification result. The hash value of this record is written to the blockchain for permanent storage. Traditional recovery processes lack rigorous verification mechanisms, which can easily lead to a rebound in risk due to premature termination. This step, however, ensures the safety of recovery through multi-dimensional monitoring and time-series verification.

[0131] Step S40 achieves adaptive isolation of the impact range through the synergy of dynamic range delineation, differentiated strategies, and closed-loop monitoring, precisely controlling the spread of fork risk. The impact diffusion simulation model, combined with the timing correctness assurance status, identifies hidden risk paths with "low time decay coefficients but high impact coefficients." For example, an interaction between a sandbox and the core isolation area occurred 2 hours ago, but the impact coefficient is 0.7. Traditional models would ignore this, but the impact diffusion simulation model, by incorporating timing correlation features, includes it in the edge area, avoiding delayed risk spread. The bandwidth adjustment of dynamic rate limiting is linked to the time decay coefficient, so that the isolation strength of the edge isolation area automatically decreases with the historical duration of the dependency relationship. For example, a dependency relationship from 1 hour ago is rate-limited by 50%, and a dependency relationship from 3 hours ago is rate-limited by 30%, which controls risk and reduces the impact on old dependent services. The verification mechanism during recovery, combined with the timing consistency verification chain in step S34, can identify timing conflicts that may be caused by the de-isolation operation itself. For example, if the timing of an operation after a sandbox is de-isolated deviates from the previous record by 2 sectors, readjustment can be used to avoid secondary risks. Step S40 upgrades the isolation mechanism from passive defense to proactive and dynamic risk management, forming a complete closed loop with the fork risk warning in step S10, the dependency mapping in S20, and the timing coordination in S30. This ensures that the entire system can accurately isolate risks while minimizing the impact on normal business operations when risks occur, significantly improving the system's risk resistance and business continuity.

[0132] Example 2

[0133] This embodiment, based on Embodiment 1, provides a blockchain-based enterprise trusted data access system, such as... Figure 6 As shown, it includes:

[0134] Fork Warning Module: Used to construct a sandbox activity coordinate system, set a fork warning area in the sandbox activity coordinate system, identify potential fork risks through the fork warning area and generate a fork risk assessment report;

[0135] Dependency Mapping Module: Used to extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependency relationships, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependency relationships, and derive the dependency impact matrix from the dependency trajectory paths;

[0136] Timing coordination module: used to construct a multi-sector timing coordination disk based on the dependency influence matrix, and to obtain the timing correctness guarantee state based on the multi-sector timing coordination disk;

[0137] Adaptive Isolation Module: Utilizes timing correctness to ensure state and dependency influence matrix, and executes an adaptive isolation mechanism for the scope of influence.

[0138] Furthermore, in the dependency mapping module, the method for generating sandbox dependency fingerprints includes:

[0139] Step S221: Collect the input dataset Din and output dataset Dout for each sandbox at time t;

[0140] Step S222: Calculate the feature hash values ​​of the input dataset and the output dataset to generate the input feature hash value and the output feature hash value;

[0141] Step S223: The input feature hash value, the output feature hash value, and time t are fused together using a hash combination function to generate a unique sandbox dependency fingerprint.

[0142] Furthermore, in the timing coordination module, the construction of a multi-sector timing coordination disk based on the dependency influence matrix, and the obtaining of the timing correctness guarantee state based on the multi-sector timing coordination disk, include:

[0143] Step S31: Design a multi-sector timing coordination disk based on the dependency influence matrix; each sector represents a time slice;

[0144] Step S32: Based on the multi-sector timing coordination disk and the dependency influence matrix, prioritize the save point operation and assign it a timing coordinate in polar coordinate form;

[0145] Step S33: Based on the timing coordinates, check for operation conflicts in adjacent sectors, apply priority arbitration rules to perform timing locking, and generate a conflict-free operation sequence;

[0146] Step S34: For each savepoint operation in the conflict-free operation sequence, generate a timing consistency verification chain;

[0147] Step S35: Traverse the timing consistency verification chain to detect timing errors, derive the timing correction vector based on the dependency influence matrix, distribute correction instructions through the blockchain network to complete timing compensation, and obtain the timing correctness guarantee state.

[0148] The methods and systems of this application may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of this application are not limited to the order specifically described above, unless otherwise specifically stated.

[0149] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0150] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A blockchain-based method for accessing trusted enterprise data, characterized in that: The method includes: A sandbox activity coordinate system is constructed, with the current moment as the origin, the horizontal axis X as the time axis (positive direction points to future time, negative direction points to historical time), and the vertical axis Y as the data change frequency deviation axis (positive direction indicates that the data change frequency is increasing relative to the baseline value, negative direction indicates that the data change frequency is decreasing relative to the baseline value). A fork warning area is set in the sandbox activity coordinate system to identify potential fork risks and generate a fork risk assessment report. Extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependencies, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependencies, and derive the dependency impact matrix from the dependency trajectory paths. A multi-sector time-series coordination disk is constructed based on the dependency and influence matrix. The multi-sector time-series coordination disk divides 24 hours into multiple equal sectors with the current time as the center. The radius of the multi-sector time-series coordination disk represents the time span. The time-series correctness guarantee state is obtained based on the multi-sector time-series coordination disk. By leveraging the temporal correctness guarantee of the state and dependency influence matrix, an adaptive isolation mechanism for the scope of influence is implemented.

2. The blockchain-based enterprise trusted data access method according to claim 1, characterized in that, The bifurcation warning area is a square with the origin of the sandbox activity coordinate system as its geometric center. The coordinates of the four vertices are (-M' / 2,-M' / 2), (-M' / 2,M' / 2), (M' / 2,M' / 2), and (M' / 2,-M' / 2), respectively, where M' is the side length of the square bifurcation warning area.

3. The blockchain-based enterprise trusted data access method according to claim 2, characterized in that, The method for identifying potential fork risks through fork warning areas includes: Continuously monitor the sandbox's data read frequency, data write frequency, savepoint creation interval, and sandbox-dependent interaction strength to form a sandbox operation feature vector sequence; A three-layer conditional judgment for bifurcation warning detection is performed on the sandbox operation feature vector sequence to identify potential bifurcation risks; the first layer of the three-layer conditional judgment is the judgment within the region, the second layer is the judgment of increasing trend, and the third layer is the judgment of threshold exceeding limit.

4. The blockchain-based enterprise trusted data access method according to claim 3, characterized in that, The method for generating the sandbox-dependent fingerprint includes: Collect the input and output datasets for each sandbox at time t; Calculate the feature hash values ​​of the input and output datasets to generate the input feature hash value and the output feature hash value; The input feature hash value, the output feature hash value, and time t are fused together using a hash combination function to generate a unique sandbox dependency fingerprint.

5. The blockchain-based enterprise trusted data access method according to claim 4, characterized in that, The method for drawing the dependent trajectory path includes: By leveraging sandbox dependency fingerprints to identify data consumption relationships between sandboxes, the data provider sandbox and the data consumer sandbox can be determined. Based on the dependency relationship three-dimensional spatial coordinate system, obtain the three-dimensional spatial coordinate positions of the data provider sandbox and the data consumer sandbox; Directed edge connections are created between the three-dimensional spatial coordinates of the data provider sandbox and the data consumer sandbox to form a dependent trajectory path.

6. The blockchain-based enterprise trusted data access method according to claim 5, characterized in that, The methods for determining the data provider sandbox and the data consumer sandbox include: For any two sandboxes i' and j', by comparing the input feature hash value of sandbox j' with the output feature hash value of sandbox i', if the first 48 bits of the two are the same, it is determined that sandbox j' has consumed the data of sandbox i'. Sandbox i' is the data provider sandbox, and sandbox j' is the data consumer sandbox. Here, i' and j' are the index variables of the sandboxes, and i' ≠ j'.

7. The blockchain-based enterprise trusted data access method according to claim 6, characterized in that, The method for deriving the dependency influence matrix from the dependency trajectory path includes: The geometric feature parameters of the dependent trajectory path are analyzed, and based on the geometric feature parameters, a dynamic threshold criterion is applied to identify three types of dependency relationships: strong dependency, deep dependency, and circular dependency. By integrating the three types of identified dependencies, a dependency influence matrix is ​​constructed to describe the mutual influence between sandboxes. The matrix element values ​​in the dependency influence matrix represent the influence coefficients.

8. The blockchain-based enterprise trusted data access method according to claim 7, characterized in that, The method for identifying three types of dependency relationships—strong dependency, deep dependency, and circular dependency—using the dynamic threshold criterion includes: Calculate the slope of each directed edge in the dependent trajectory path. When the absolute value of the slope exceeds the set slope threshold, it is marked as a strong dependency type. Calculate the length of the dependency trajectory path from the source sandbox to the target sandbox. When the number of hops through the sandboxes traversed by the dependency trajectory path exceeds the set hop threshold, it is marked as a deep dependency type. Detect whether the dependency path forms a closed loop. When it starts from the starting sandbox and eventually returns to the starting sandbox, it is marked as a circular dependency type.

9. A blockchain-based enterprise trusted data access system, used to implement the blockchain-based enterprise trusted data access method according to any one of claims 1-8, characterized in that, The system includes: Fork Warning Module: Used to construct a sandbox activity coordinate system, set a fork warning area in the sandbox activity coordinate system, identify potential fork risks through the fork warning area and generate a fork risk assessment report; Dependency Mapping Module: Used to extract the list of affected sandboxes from the fork risk assessment report, create a three-dimensional coordinate system of dependency relationships, generate sandbox dependency fingerprints and draw dependency trajectory paths within the three-dimensional coordinate system of dependency relationships, and derive the dependency impact matrix from the dependency trajectory paths; Timing coordination module: used to construct a multi-sector timing coordination disk based on the dependency influence matrix, and to obtain the timing correctness guarantee state based on the multi-sector timing coordination disk; Adaptive Isolation Module: Utilizes timing correctness to ensure state and dependency influence matrix, and executes an adaptive isolation mechanism for the scope of influence.

Images