Data tracing method and system based on big data and multi-dimensional characteristics of blockchain

Abstract

The present invention discloses a data tracing method and system based on big data and multidimensional features of blockchain, which belongs to the field of big data technology, including: obtaining a target data set; calculating the responsibility value of each relevant node and the interaction weight between adjacent nodes, and constructing a multidimensional feature matrix; inputting the multidimensional feature matrix into a graph convolutional network for feature fusion, and constructing a traceability graph with relevant nodes as vertices and interaction weights as edges; calculating the responsibility propagation probability of each node in the traceability graph, and determining abnormal data; determining the responsible node according to the transmission path of the abnormal data, and calculating the final responsibility score of each responsible node; attributing responsibility according to the final responsibility score, and constructing a node-edge weight visualization model according to the responsibility attribution result for visual traceability. The present application solves the problem of uncertainty in responsibility attribution in traditional solutions, and improves the accuracy and practicality of the traceability method.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to a data tracing method and system based on big data and multi-dimensional features of blockchain. Background Art

[0002] With the rapid development of information technology, big data technology has been widely used in all walks of life, realizing the integration and analysis of massive, multi-source, and heterogeneous data. This technological innovation has greatly promoted the intelligence of complex systems and the improvement of decision support capabilities. However, in practical applications, problems such as data inconsistency, unclear sources, and error propagation have seriously affected the reliability of the system and the accuracy of analysis results.

[0003] The diversity and complexity of data make it very easy for deviations and errors to occur during data collection, transmission, processing and analysis. These problems not only come from defects in the data itself, but may also be closely related to each link in the data transmission path. Therefore, how to effectively track the source, transmission path and change process of data has become the key to ensuring data quality and the accuracy of analysis results.

[0004] In response to the above problems, the industry has proposed the concept of data traceability, which aims to achieve monitoring and management of the entire life cycle of data by establishing a complete data tracking mechanism. Existing data traceability methods usually include data collection and preprocessing, multi-dimensional feature modeling, dynamic traceability path construction, anomaly detection and responsibility attribution, and visual traceability and traceability result storage.

[0005] During the data collection and preprocessing phase, the system obtains data flow information from multiple heterogeneous data sources, cleans and normalizes the format, and generates a unique identifier for each piece of data for subsequent tracking. Multidimensional feature modeling constructs a multidimensional feature matrix based on data content, transmission path, and context features, uses a deep learning model to embed features into the data, and represents the multidimensional features of the data in the form of a low-dimensional vector, providing a basis for subsequent traceability analysis.

[0006] In the dynamic traceability path construction phase, the system uses a graph-based data traceability model to establish a dynamic traceability path based on the data source and target node, and uses an improved graph neural network to extract the relationship features between the paths to achieve global tracking of the data flow. This step is crucial for understanding the data propagation path and change process.

[0007] In the anomaly detection and responsibility attribution stage, the system establishes a data flow anomaly detection mechanism based on historical data and feature models. When abnormal data is detected, the path backtracking algorithm is used to identify the responsible node and generate a data traceability report. However, when the data transmission path contains multiple possible responsible nodes, the responsibility attribution based on a simple backtracking algorithm may not be able to accurately determine the single responsible party, which is a major challenge facing current data traceability technology.

[0008] Finally, the visual traceability and traceability result storage stage displays the data flow and traceability path through a graphical interface, supports the query, storage and historical comparative analysis of traceability results, and provides an intuitive and convenient tool for data quality management and decision-making.

[0009] In summary, although the existing data traceability methods have solved the problems of data inconsistency, unknown sources and error propagation to a certain extent, they are still insufficient in the accuracy of responsibility attribution. Summary of the invention

[0010] The purpose of the present invention is to provide a data tracing method and system based on big data and multi-dimensional characteristics of blockchain, so as to solve the problem of inaccurate attribution of responsibility during data tracing in the prior art.

[0011] To achieve the above objectives, this application adopts the following technical solutions:

[0012] In the first aspect, the present application provides a data tracing method based on big data and multi-dimensional features of blockchain, comprising the following steps:

[0013] Acquire a target data set, wherein the set includes a plurality of target data records and a context state of each target data record and a historical behavior record of a related node, wherein the target data record includes data content and a transmission path;

[0014] Calculate the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and construct a multidimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight;

[0015] Input the multidimensional feature matrix into a graph convolutional network for feature fusion to obtain a global feature graph, and construct a traceability graph with relevant nodes as vertices and interaction weights as edges based on the global feature graph;

[0016] Calculate the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and use a multi-level detection mechanism to determine abnormal data in the target data set;

[0017] Determine the responsible node in the traceability graph according to the transmission path of the abnormal data, and calculate the final responsibility score of each responsible node according to the responsibility propagation probability and historical behavior records;

[0018] Responsibility is attributed according to the final responsibility score, and a node-edge weight visualization model is constructed according to the responsibility attribution result for visual traceability.

[0019] In the second aspect, the present application provides a data traceability system based on big data and multi-dimensional features of blockchain, including:

[0020] An acquisition module, used to acquire a target data set, wherein the set includes a plurality of target data records and the context state of each target data record and the historical behavior records of its related nodes, wherein the target data record includes data content and a transmission path;

[0021] A first construction module is used to calculate the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and to construct a multidimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight;

[0022] A second construction module is used to input the multidimensional feature matrix into a graph convolutional network for feature fusion to obtain a global feature map, and to construct a traceability graph with relevant nodes as vertices and interaction weights as edges according to the global feature map;

[0023] A detection module, used to calculate the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and to determine the abnormal data in the target data set using a multi-level detection mechanism;

[0024] A calculation module, used to determine the responsibility node in the traceability graph according to the transmission path of the abnormal data, and calculate the final responsibility score of each responsibility node according to the responsibility propagation probability and historical behavior records;

[0025] A visualization module is used to attribute responsibility according to the final responsibility score, and to build a node-edge weight visualization model according to the responsibility attribution result for visual tracing.

[0026] In a third aspect, the present application provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to implement a data tracing method based on big data and multi-dimensional features of blockchain as described in any one of the above when executed.

[0027] The present invention has the following beneficial effects:

[0028] By introducing the responsibility propagation probability model and Bayesian reasoning, the accuracy of determining the responsible nodes in the traceability path is enhanced, and the multi-dimensional feature modeling is optimized by combining contextual information and historical behavior characteristics. At the same time, the final responsibility score of the node is calculated based on the dynamic responsibility allocation model, which supports hierarchical attribution and generates intuitive visual results, solving the problem of uncertain responsibility attribution in traditional solutions and improving the accuracy and practicality of the traceability method. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 This is a flow chart of a data tracing method based on big data and multi-dimensional features of blockchain provided in an embodiment of the present application;

[0031] Figure 2 It is a multi-dimensional feature matrix relationship diagram provided by an embodiment of the present application;

[0032] Figure 3 This is a schematic diagram of a dynamic tracing path provided in an embodiment of the present application;

[0033] Figure 4 is a schematic diagram of a visualization interface provided in an embodiment of the present application;

[0034] Figure 5 This is a structural example diagram of a data traceability system based on big data and multi-dimensional features of blockchain provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] To make the technical solution of the present application clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The terms "first", "second", etc. in the claims and specification of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances. This is only to describe the distinction method used when describing objects with the same attributes in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that a process, method, system, product or device containing a series of units is not necessarily limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or devices.

[0036] In one embodiment, if Figure 1As shown, the present application provides a data tracing method based on big data and multi-dimensional features of blockchain, comprising the following steps:

[0037] S110, obtaining a target data set, the set including a number of target data records and the context state of each target data record and the historical behavior records of its related nodes, the target data record including data content and transmission path;

[0038] S120, calculating the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and constructing a multidimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight;

[0039] S130, inputting the multidimensional feature matrix into the graph convolutional network for feature fusion to obtain a global feature graph, and constructing a traceability graph with relevant nodes as vertices and interaction weights as edges according to the global feature graph;

[0040] S140, calculating the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and using a multi-level detection mechanism to determine abnormal data in the target data set;

[0041] S150, determining the responsible node in the traceability graph according to the transmission path of the abnormal data, and calculating the final responsibility score of each responsible node according to the responsibility propagation probability and historical behavior records;

[0042] S160. Perform responsibility attribution based on the final responsibility score, and build a node-edge weight visualization model based on the responsibility attribution result for visual traceability.

[0043] This embodiment includes five steps, namely data collection and preprocessing, multi-dimensional feature modeling, dynamic traceability path construction, anomaly detection and responsibility attribution, and visual traceability and traceability result storage.

[0044] The first step is data collection and preprocessing.

[0045] Data collection and preprocessing are the basis of the entire traceability process. It mainly involves extracting necessary information from multi-source heterogeneous data related to blockchain transactions and performing cleaning and normalization processing, so as to provide high-quality data support for subsequent multi-dimensional feature modeling and responsibility attribution through comprehensive and accurate data input.

[0046] Further, a number of original data records are extracted from the multi-source heterogeneous data streams, and each original data record is cleaned and format-standardized to obtain a target data record;

[0047] Obtain the context state of each target data record and the historical behavior records of each node related to it. The historical behavior records include the historical abnormality frequency and stability score.

[0048] The target data set is constructed based on each target data record, its context environment state, and the historical behavior records of each related node.

[0049] Data collection and preprocessing are mainly divided into two parts: data collection and cleaning and context information supplementation. Among them, data collection refers to extracting several original data records from multi-source heterogeneous data streams. Each original data record contains information such as timestamp, data content, source node and transmission path during data transmission. In this embodiment, the timestamp covers all key time points in the transaction process, including but not limited to the transaction creation timestamp and transaction verification timestamp. The data content specifically includes the transaction amount, the information of the transaction initiator and the results of public and private key verification. The source node is the blockchain node of the transaction initiator, and the transmission path is all the blockchain nodes that the data passes through during the transaction. At the same time, during the collection, each original data record will be assigned a unique identifier UID so that its source and flow can be traced in the subsequent traceability process, thereby realizing the tracing of data and the monitoring and analysis of data flow.

[0050] After data collection is completed, the data needs to be cleaned and formatted to obtain the target data records. The purpose is to eliminate noise data and duplicate data and ensure the consistency of data formats between various data sources. Among them, format normalization includes formatting the timestamp to ensure that the time data can be accurately parsed and used in the subsequent calculation and modeling process. Through data cleaning and format normalization, data quality can be effectively guaranteed, thereby ensuring that subsequent calculations and modeling are carried out efficiently and accurately.

[0051] The context information supplement is to further obtain the context environment status of the target data record and the historical behavior records of each node related to it based on the cleaned and normalized data. Among them, the context environment status includes data such as network delay, transmission bandwidth and node load, which are quantified by the context environment evaluation value, and the data such as network delay, transmission bandwidth and node load can be obtained through network monitoring tools and system performance monitoring tools, such as measuring the round-trip time of communication between nodes through network testing tools to determine network delay, obtaining the current available bandwidth from the interface provided by the network device or operating system to determine the transmission bandwidth, and obtaining the usage of resources such as CPU and memory from the node's resource management system to determine the node load; the node's historical behavior record includes historical abnormality frequency and stability score, which can be obtained from data sources such as system log records and previous abnormality detection results. For example, the historical abnormality frequency of the node is obtained by counting the ratio of the number of abnormalities that occurred in the node in the past period of time to the total number of operations, and the stability score is calculated based on the historical abnormality frequency. These context information provides important auxiliary decision-making basis for subsequent responsibility deduction.

[0052] Add all target data records, the context features of each target data record, and the historical behavior records of its related nodes to an empty set to obtain a target data set.

[0053] Furthermore, the context environment evaluation value corresponding to each context environment state is determined, and the corresponding historical abnormality frequency, stability score and context environment evaluation value are weighted averaged to obtain the responsibility value of each relevant node.

[0054] Another important algorithm in data collection and preprocessing is the historical behavior feature modeling of the responsible node. The purpose of this algorithm is to construct the responsibility value of each node in the transmission path by comprehensively evaluating the historical abnormal behavior, transmission environment and stability of the node. To this end, a weighted calculation formula as shown below is introduced to evaluate the responsibility value of the node, and in the subsequent steps, the responsibility propagation mechanism will trace and determine the responsibility of the data flow based on this responsibility value.

[0055] Specifically, .

[0056] Among them, R i represents the responsibility value of node i, H i is the historical abnormal frequency of node i, C i is the context evaluation value of node i, S i is the stability score of node i, w1, w2 and w3 are weight coefficients, reflecting the importance of each feature in responsibility assessment. i , C i , S iThe value range of w1, w2 and w3 is [0,1], and w1+w2+w3=1.

[0057] It should be noted here that the historical abnormal frequency is usually obtained by evaluating the historical behavior of the node within 3-6 months. This time period can cover a certain number of transactions or data transmission events, thereby more comprehensively reflecting the historical behavior of the node, and will not make the early data lose its timeliness or be overly diluted by recent data due to too long a time period.

[0058] For example, the historical abnormal frequency of node i is 0.1, the context evaluation value is 0.8, the stability score is 0.9, the weight coefficients w1=0.5, w2=0.3, w3=0.2, then the responsibility value of node i is:

[0059] .

[0060] The responsibility value can accurately reflect the degree of responsibility of the node in different data transmission scenarios, and will serve as the basis for subsequent traceability path construction and responsibility derivation.

[0061] The second step is multi-dimensional feature modeling.

[0062] Multidimensional feature modeling is the core step to ensure that features such as data sources, transmission paths, and node historical behaviors can be effectively integrated and analyzed in depth. The goal of this step is to establish a multidimensional feature matrix that can integrate data content features, transmission path features, context features, and features related to responsibility deduction, and based on this matrix, use a deep learning model to perform feature fusion to generate a global feature graph to provide support for subsequent traceability path construction and responsibility attribution.

[0063] Further, feature extraction is performed on the data content, transmission path, context environment state, and historical behavior records to obtain data content features, transmission path features, context environment features, and historical behavior features;

[0064] A multidimensional feature matrix is ​​constructed based on data content features, transmission path features, context environment features, historical behavior features and interaction weights.

[0065] In this step, feature extraction is first performed on the collected data content, transmission path and context information. For example, feature extraction is performed on the data content to obtain the legality of the amount and the public key matching rate. These features can better describe the impact of the data content on traceability and responsibility determination; the transmission path is extracted to obtain the path length and the transmission delay of each path. These features can better reflect the impact of the transmission path on data traceability; feature extraction is performed on the context information to obtain context environment features and historical behavior features. Among them, context environment features include node load and delay fluctuations, and historical behavior features include historical abnormal records. These features can better reflect the impact of the network environment on data traceability.

[0066] Because historical behavior features, contextual environment features, and interaction weights between nodes all play a key role in the responsibility deduction process, these features are also referred to as responsibility deduction-related features in this embodiment. Specifically, historical behavior features can reflect past abnormalities of nodes, contextual environment features can reflect the impact of the network environment, and interaction weights provide a basis for responsibility judgment from the perspective of the relationship between nodes. They are combined to provide rich and indispensable information for responsibility attribution.

[0067] like Figure 2 As shown, in order to effectively fuse the extracted features, in this embodiment, a multi-dimensional feature matrix is ​​first constructed based on the extracted features. Each dimension in the multi-dimensional feature matrix represents a feature of a certain aspect of the data stream, such as the historical behavior of the node, transmission delay, node load, etc., and then a graph convolutional network (GCN) is used for feature fusion. GCN can process the complex relationship between nodes through a graph structure, and effectively extract the dependency features between nodes to form a low-dimensional representation of the data.

[0068] In the process of multi-dimensional feature modeling, the interaction relationship between nodes is represented by graph structure weighting. Specifically, the interaction weight of each edge is dynamically adjusted based on factors such as the interaction frequency between adjacent nodes on the transmission path, transmission delay, and data transmission reliability.

[0069] Furthermore, the interaction frequency, transmission delay and reliability score between adjacent nodes on the transmission path are obtained, and the interaction weights between all adjacent nodes are calculated based on the interaction frequency, transmission delay and reliability score.

[0070] Assume that the interaction weight between node i and node j is w ij , the calculation formula is as follows:

[0071] .

[0072] Among them, f(E ij ,T ij ,Q ij) is the interaction frequency E based on node i and node j ij , transmission delay T ij and reliability score Q ij A comprehensive function that is used to calculate the interaction strength between two nodes. The specific function formula can be set according to actual needs. The denominator is the sum of the interaction strengths of node i and all its neighbor nodes N(i). u is an index variable used to traverse all neighbor nodes of node i, belonging to N(i).

[0073] It should be noted here that the reliability score is an indicator used to measure the reliability of interaction between adjacent nodes by comprehensively considering multiple factors such as the stability and accuracy of data transmission between the two.

[0074] The final interaction weight can more accurately reflect the relationship between paths and nodes during data flow transmission, so that GCN can better embed and fuse data.

[0075] For example, the interaction frequency between node i and node j is 0.8, the transmission delay is 0.5, the reliability score is 0.9, and the sum of the interaction strengths between node i and all its neighboring nodes is 3.5, then its interaction weight is:

[0076] .

[0077] The calculation result is f(0.8,0.5,0.9)=1.2, then

[0078] .

[0079] That is, the interaction weight of the edge between node i and node j in the graph is 0.343, and it is determined that there is a strong interaction relationship between the two nodes according to the actual situation. Through GCN, the interaction weight will affect the feature fusion of the nodes and improve the accuracy of determining the responsible nodes.

[0080] By integrating the historical behavior features, contextual features, and interaction weights of nodes into modeling, and combining graph convolutional networks to deeply embed complex features, we can fully capture the dynamic features of the data and improve adaptability to a variety of scenarios.

[0081] The third step is to build a dynamic traceability path.

[0082] like Figure 3 As shown in the figure, the purpose of dynamic traceability path construction is to model multiple nodes and paths through a graph structure so as to find the real responsible node among multiple possible paths through which the data flows. It also solves the problem that the responsibility attribution in the traditional solution cannot effectively judge the existence of multiple responsible nodes by introducing the responsibility propagation probability model.

[0083] Furthermore, the conditional probability and joint probability corresponding to the historical behavior characteristics and contextual environment characteristics of each relevant node under the given responsible party are obtained;

[0084] The prior responsibility probability of each relevant node is determined according to the responsibility value, and the responsibility propagation probability of each relevant node is calculated according to the conditional probability, joint probability and prior responsibility probability.

[0085] Among them, the introduction of the responsibility propagation probability model is to solve the problem of how to accurately calculate the possibility of each node being the responsible party when the data flow passes through multiple nodes. The model is based on the Bayesian network and constructs the responsibility causal chain through the historical behavior characteristics, contextual environment characteristics and responsibility propagation probability between nodes.

[0086] Specifically, the probability of responsibility propagation of node i in a certain path is calculated by the following formula:

[0087] .

[0088] Among them, P(R i ∣X i Y i ) represents the probability of node i being the responsible party in the path, X i and Y i denote the historical behavior characteristics and contextual environment characteristics of node i, respectively, P(R i ) is the prior responsibility probability of node i. The prior responsibility probability is essentially the responsibility value R i The probability of the responsibility value R calculated above i =0.47, the prior probability of responsibility here is 47%. The prior probability of responsibility can also be understood as the probability that node i is the direct responsible node in each event that has occurred in the past, P(X i ∣R i ) and P(Y i ∣R i ) are the conditional probabilities of the historical behavior characteristics and contextual environment characteristics of node i given the responsible party, and the denominator P(X i ,Y i ) is the joint probability of the features and is used for normalization.

[0089] Specifically, conditional probability refers to the probability of event A occurring under the condition that factor B occurs, and the calculation formula is P(A∣B)=P(AB)P(B)P(A∣B)=P(B)P(AB), where P(AB) is the joint probability, that is, the probability of A and B occurring at the same time. If a transaction is associated with a known malicious address, then the conditional probability that the transaction is malicious may increase.

[0090] Joint probability refers to the probability of events occurring at the same time. For example, the probability of two events A and B occurring at the same time is recorded as P(A,B) or P(A∩B) or P(AB). If events A and B are independent of each other, then P(A,B)=P(A)P(B)P(A,B)=P(A)P(B); if a transaction has both a large transfer and is associated with a known malicious address, then when these two features appear at the same time, the joint probability that the transaction is malicious will be higher.

[0091] Suppose that in a blockchain, a user has a smart contract for detecting abnormal transactions. The user sets a priori probability of responsibility for a node to 5% based on historical data. When a new transaction occurs, the user finds that the transaction is associated with a known malicious address, and this association increases the conditional probability that the transaction is malicious to 20%. At the same time, if the transaction amount exceeds a certain threshold, the joint probability of this threshold and the malicious address may further increase to 40%. In this way, the user can adjust the assessment of the probability of responsibility propagation for this transaction based on these prior probability of responsibility, conditional probability, and joint probability, thereby more accurately identifying abnormal transactions.

[0092] The fourth step is anomaly detection and responsibility attribution.

[0093] Anomaly detection and responsibility attribution are the core links for the system to accurately determine data anomalies and responsibility nodes. By establishing an efficient anomaly detection mechanism and combining it with a dynamic responsibility allocation model, it gradually realizes the location of data anomalies and the hierarchical display of responsibility attribution.

[0094] Further, the probability that each data point in the multidimensional feature matrix belongs to the kth Gaussian distribution is calculated using the Gaussian mixture model, and the probability is compared with the first set threshold, where k is an integer greater than or equal to 0;

[0095] When the Gaussian distribution probability is less than the first set threshold, the data point is determined to be suspicious;

[0096] The isolation forest algorithm is used to calculate the anomaly score of the corresponding data point, and the final anomaly score is obtained by weighted average of the anomaly score and its Gaussian distribution probability;

[0097] The final anomaly score is compared with a second set threshold, and when the final anomaly score is less than the second set threshold, the data point is determined to be an abnormal data.

[0098] First, in order to improve the accuracy and robustness of anomaly detection, this embodiment adopts a multi-level detection mechanism. In the initial stage, the data content features and context features are detected based on the Gaussian mixture model. The Gaussian mixture model can adapt to complex data patterns by assuming that the data distribution is a linear combination of multiple Gaussian distributions. Assume that the multidimensional feature matrix generated in step 2 is X, with a dimension of n×m, where n is the number of data items and m is the number of features. Then, for each data point The probability of belonging to the kth Gaussian distribution is:

[0099] .

[0100] in, is the weight of the kth Gaussian distribution, is the mean , covariance Gaussian distribution density function, e, o and k are all integers greater than or equal to 0.

[0101] It should be noted here that and It is obtained by using the expectation maximization (EM) algorithm, which is a prior art and will not be described in detail here.

[0102] The calculated probability is compared with the set threshold. If it is less than the set threshold, the data point x is determined to be i Suspicious and belongs to the kth category of abnormal distribution. This category is used to determine the distribution of data anomalies, affecting the calculation of relevant parameters in the responsibility allocation formula, such as the probability of responsibility propagation related to data characteristics under this distribution.

[0103] On this basis, the isolation forest is introduced as a supplementary detection. Specifically, it includes: first randomly selecting a feature from a given data set, randomly selecting a split value between the maximum and minimum values ​​of the selected feature, dividing the data set into two subsets, so that the data points less than the split value are in the left subset, and the data points greater than the split value are in the right subset, and repeating the above steps for each subset until the stop condition is met (such as the number of data points in the subset is less than a certain threshold or the tree reaches the preset maximum depth), thereby constructing an isolated tree; then repeating the above process of constructing an isolated tree to obtain multiple isolated trees to form an isolation forest; then, for each data point, record its path length in each tree in the isolation forest (the number of edges from the root node to the leaf node where the data point is located), and calculate the average path length and anomaly score of the data point.

[0104] Then, the Gaussian distribution probability and the isolation forest anomaly score of each data point are weighted averaged, and the corresponding data point is judged whether it is abnormal based on the weighted average result. By combining these two methods, the coverage and accuracy of data anomaly detection are improved.

[0105] For the detected abnormal data, all the nodes it flows through in the traceability graph are determined, and these nodes are regarded as responsibility nodes, and then hierarchical analysis is performed through the dynamic responsibility allocation model.

[0106] The responsibility allocation model integrates the historical behavior characteristics, contextual environment characteristics and responsibility propagation probability of the node, calculates the contribution of each responsible node to the abnormal data, and provides support for subsequent hierarchical attribution reports. Assuming that node i is the responsible node, the model is as follows:

[0107] .

[0108] Among them, r i is the final responsibility score of node i, P(R i ∣X i ,Y i ) is the responsibility propagation probability of node i calculated in step 3, H i is the historical abnormal frequency of node i, α and β are adjustment weights used to balance the influence of propagation probability and historical behavior, and the value range of α and β is between 0 and 1. The specific value can be determined according to the characteristics of the data and its importance in the actual application scenario.

[0109] The combination of Gaussian mixture model and isolation forest improves the accuracy and robustness of anomaly detection, significantly reduces the possibility of false detection and missed detection, and provides more stable input data for the traceability system.

[0110] Furthermore, the node with the highest final responsibility score in the traceability graph is listed as the direct responsibility node, and the other nodes are listed as secondary responsibility nodes.

[0111] The responsibility attribution results are displayed in a hierarchical form, including direct responsibility nodes and secondary responsibility nodes. The direct responsibility node refers to the node with the highest final responsibility score, which is usually the main source of abnormal data; the secondary responsibility node refers to the node that indirectly participates in the transmission of abnormal data. Its final responsibility score is slightly lower than that of the direct responsibility node, but it still needs attention.

[0112] For example, in a data anomaly, the traceability graph contains nodes A, B, and C, and the responsibility propagation probability of node A is 0.6, and the historical anomaly frequency is 0.8; the responsibility propagation probability of node B is 0.4, and the historical anomaly frequency is 0.5; the responsibility propagation probability of node C is 0.2, and the historical anomaly frequency is 0.4. Assuming α=0.7 and β=0.3, the final responsibility score of node A is:

[0113] .

[0114] Following similar steps, the final responsibility scores of nodes B and C are 0.43 and 0.32 respectively. Therefore, node A is determined to be the directly responsible node, and nodes B and C are secondary responsible nodes.

[0115] The dynamic responsibility allocation model combines multi-dimensional features with the probability of responsibility propagation to generate a hierarchical attribution report, clearly displaying the direct responsibility and secondary responsibility nodes. It not only improves the accuracy of responsibility attribution, but also provides users with strong explainability support.

[0116] Step 5: Visualize the traceability and store the traceability results.

[0117] Visual traceability and traceability result storage are designed to present data flow and responsibility attribution results in an intuitive form, and provide efficient traceability result storage and analysis capabilities, such as Figure 4 As shown in the figure, it mainly provides users with a comprehensive traceability operation experience by combining dynamic visualization and storage management, ensuring the traceability and reusability of data analysis results.

[0118] Furthermore, the responsible node with the highest final responsibility score is listed as the direct responsible node, and the other nodes are listed as secondary responsible nodes;

[0119] The priori responsibility probability is normalized to obtain the color value of each responsible node, and the transfer weight of the edge is calculated according to the data content characteristics, transfer path characteristics, context environment characteristics and historical behavior characteristics between adjacent nodes;

[0120] A node-edge weight visualization model is constructed based on color values ​​and transfer weights to visualize the critical paths through which abnormal data flows and the final responsibility score of each responsible node.

[0121] The core of dynamic visualization is to present the traceability path, node responsibility distribution and historical change trend through a multi-dimensional graphical interface. To this end, a node-edge weight visualization model is constructed. In the graph structure of the traceability path, the prior responsibility probability P(R i ) and the transfer weight of the edge It is used as the key parameter for dynamic visualization. The graphical interface intuitively displays the node responsibility and the main data flow through color depth and edge width.

[0122] The display color depth of the node is calculated by the following formula:

[0123] .

[0124] Among them, C i is the color value of node i, Normalize represents the prior responsibility probability P(R i) to ensure that the node colors have a uniform visual range in the graph. With this formula, areas with high responsibility values ​​will appear darker, intuitively highlighting potential responsible nodes.

[0125] The width of the edge is adjusted according to the transfer weight, and the transfer weight calculation formula is:

[0126] .

[0127] Among them, X ij and Y ij They represent the data transmission characteristics and context environment characteristics between nodes i and j respectively. Function f is a weighted and mapping method that can convert feature input into edge weight values. Paths with larger edge widths are considered to be key paths through which data anomalies may flow.

[0128] Specifically, .

[0129] in, , Represents different aspects of data transfer characteristics, , Represents different factors of contextual environment characteristics. z is the index variable used to traverse the elements in a specific range associated with nodes i and j, which belongs to V(ij). It is used to sum up each item to determine the transfer weight of the edge. a, b, c, and d are corresponding weight coefficients, which are used to adjust the influence of different factors on the edge weight.

[0130] It should be noted here that data transmission characteristics include data content characteristics and transmission path characteristics.

[0131] Through dynamic visualization, users can quickly identify high-responsibility nodes, critical paths and corresponding responsibility distribution in the interface. At the same time, the interface also supports a time slider, which allows users to dynamically view the time change trend of the final responsibility distribution of the node.

[0132] The purpose of tracing result storage is to store analysis data in a structured manner for easy query and historical comparison.

[0133] Three core data tables are designed for result storage, namely, node responsibility score table, path weight table and time responsibility change table. Each table is associated with a unique identifier of the event to facilitate comparative analysis.

[0134] At the same time, in order to support efficient storage and fast retrieval, an optimized storage strategy based on a time series database is also adopted. The node's final responsibility score, path weight, and event change data are indexed by events and stored in chronological order, thereby supporting efficient time range queries and dynamic updates.

[0135] For example, a traceability event includes nodes A, B and C, and the traceability path is A→B→C. The final responsibility scores of the nodes are 0.8, 0.5 and 0.3, respectively, and the transfer weights are 0.7 and 0.4, respectively. The color values ​​of nodes A, B and C calculated by the normalization formula are 1.0, 0.625 and 0.375, respectively. That is, in the visualization interface, node A is displayed in dark color, node C is displayed in light color, the width of edge A→B is larger, and the width of edge B→C is smaller, which highlights that the critical path of the data anomaly is from A to B.

[0136] In the traceability result storage, the node responsibility score table records the final responsibility scores of nodes A, B and C and their calculation sources respectively; the path weight table records the transfer weight values ​​of A→B and B→C respectively; the time responsibility change table shows the changes in the final responsibility scores of nodes A, B and C at different time points, providing a basis for subsequent analysis.

[0137] This embodiment enhances the accuracy of determining responsible nodes in the traceability path by introducing a probability model of responsibility propagation and Bayesian reasoning, and optimizes multi-dimensional feature modeling by combining contextual information and historical behavior characteristics. At the same time, the final responsibility score of the node is calculated based on a dynamic responsibility allocation model, and hierarchical attribution is supported to generate intuitive visual results, which solves the problem of uncertain responsibility attribution in traditional solutions and improves the accuracy and practicality of the traceability method.

[0138] In another embodiment, if Figure 5 As shown, the present application also provides a data tracing system based on big data and multi-dimensional features of blockchain, including:

[0139] An acquisition module is used to acquire a target data set, which includes a number of target data records and the context state of each target data record and the historical behavior records of its related nodes. The target data record includes data content and a transmission path;

[0140] The first construction module is used to calculate the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and to construct a multi-dimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight;

[0141] The second construction module is used to input the multi-dimensional feature matrix into the graph convolutional network for feature fusion to obtain a global feature map, and to construct a traceability graph with related nodes as vertices and interaction weights as edges based on the global feature map;

[0142] A detection module, used to calculate the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and to determine the abnormal data in the target data set using a multi-level detection mechanism;

[0143] The calculation module is used to determine the responsible nodes in the traceability graph according to the transmission path of the abnormal data, and calculate the final responsibility score of each responsible node according to the responsibility propagation probability and historical behavior records;

[0144] The visualization module is used to attribute responsibility based on the final responsibility score and build a node-edge weight visualization model based on the responsibility attribution results for visual traceability.

[0145] This embodiment is used to implement the method provided by the above embodiment, and has the corresponding beneficial effects of the above method. For technical details not fully described in this embodiment, please refer to the methods provided by all the above embodiments of the present invention.

[0146] In another embodiment, the present application also provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to implement a data tracing method based on big data and multi-dimensional features of blockchain as described above when executed.

[0147] Exemplarily, a computer program may be divided into one or more modules / units, one or more modules / units are stored in a memory and executed by a processor, and the I / O interface transmission of data is completed by an input interface and an output interface to complete the present invention. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in a computer device.

[0148] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may include, but is not limited to, a memory and a processor. Those skilled in the art may understand that this embodiment is only an example of a computer device and does not constitute a limitation on the computer device. The computer device may include more or fewer components, or a combination of certain components, or different components. For example, the computer device may also include an input device, a network access device, a bus, etc.

[0149] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0150] The memory may be an internal storage unit of a computer device, such as a hard disk or memory of a computer device. The memory may also be an external storage device of a computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device. Furthermore, the memory may include both an internal storage unit of a computer device and an external storage device. The memory is used to store computer programs and other programs and data required by the computer device. The memory 601 may also be used to temporarily store data in an output device. The aforementioned storage media include various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory ROM, a random access memory RAM, a disk or an optical disk.

[0151] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A data tracing method based on big data and multi-dimensional features of blockchain, characterized in that: The following steps are involved: Acquire a target data set, wherein the set includes a plurality of target data records and a context state of each target data record and a historical behavior record of its related nodes, wherein the target data record includes data content and a transmission path; Calculate the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and construct a multidimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight; Input the multidimensional feature matrix into a graph convolutional network for feature fusion to obtain a global feature graph, and construct a traceability graph with relevant nodes as vertices and interaction weights as edges based on the global feature graph; Calculate the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and use a multi-level detection mechanism to determine abnormal data in the target data set; Determine the responsible node in the traceability graph according to the transmission path of the abnormal data, and calculate the final responsibility score of each responsible node according to the responsibility propagation probability and historical behavior records; Responsibility is attributed according to the final responsibility score, and a node-edge weight visualization model is constructed according to the responsibility attribution result for visual traceability.

2. According to claim 1, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The step of obtaining a target data set includes: Extract several raw data records from multi-source heterogeneous data streams, and clean and standardize the format of each raw data record to obtain the target data record; Obtaining the context state of each target data record and the historical behavior record of each node related thereto, wherein the historical behavior record includes the historical abnormality frequency and stability score; The target data set is constructed based on each target data record, its context environment state, and the historical behavior records of each related node.

3. According to claim 2, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The calculating of the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record includes: Determine the context environment evaluation value corresponding to each context environment state, and perform weighted average of the corresponding historical abnormal frequency, stability score and context environment evaluation value to obtain the responsibility value of each relevant node; The interaction frequency, transmission delay and reliability score between adjacent nodes on the transmission path are obtained, and the interaction weights between all adjacent nodes are calculated according to the interaction frequency, transmission delay and reliability score.

4. According to claim 1, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The multi-dimensional feature matrix is ​​constructed according to the data content, transmission path, context environment state, historical behavior record and interaction weight, including: Extracting features of the data content, transmission path, context environment state, and historical behavior records to obtain data content features, transmission path features, context environment features, and historical behavior features; A multidimensional feature matrix is ​​constructed according to the data content features, transmission path features, context environment features, historical behavior features and interaction weights.

5. According to claim 1, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The calculating the responsibility propagation probability of each node in the traceability graph according to the responsibility value includes: Obtain the conditional probability and joint probability corresponding to the historical behavior characteristics and contextual environment characteristics of each relevant node under a given responsible party; The priori responsibility probability of each relevant node is determined according to the responsibility value, and the responsibility propagation probability of each relevant node is calculated according to the conditional probability, the joint probability and the priori responsibility probability.

6. According to claim 1, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The multi-level detection mechanism is used to determine abnormal data, including: Calculate the probability that each data point in the multidimensional feature matrix belongs to the kth Gaussian distribution by using a Gaussian mixture model, and compare it with a first set threshold, where k is an integer greater than or equal to 0; When the Gaussian distribution probability is less than the first set threshold, the data point is determined to be suspicious; The anomaly score of the corresponding data point is calculated using the isolation forest algorithm, and the final anomaly score is obtained by weighted average of the anomaly score and its Gaussian distribution probability; The final anomaly score is compared with a second set threshold, and when the final anomaly score is less than the second set threshold, the data point is determined to be an abnormal data.

7. According to claim 2, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: Determining the responsible node in the traceability graph according to the transmission path of the abnormal data, and calculating the final responsibility score of each responsible node according to the responsibility propagation probability and historical behavior records, includes: All nodes through which the abnormal data flows in the traceability graph are regarded as responsible nodes, and the final responsibility score of each responsible node is calculated according to the responsibility propagation probability and historical abnormality frequency.

8. According to claim 5, a data tracing method based on big data and multi-dimensional features of blockchain is characterized in that: The attribution of responsibility is performed according to the final responsibility score, and a node-edge weight visualization model is constructed according to the responsibility attribution result for visual tracing, including: The responsible node with the highest final responsibility score is listed as the direct responsible node, and the other nodes are listed as secondary responsible nodes; The priori responsibility probability is normalized to obtain the color value of each responsible node, and the transfer weight of the edge is calculated according to the data content characteristics, transfer path characteristics, context environment characteristics and historical behavior characteristics between adjacent nodes; A node-edge weight visualization model is constructed according to the color values ​​and transfer weights to visualize the critical paths through which the abnormal data flows and the final responsibility score of each responsible node.

9. A data tracing system based on big data and multi-dimensional features of blockchain, characterized in that: include: An acquisition module, used to acquire a target data set, wherein the set includes a plurality of target data records and the context state of each target data record and the historical behavior records of its related nodes, wherein the target data record includes data content and a transmission path; A first construction module is used to calculate the responsibility value of each relevant node and the interaction weight between adjacent nodes based on the target data record, context environment state and historical behavior record, and to construct a multidimensional feature matrix according to the data content, transmission path, context environment state, historical behavior record and interaction weight; A second construction module is used to input the multidimensional feature matrix into a graph convolutional network for feature fusion to obtain a global feature map, and to construct a traceability graph with relevant nodes as vertices and interaction weights as edges according to the global feature map; A detection module, used to calculate the responsibility propagation probability of each node in the traceability graph according to the responsibility value, and to determine the abnormal data in the target data set using a multi-level detection mechanism; A calculation module, used to determine the responsibility node in the traceability graph according to the transmission path of the abnormal data, and calculate the final responsibility score of each responsibility node according to the responsibility propagation probability and historical behavior records; A visualization module is used to attribute responsibility according to the final responsibility score, and to build a node-edge weight visualization model according to the responsibility attribution result for visual tracing.

10. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables the computer to implement a data tracing method based on big data and multi-dimensional features of blockchain as described in any one of claims 1 to 8 when executed.

Images