Unstructured data storage optimization method and system based on information entropy and blockchain

Abstract

The application provides an unstructured data storage optimization method and system based on information entropy and a blockchain, relates to the technical field of data storage, and comprises the following steps: establishing a blockchain storage network to obtain data, identifying redundant data by dimension reduction, mapping and a hyperplane, removing the redundant data based on fuzzy similarity, dividing the data to be stored into blocks, calculating access frequency, determining a storage location by executing a consistent hash algorithm, establishing an access index and a verification sequence, and executing data integrity verification, so that the data access efficiency can be improved and the storage security can be enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data storage, in particular to a non-structured data storage optimization method and system based on information entropy and blockchain. BACKGROUND

[0002] Non-structured data includes images, videos, texts, audios and other forms, and traditional data storage systems face problems such as low storage efficiency, poor security, complex management and the like when processing a large amount of non-structured data, and the distributed, tamper-proof, traceable and other characteristics of blockchain technology provide new ideas and methods for the storage of non-structured data.

[0003] However, existing non-structured data storage technologies have problems such as difficulty in effectively identifying and processing redundant information in non-structured data, resulting in a large amount of repeated or similar data being repeatedly stored, which not only wastes storage resources but also increases the complexity of data management, lack of effective mechanisms for data distribution and access optimization, inability to intelligently allocate data according to data access frequency and characteristics, leading to hot data access congestion and resource utilization imbalance, reducing the performance and response speed of the entire storage system, and difficulty in ensuring the security and reliability of non-structured data in a distributed environment, especially lack of efficient data access verification and integrity proof methods, making the system face the risk of data tampering and unauthorized access. SUMMARY

[0004] The embodiments of the present application provide a non-structured data storage optimization method and system based on information entropy and blockchain, which can at least solve some of the problems in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a non-structured data storage optimization method based on information entropy and blockchain, comprising:

[0006] establishing a blockchain storage network and obtaining non-structured data, constructing a covariance matrix corresponding to the non-structured data and performing eigenvalue decomposition to obtain reduced dimension data;

[0007] mapping the reduced dimension data to a high-dimensional feature space through a preset kernel function, constructing an optimal classification hyperplane, calculating the distance vector of each reduced dimension data to the optimal classification hyperplane to obtain an initial redundant data set, generating a binary code sequence based on the data features in the initial redundant data set, performing selection, crossover and mutation operations on the binary code sequence to obtain an optimized feature sequence, marking redundant data according to the optimized feature sequence, calculating the fuzzy similarity of different redundant data, and removing data with fuzzy similarity greater than a preset fuzzy similarity threshold to obtain a data set to be stored;

[0008] The data set to be stored is divided into a data block set according to a preset size, the access frequency of each data block in the data block set is calculated, a consistent hash algorithm is executed in combination with the access frequency to allocate the data block set to obtain a data block storage location table, a data access index is established based on the data block storage location table, each data block is segmented and encoded to calculate a hash value to obtain a data block feature table, and a data access verification sequence is generated according to the data block feature table and the data access index;

[0009] According to the data access verification sequence, data integrity verification is performed, a proof containing data identification and signature is generated, and the verification result is recorded to a block chain.

[0010] In an optional implementation,

[0011] A block chain storage network is established, and unstructured data is obtained, a covariance matrix corresponding to the unstructured data is constructed, and eigenvalue decomposition is performed to obtain reduced dimension data, including:

[0012] Unstructured data is obtained, and a block chain storage network composed of multiple nodes is established;

[0013] Frequency features and time sequence features in the unstructured data are extracted, a feature vector is calculated according to the frequency features and the time sequence features, and a covariance matrix is constructed based on the feature vector;

[0014] Eigenvalue decomposition is performed on the covariance matrix to obtain initial eigenvalues, and the unstructured data is reconstructed based on the initial eigenvalues to obtain reduced dimension data, and the reduced dimension data is stored in the block chain storage network.

[0015] In an optional implementation,

[0016] The reduced dimension data is mapped to a high-dimensional feature space by a preset kernel function, an optimal classification hyperplane is constructed, a distance vector of each reduced dimension data to the optimal classification hyperplane is calculated to obtain an initial redundant data set, a binary coding sequence is generated based on the data features in the initial redundant data set, and selection, crossover and mutation operations are performed on the binary coding sequence to obtain an optimized feature sequence, including:

[0017] Reduced dimension data is obtained, the Euclidean distance between different data points in the reduced dimension data is calculated, a Gaussian kernel function is constructed based on the Euclidean distance and a pre-set kernel function bandwidth parameter, and the reduced dimension data is mapped to a high-dimensional feature space by using the Gaussian kernel function to obtain high-dimensional feature data;

[0018] constructing an optimization objective function based on the high-dimensional feature data, wherein the optimization objective function comprises a normal vector length minimization term and a constraint relaxation term weighted based on a pre-set penalty factor, and solving the optimization objective function to obtain a normal vector of an optimal classification hyperplane and a bias term;

[0019] calculating a distance value of each data point in the reduced dimension data to the optimal classification hyperplane based on the normal vector and the bias term, sorting the distance values from small to large, selecting data points located in the top 10% to construct an initial redundant data set, extracting a feature vector corresponding to each data point from the initial redundant data set, calculating the mean value of each dimension feature component of the feature vector, comparing the feature component with the corresponding dimension mean value, assigning a binary value 1 to the feature component greater than the mean value and a binary value 0 to the feature component less than the mean value, and generating a binary coding sequence;

[0020] calculating the fitness value of each sequence in the binary coding sequence, performing roulette selection operation based on the fitness value to construct a high-quality sequence, performing single-point crossover operation on the high-quality sequence to generate a crossover sequence, and obtaining an optimized feature sequence by performing random inversion operation on the bit position of the crossover sequence.

[0021] In an optional embodiment,

[0022] According to the optimized feature sequence, the redundancy data is labeled, the fuzzy similarity of different redundant data is calculated, the data with a fuzzy similarity greater than a pre-set fuzzy similarity threshold is removed, and a to-be-stored data set is obtained, comprising:

[0023] obtaining an optimized feature sequence and mapping each feature dimension in the optimized feature sequence to a fuzzy membership value, constructing a triangular membership function based on the fuzzy membership value, dynamically adjusting the parameters of the triangular membership function based on the data distribution characteristics corresponding to the unstructured data, and obtaining an adaptive fuzzy feature;

[0024] constructing a multi-dimensional fuzzy rule base based on the adaptive fuzzy feature, assigning an initial weight coefficient to each rule in the multi-dimensional fuzzy rule base, calculating the fuzzy similarity between different data according to the multi-dimensional fuzzy rule base, and labeling the data with a fuzzy similarity greater than a pre-set fuzzy similarity threshold as preliminary screening redundant data;

[0025] calculating the local density value corresponding to the preliminary screening redundant data based on the distance between data points and a cut-off distance parameter, and performing fine screening on the preliminary screening redundant data according to the local density value, and labeling the data with a local density value greater than a pre-set density threshold as candidate redundant data;

[0026] Calculate the integrity score of the candidate redundant data, eliminate the candidate redundant data whose integrity score is lower than the pre-set integrity threshold, and construct the to-be-stored data set based on the retained candidate redundant data.

[0027] In an optional implementation,

[0028] The to-be-stored data set is divided into a data block set according to a preset size, the access frequency of each data block in the data block set is calculated, a consistent hash algorithm is executed to allocate the data block set to obtain a data block storage location table, and a data access index is established based on the data block storage location table, including:

[0029] The to-be-stored data set is preliminarily divided according to a preset size, an optimal block factor including the total size of the data set and the data density variance is calculated, and the to-be-stored data set is adaptively divided according to the optimal block factor to obtain a data block set;

[0030] For each data block in the data block set, the access frequency of each data block in the data block set is calculated based on the historical access number in a preset time window and a pre-set time decay factor;

[0031] The storage node weight value is calculated based on the access frequency and the basic weight value corresponding to each storage node, the storage node weight value is used to virtually extend the storage node to obtain a virtual node distribution sequence, the data block hash value of each data block in the data block set is calculated, the consistent hash algorithm is executed to map the weighted sum of the data block hash value and the access frequency to the hash space corresponding to the virtual node distribution sequence to obtain a data block storage location table;

[0032] The main index table recording the mapping relationship between the data block and the physical node is established based on the data block storage location table, the access frequency threshold is set based on the access frequency and the secondary index table of hot data access is established, the physical distance and the logical distance are calculated, the routing weight is constructed, and the routing table is generated, and the main index table, the secondary index table and the routing table are combined to form a data access index.

[0033] In an optional implementation,

[0034] The data block is segmented and encoded to obtain a data block feature table, and the data access verification sequence is generated based on the data block feature table and the data access index, including:

[0035] The historical access record of the data block is obtained, and the access mode time sequence is generated based on the historical access record;

[0036] The content entropy, boundary sensitivity and data density of the content in the data block are calculated to construct a data block content feature vector, the optimal segmentation size is calculated based on the information entropy of the data block content feature vector and the variance of the access mode time sequence, and each data block is segmented according to the optimal segmentation size to obtain a segmentation sequence;

[0037] The historical error record of the data block is obtained to calculate the data block error rate, the information symbol number is calculated based on the access frequency of the data block in the segmentation sequence, the check symbol number is calculated according to the data block error rate and the information symbol number, the encoding block set is obtained by encoding processing the segmentation sequence, the preset minimum segmentation size of the data block is obtained, the tree structure depth is calculated according to the data block size and the tree structure is constructed, the node number of the encoding block in the encoding block set is allocated to generate the parent node pointer, and the hierarchical hash value of the encoding block set is calculated, the hierarchical hash value, the tree structure depth information and the parent node pointer are combined to form a data block feature table;

[0038] The access frequency of the data block in the data access index is obtained to calculate the cache threshold, the data node with the access frequency greater than the cache threshold is cached and pre-fetched to establish a verification path, the current time of the system is obtained as a time stamp, the read-write state, the integrity state and the access permission state of the data block are obtained to form node state information, the hierarchical hash value, the time stamp and the node state information are signed to obtain a verification mark, and the data access verification sequence is generated based on the verification mark, the verification path and the data block feature table.

[0039] In an optional implementation,

[0040] According to the data access verification sequence, data integrity verification is performed, a proof containing a data identifier and a signature is generated, and the verification result is recorded to a block chain, which includes:

[0041] The data access verification sequence is obtained, the hierarchical hash value is recursively verified based on the verification path in the data access verification sequence, and a verification result is generated;

[0042] According to the verification result, a data identifier is generated, and a signature processing is performed on the data identifier to obtain an integrity signature, and the data identifier and the integrity signature are combined to form a verification proof;

[0043] The verification proof is submitted to a block chain network and recorded.

[0044] In a second aspect of the embodiment of the application, an unstructured data storage optimization system based on information entropy and a block chain is provided, which includes:

[0045] A first unit is configured to establish a block chain storage network, obtain unstructured data, construct a covariance matrix corresponding to the unstructured data, perform eigenvalue decomposition, and obtain reduced dimension data;

[0046] The second unit is configured to map the reduced dimension data to a high dimension feature space by a preset kernel function, construct an optimal classification hyperplane, calculate a distance vector of each reduced dimension data to the optimal classification hyperplane to obtain an initial redundant data set, generate a binary coding sequence based on data features in the initial redundant data set, perform selection, crossover and mutation operations on the binary coding sequence to obtain an optimized feature sequence, mark redundant data according to the optimized feature sequence, calculate fuzzy similarity of different redundant data, and remove data with fuzzy similarity greater than a preset fuzzy similarity threshold to obtain a data set to be stored.

[0047] The third unit is configured to divide the data set to be stored into data block sets according to a preset size, calculate access frequencies of data blocks in the data block sets, perform a consistent hash algorithm on the data block sets according to the access frequencies to obtain a data block storage location table, establish a data access index based on the data block storage location table, segmentally encode each data block and calculate a hash value to obtain a data block feature table, and generate a data access verification sequence according to the data block feature table and the data access index.

[0048] The fourth unit is configured to perform data integrity verification according to the data access verification sequence, generate a proof containing data identification and a signature, and record a verification result to a block chain.

[0049] In a third aspect of the embodiments of the present application, an electronic device is provided, comprising:

[0050] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0051] In a fourth aspect of the embodiments of the present application, a computer readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0052] In the present application, by constructing a covariance matrix in a block chain network and performing eigenvalue decomposition, effective dimension reduction processing of unstructured data is realized, data redundancy and storage space occupation are reduced, data processing efficiency is improved, a fuzzy similarity calculation mechanism and a genetic algorithm optimized feature sequence are introduced, redundant data with high similarity are effectively identified and removed, the block chain storage structure is significantly optimized while ensuring data integrity, the problem of occupying storage resources by a large amount of repeated data is avoided, an efficient data access index system is constructed by combining data block access frequency and a consistent hash algorithm, and data integrity verification is realized by data block segmentation coding and hash calculation, thereby ensuring the safety and reliability of unstructured data in the block chain storage network and improving the query and access efficiency of data. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart of the method for optimizing unstructured data storage based on information entropy and a blockchain according to an embodiment of the present application is shown in

[0054] Figure 2 A data block verification and processing flowchart of the method for optimizing unstructured data storage based on information entropy and a blockchain according to an embodiment of the present application is shown in DETAILED DESCRIPTION

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in connection with the drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0056] The technical solutions of the present application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and some embodiments can not be described again for the same or similar concepts or processes.

[0057] Figure 1 A flowchart of the method for optimizing unstructured data storage based on information entropy and a blockchain according to an embodiment of the present application is shown in Figure 1 The method comprises:

[0058] A blockchain storage network is established, and unstructured data is obtained. A covariance matrix corresponding to the unstructured data is constructed, and eigenvalue decomposition is performed to obtain reduced dimension data.

[0059] The reduced dimension data is mapped to a high-dimensional feature space through a preset kernel function, an optimal classification hyperplane is constructed, a distance vector of each reduced dimension data to the optimal classification hyperplane is calculated to obtain an initial redundant data set, a binary code sequence is generated based on the data features in the initial redundant data set, selection, crossover, and mutation operations are performed on the binary code sequence to obtain an optimized feature sequence, the redundant data is marked according to the optimized feature sequence, the fuzzy similarity of different redundant data is calculated, the data with a fuzzy similarity greater than a preset fuzzy similarity threshold is removed, and a to-be-stored data set is obtained.

[0060] The data set to be stored is divided into a data block set according to a preset size, the access frequency of each data block in the data block set is calculated, a consistent hash algorithm is executed in combination with the access frequency to allocate the data block set to obtain a data block storage location table, a data access index is established based on the data block storage location table, each data block is segmented and encoded to calculate a hash value to obtain a data block feature table, and a data access verification sequence is generated according to the data block feature table and the data access index;

[0061] Data integrity verification is performed according to the data access verification sequence, a proof containing data identification and a signature is generated, and the verification result is recorded to a block chain.

[0062] In an optional implementation,

[0063] A block chain storage network is established and unstructured data is obtained, a covariance matrix corresponding to the unstructured data is constructed and eigenvalue decomposition is performed, and reduced dimension data is obtained, including:

[0064] Unstructured data is obtained, and a block chain storage network composed of multiple nodes is established;

[0065] Frequency features and time sequence features in the unstructured data are extracted, a feature vector is calculated according to the frequency features and the time sequence features, and a covariance matrix is constructed based on the feature vector;

[0066] Eigenvalue decomposition is performed on the covariance matrix to obtain initial eigenvalues, and the unstructured data is reconstructed based on the initial eigenvalues to obtain reduced dimension data, and the reduced dimension data is stored in the block chain storage network.

[0067] Unstructured data is obtained from various data sources, such as text documents, images, audio, and video. For text data, web crawler technology can be used to crawl from specified websites; for audio data, a microphone device can be used for collection; for video data, a camera device can be used for recording. The obtained unstructured data is stored in a temporary storage space to prepare for subsequent processing.

[0068] A block chain storage network is established, which is composed of multiple nodes, each node has data storage, verification and synchronization functions. The block chain network adopts a distributed architecture, and the nodes communicate with each other through a P2P protocol. Each node maintains a complete block chain ledger to ensure data consistency. The block structure includes a block header and a block body, the block header includes a previous block hash value, a timestamp, a difficulty target and a random number; the block body includes transaction data, i.e. the reduced dimension unstructured data in the present technical solution. Consensus algorithms are used between nodes to ensure data consistency, and proof of work, proof of stake or practical Byzantine fault tolerance algorithms can be selected according to actual needs.

[0069] Frequency feature extraction of unstructured data involves analyzing the frequency of occurrence of each component element in the data. For text data, word frequency or character frequency can be calculated; for image data, pixel value distribution can be analyzed; for audio data, the energy distribution of each frequency band can be calculated. Taking text data as an example, use word segmentation technology to divide the text into a sequence of words, and calculate the frequency of each word in the document. The dimension of the frequency feature vector is comparable to the size of the vocabulary, and each dimension represents the frequency of the corresponding word. In practical applications, low-frequency words need to be filtered, and only high-frequency word features are retained, thereby reducing the vector dimension.

[0070] Temporal feature extraction mainly focuses on the law of data elements changing over time. For time series data, features such as trend, periodicity, autocorrelation coefficient, etc. can be extracted; for text data, the transition relationship between sentences or paragraphs can be analyzed; for video data, inter-frame change features can be extracted. Taking time series data as an example, statistical quantities such as mean, variance, kurtosis, skewness, etc. can be calculated within different time windows to form a time series feature vector.

[0071] Based on the extracted frequency features and time series features, the feature vector is calculated, and the two types of features are fused. The frequency feature vector and the time series feature vector can be directly concatenated to form a comprehensive feature vector using the feature concatenation method; or the two types of features can be fused by assigning different weights to them. To improve the feature expression ability, feature transformation techniques such as logarithmic transformation and normalization processing can be introduced. After the feature vector is calculated, each unstructured data sample corresponds to a fixed-dimensional feature vector.

[0072] The feature vectors of all samples form a feature matrix, and the covariance relationship between the features is calculated. Each element of the covariance matrix represents the correlation between two features. During the calculation, the feature vectors are first centralized, i.e. the mean of each feature is subtracted, so that the feature distribution center is at the origin. After centralization, the covariance between the features is calculated, and the covariance matrix is filled. The covariance matrix is a symmetric matrix, the diagonal elements are the variances of the features, and the non-diagonal elements are the covariances between the feature pairs.

[0073] By calculating the eigenvalues and eigenvectors of the covariance matrix, the main variation direction of the data is found. The size of the eigenvalue represents the degree of data change in the corresponding eigenvector direction, and the larger the eigenvalue, the more significant the data change in that direction. In actual operation, numerical calculation methods such as power iteration method and QR decomposition method can be used to solve the eigenvalues and eigenvectors.

[0074] According to the size of the eigenvalue, sort and select the top-ranked eigenvectors to form a dimension reduction projection matrix. The number of columns of the projection matrix determines the dimension of the reduced data, and the number of dimensions to be retained can be determined according to the cumulative variance contribution rate. For example, select the top K eigenvectors, and the sum of their corresponding eigenvalues accounts for more than 95% of the total eigenvalue sum.

[0075] The original feature vector is transformed by using a projection matrix to obtain a reduced dimension data representation. The reduced dimension data not only retains the main information of the original data, but also greatly reduces the storage space requirement. For example, for 100MB of text data, only 20MB of storage space may be required after applying the method, greatly improving the storage efficiency.

[0076] The reduced dimension data is stored in the blockchain network, and the reduced dimension data is packaged into a transaction and added to a block after being verified by a network node. To improve storage efficiency, the reduced dimension data can be further compressed or fragmented. Each data block is attached with meta-information such as timestamp and data hash, facilitating subsequent retrieval and verification. After data storage is completed, the network nodes automatically synchronize the ledger to ensure consistency of data across the network.

[0077] In this embodiment, after obtaining the unstructured data, the frequency features and time sequence features are extracted and the feature vectors are calculated to effectively depict the dynamic change rule and internal correlation of the unstructured data, avoiding the high-dimensional redundancy problem caused by directly processing the original data. By constructing a covariance matrix and performing eigenvalue decomposition, the most representative important feature dimension in the data can be identified. The reduced dimension data is stored in a blockchain storage network composed of multiple nodes, which can realize decentralized storage of data in a distributed environment, ensure data tamper resistance and traceability, improve data security and credibility, and based on the multi-node collaborative storage mechanism of the blockchain, the stability and disaster recovery capability of data storage and access are also improved, so that the unstructured data has higher utilization value under the double protection of efficient compression and safe storage.

[0078] In an alternative embodiment,

[0079] The reduced dimension data is mapped to a high-dimensional feature space by a preset kernel function, an optimal classification hyperplane is constructed, a distance vector of each reduced dimension data to the optimal classification hyperplane is calculated to obtain an initial redundant data set, a binary code sequence is generated based on the data features in the initial redundant data set, and selection, crossover and mutation operations are performed on the binary code sequence to obtain an optimized feature sequence.

[0080] The reduced dimension data is obtained, the Euclidean distance between different data points in the reduced dimension data is calculated, a Gaussian kernel function is constructed based on the Euclidean distance and a pre-set kernel function bandwidth parameter, and the reduced dimension data is mapped to a high-dimensional feature space by using the Gaussian kernel function to obtain high-dimensional feature data.

[0081] An optimization objective function is constructed based on the high-dimensional feature data, wherein the optimization objective function includes a normal vector length minimization term and a constraint relaxation term weighted based on a pre-set penalty factor, and a normal vector and a bias term of an optimal classification hyperplane are obtained by solving the optimization objective function.

[0082] Based on the normal vector and the bias term, the distance value of each data point in the reduced dimension data to the optimal classification hyperplane is calculated, the distance value is sorted from small to large, the data points located in the front 10% are selected to construct an initial redundant data set, the feature vector corresponding to each data point is extracted from the initial redundant data set, the mean value of each dimension feature component of the feature vector is calculated, the feature component is compared with the corresponding dimension mean value, the feature component greater than the mean value is assigned a binary value 1, and the feature component less than the mean value is assigned a binary value 0, and a binary coding sequence is generated;

[0083] The fitness value of each sequence in the binary coding sequence is calculated, and roulette selection operation is performed based on the fitness value to construct a high-quality sequence, single-point crossover operation is performed on the high-quality sequence to generate a crossover sequence, and random inversion operation is performed on the bit position of the crossover sequence to obtain an optimized feature sequence.

[0084] Obtain reduced dimension data, which can be a low-dimensional representation extracted from original unstructured data by principal component analysis, linear discriminant analysis or autoencoder, etc. In practical applications, the reduced dimension data can be a numerical vector after feature extraction and dimension reduction of unstructured data such as text, image or audio. For example, a 1000-word text can be converted into a 50-dimensional vector after word frequency statistics and dimension reduction, and each element in the vector represents the projection value of the original text in a certain feature direction.

[0085] Calculate the Euclidean distance between different data points in the reduced dimension data. The Euclidean distance calculation method is to square the difference value of the corresponding dimensions of two data points, sum them up, and then take the square root. Taking two 5-dimensional vectors as an example, if vector A is [0.2, 0.5, 0.8, 0.3, 0.6] and vector B is [0.4, 0.3, 0.7, 0.5, 0.2], the Euclidean distance between them is calculated as follows: first, square the sum of the differences in corresponding positions: (0.2-0.4) 2 +(0.5-0.3) 2 +(0.8-0.7) 2 +(0.3-0.5) 2 +(0.6-0.2)2=0.04+0.04+0.01+0.04+0.16=0.29, and then take the square root to get the Euclidean distance of about 0.54.

[0086] The Gaussian kernel function is constructed based on the Euclidean distance and a pre-set kernel bandwidth parameter, realizing the nonlinear transformation of data. The essence of the Gaussian kernel function is to map the Euclidean distance to a similarity value between 0 and 1 through an exponential function. The kernel bandwidth parameter controls the shape of the Gaussian function. A smaller bandwidth parameter makes the function curve steeper and more sensitive to distance changes. In practical applications, the bandwidth parameter can be determined by cross-validation method. For example, candidate values can be set in the range of 0.1 to 10, and the value that optimizes the classification performance is selected. Assuming that the selected bandwidth parameter is 2, the Gaussian kernel function value between two data points with a distance of 0.54 is e^(-0.54^2 / 2*2^2), which is approximately 0.93, indicating that the two points have a high similarity in high-dimensional space.

[0087] The reduced dimension data is mapped to a high-dimensional feature space using the Gaussian kernel function. The high-dimensional features do not need to be explicitly calculated, but are implicitly realized through the kernel trick. The data points in the high-dimensional feature space have better linear separability, which is beneficial for subsequent classification tasks. After data mapping, data points that are difficult to linearly separate in low-dimensional space may become linearly separable in high-dimensional space.

[0088] An optimization objective function is constructed based on the high-dimensional feature data. The function includes a normal vector length minimization term and a constraint relaxation term weighted based on a pre-set penalty factor. The purpose of the normal vector length minimization term is to maximize the classification boundary, and the constraint relaxation term allows part of the samples to violate the classification constraint, improving the model generalization ability. The penalty factor is a trade-off parameter that controls the tolerance of the model to misclassified samples. A larger penalty factor will make the model focus more on reducing misclassification, and a smaller penalty factor will make the model focus more on maximizing the boundary. In practical applications, the penalty factor can be set to a value between 0.1 and 10, and the optimal value is determined by grid search.

[0089] The normal vector and bias term of the optimal classification hyperplane are obtained by solving the optimization objective function. The sequential minimal optimization algorithm is used to optimize the Lagrange multipliers through iteration to obtain the normal vector and bias term. The normal vector determines the direction of the hyperplane, and the bias term determines the position of the hyperplane. Taking two-dimensional data as an example, if the normal vector is [0.8, 0.6] and the bias term is -0.5, the classification hyperplane equation is 0.8x + 0.6y - 0.5 = 0, where x and y are the two feature components of the data points.

[0090] The distance value of each data point in the reduced dimension data to the optimal classification hyperplane is calculated based on the normal vector and the bias term. The distance calculation method is the dot product of the feature vector of the data point and the normal vector plus the bias term, and then divided by the length of the normal vector. The positive and negative of the distance value indicates which side of the data point is located on the hyperplane, and the absolute value indicates the vertical distance of the data point to the hyperplane. The data point with a distance value close to zero is located near the decision boundary, and often contains more information and has a greater impact on the classification result.

[0091] The distance values are sorted from small to large, and the data points located in the top 10% are selected to construct the initial redundant data set. The data point closest to the decision boundary is considered as a candidate point with higher information redundancy. In the case of 10,000 data points, the 1,000 points with the smallest distance will form the initial redundant data set.

[0092] The feature vector corresponding to each data point in the initial redundant data set is extracted, and the mean value of each dimension feature component of the feature vector is calculated. For a 50-dimensional feature vector, the average value of 1,000 data points in each dimension is calculated to obtain a 50-dimensional mean vector. Compare the feature component with the corresponding dimension mean value, and assign binary value 1 to the feature component greater than the mean value, and assign binary value 0 to the feature component less than the mean value to generate a binary encoding sequence.

[0093] The fitness value of each sequence in the binary encoding sequence is calculated, which reflects the redundancy degree of the corresponding data point. Fitness calculation can be based on information entropy or similarity with other data points. The lower the information entropy or the higher the similarity, the higher the data redundancy, and the greater the fitness value. Based on the fitness value, roulette selection operation is performed to construct high-quality sequences. Roulette selection allocates selection probability in proportion to the fitness value, and the probability of being selected is greater for sequences with higher fitness values. Among the 1,000 sequences, 500 high-quality sequences may be selected for the next stage of processing.

[0094] Single-point crossover operation is performed on the high-quality sequences to generate crossover sequences. Single-point crossover randomly selects a crossover point between two parent sequences and exchanges all bits after the point to generate two new child sequences. For example, two 50-bit parent sequences are crossed at the 20th bit to generate two new sequences, each containing the first 20 bits of one parent sequence and the last 30 bits of the other parent sequence. Random inversion operation is performed on the bit points of the crossover sequence to obtain the optimized feature sequence. Random inversion is a bit inversion operation with a certain probability (such as 0.01) on each bit in the sequence, 0 to 1, 1 to 0, which increases the diversity of the population and avoids falling into local optimal solutions.

[0095] In this embodiment, the data is mapped to a high-dimensional feature space by calculating the Euclidean distance between different data points and combining the kernel function bandwidth parameter to construct a Gaussian kernel function, which effectively improves the separability and discriminability of the original reduced dimension data. The distance of each data point to the classification boundary is calculated using the hyperplane, and the top 10% of data points near the boundary are selected as the initial redundant data set, which helps to identify potential redundant or boundary ambiguous data samples in the feature distribution, improves the relevance of subsequent optimization processing, and preserves the better feature patterns by calculating the fitness value of the coding sequence and using the roulette wheel selection mechanism. Combined with single-point crossover and random inversion operation to further explore the feature space, the generated optimized feature sequence can highlight the optimal feature combination while maintaining diversity, thereby eliminating redundancy and improving feature representation ability.

[0096] In an alternative embodiment,

[0097] According to the optimized feature sequence, the redundant data is labeled, the fuzzy similarity of different redundant data is calculated, the data with fuzzy similarity greater than a preset fuzzy similarity threshold is removed, and a to-be-stored data set is obtained, including:

[0098] An optimized feature sequence is obtained, and each feature dimension in the optimized feature sequence is mapped to a fuzzy membership value. A triangular membership function is constructed based on the fuzzy membership value, and the parameters of the triangular membership function are dynamically adjusted based on the data distribution characteristics corresponding to the unstructured data to obtain an adaptive fuzzy feature.

[0099] A multi-dimensional fuzzy rule base is constructed based on the adaptive fuzzy feature, and an initial weight coefficient is assigned to each rule in the multi-dimensional fuzzy rule base. The fuzzy similarity between different data is calculated according to the multi-dimensional fuzzy rule base, and the data pair with fuzzy similarity greater than a preset fuzzy similarity threshold is labeled as preliminary screening redundant data.

[0100] The local density value corresponding to the preliminary screening redundant data is calculated based on the distance between data points and the cutoff distance parameter, and the preliminary screening redundant data is refined according to the local density value. The data with a local density value greater than a preset density threshold is labeled as candidate redundant data.

[0101] The integrity score of the candidate redundant data is calculated, and the candidate redundant data with an integrity score lower than a pre-set integrity threshold is removed. A to-be-stored data set is constructed based on the retained candidate redundant data.

[0102] An optimized feature sequence is obtained, which is usually generated by a genetic algorithm or other optimization algorithm, and contains key feature information of unstructured data. The feature sequence can be in binary form, such as "10101100", or in real number form, such as [0.3, 0.7, 0.2, 0.9]. When mapping each feature dimension in the optimized feature sequence to a fuzzy membership value, the distribution range and semantic meaning of the feature value need to be considered. For binary features, 0 can be directly mapped to 0.0 and 1 can be directly mapped to 1.0; for continuous features, they can be mapped to the [0, 1] interval through normalization processing. For example, for a feature value of 15 in the range [-10, 30], the membership value can be calculated by linear normalization as (15-(-10)) / (30-(-10)) = 0.625.

[0103] A triangular membership function is constructed based on the fuzzy membership value, which is defined by three parameters: left endpoint, vertex, and right endpoint. Assuming that the membership value of a certain feature dimension is 0.6, the left endpoint of the triangular membership function can be initially set to 0.4, the vertex to 0.6, and the right endpoint to 0.8. This means that when the feature value is exactly equal to 0.6, the membership degree is 1; when the feature value deviates from 0.6 but is within the range [0.4, 0.8], the membership degree decreases linearly with the distance; when the feature value is less than 0.4 or greater than 0.8, the membership degree is 0.

[0104] The parameters of the triangular membership function are dynamically adjusted based on the data distribution characteristics corresponding to the unstructured data, so that the fuzzy reasoning is more consistent with the actual data distribution. The adjustment method can be based on statistical characteristics of the data, such as mean, variance, skewness, etc. For uniformly distributed features, the three parameters of the triangular membership function can be set to be uniformly spaced; for skewed distribution features, the triangular membership function can be shifted towards the data-intensive area. For example, if the data points of a certain feature are concentrated in the range [0.5, 0.7], the left endpoint, vertex, and right endpoint of the triangular membership function can be adjusted to 0.45, 0.6, and 0.75 respectively, so that the function better covers the high-density area.

[0105] A multi-dimensional fuzzy rule base is constructed based on adaptive fuzzy features. The multi-dimensional fuzzy rules are in the form of "if... then...", describing the data redundancy degree under different feature combinations. For data with n feature dimensions, theoretically, 3^n rules can be constructed, but in practical applications, only important rules are usually retained to reduce computational complexity. For example, for a two-dimensional feature space, 9 rules can be constructed: if feature 1 is low and feature 2 is low, then the redundancy degree is high; if feature 1 is low and feature 2 is medium, then the redundancy degree is medium, and so on. The specific form of the rules depends on the semantic meaning and distribution characteristics of the data.

[0106] An initial weight coefficient is assigned to each rule in the multi-dimensional fuzzy rule base to reflect the importance of the rule. The initial weight can be set based on domain knowledge or derived from training data. For example, for text data, rules describing keyword frequency can have a higher weight; for image data, rules describing texture features can be more important. The initial weight can be set in the interval [0, 1], such as assigning weights [0.9, 0.7, 0.5, 0.7, 0.5, 0.3, 0.5, 0.3, 0.1] to 9 rules, indicating the importance of different rule combinations.

[0107] To calculate the fuzzy similarity between different data based on the multi-dimensional fuzzy rule base, the trigger strength and weight of each rule need to be considered. The specific calculation process includes calculating the trigger strength of each rule for each pair of data points; multiplying the trigger strength by the rule weight to obtain the weighted trigger strength; summing the weighted trigger strengths of all rules and dividing by the total weight to obtain the normalized fuzzy similarity. The fuzzy similarity value ranges from 0 to 1, with a larger value indicating that the two data points are more similar.

[0108] Data pairs with a fuzzy similarity greater than a pre-set fuzzy similarity threshold are marked as preliminary screening redundant data. The setting of the fuzzy similarity threshold depends on the redundancy tolerance of the application scenario, usually between 0.7 and 0.9. For example, if the threshold is set to 0.85, data pairs with a fuzzy similarity greater than 0.85 are considered potential redundancies and added to the preliminary screening redundant data set. In practical applications, the preliminary screening process may generate a large number of candidate redundant pairs, and to improve efficiency, hash or index technology can be used to speed up similarity calculation.

[0109] Based on the distance between data points and the cut-off distance parameter, the local density value corresponding to the preliminary screening redundant data is calculated to further screen redundant data. The local density reflects the data distribution around the point, and the calculation method is to count the number or weight sum of points with a distance less than the cut-off distance. The cut-off distance is a pre-set parameter that determines the range of local density calculation and can be determined through data analysis or cross-validation. For example, if the average neighboring distance of a data set is 0.1, the cut-off distance can be set to 0.15 to ensure that the local density calculation of each point includes enough neighboring points.

[0110] According to the local density value, the preliminary screening redundant data is refined, and data with a local density value greater than a pre-set density threshold is marked as candidate redundant data. The density threshold is set based on the overall density distribution of the data set, which can be the upper quartile or a specific percentile of the local density distribution of the data set. For example, if the local density values of the data set are distributed in the range [1, 10] with a median of 4, the density threshold can be set to 6, and data points with a local density greater than 6 are marked as candidate redundant points. The purpose of density screening is to retain points located in high-density areas of the data set, which usually have high information redundancy and can be considered as representatives of the area.

[0111] The integrity score of the candidate redundant data is calculated to evaluate whether it is suitable for elimination from the storage set. The integrity score reflects the completeness of the information contained in the data point, which can be calculated based on information entropy, feature coverage or data quality indicators. The calculation method of the integrity score varies with the data type: for text data, vocabulary coverage and semantic integrity can be considered; for image data, clarity and content richness can be considered; for sensor data, sampling integrity and signal-to-noise ratio can be considered. The integrity score is usually normalized to the interval [0, 1], and the higher the value, the better the data integrity.

[0112] The candidate redundant data with an integrity score lower than a pre-set integrity threshold is eliminated, ensuring that the data marked as redundant does not result in information loss. The setting of the integrity threshold needs to balance storage efficiency and information preservation, usually between 0.3 and 0.7. For example, if the threshold is set to 0.5, candidate redundant data with an integrity score lower than 0.5 is retained (not eliminated), and candidate redundant data with an integrity score higher than 0.5 is confirmed as true redundant data, which can be eliminated from the storage set. The retained candidate redundant data is used to construct the to-be-stored data set, i.e. the data set after the confirmed redundant data is eliminated, which will be finally stored in the blockchain network.

[0113] In this embodiment, by mapping each feature dimension in the optimized feature sequence to a fuzzy membership value and dynamically adjusting the parameters of the triangular membership function based on the data distribution characteristics, the unstructured data features are adaptively and flexibly characterized, so that the fuzzy feature differences of different data can be more accurately reflected, and data pairs with high similarity are identified through fuzzy similarity calculation, so that most redundant data can be effectively screened out in the early stage, thereby reducing the complexity of subsequent processing. The local density value is calculated using the distance between data points and the truncation distance parameter, and the preliminary screening redundant data is refined based on the density threshold, so that data points with denser local distribution and stronger representativeness can be retained, improving the accuracy of redundant data determination. While maintaining data integrity, the amount of redundant data is effectively reduced, realizing high-quality construction of the data set, thereby improving the efficiency of data storage and management, and providing a more reliable data basis for subsequent data analysis and modeling.

[0114] In an optional implementation,

[0115] The to-be-stored data set is divided into a data block set according to a pre-set size, the access frequency of each data block in the data block set is calculated, a consistent hash algorithm is executed to allocate the data block set to obtain a data block storage location table, and a data access index is established based on the data block storage location table, including:

[0116] The data set to be stored is initially blocked according to a preset size, an optimal blocking factor containing the total size of the data set and the variance of the data density is calculated, and the data set to be stored is adaptively blocked according to the optimal blocking factor to obtain a data block set;

[0117] For each data block in the data block set, the access frequency of each data block in the data block set is calculated based on the historical access times in a preset time window and a preset time decay factor;

[0118] The storage node weight value is calculated based on the access frequency and the base weight value corresponding to each storage node, the storage nodes are virtually extended based on the storage node weight value, a virtual node distribution sequence is obtained, the data block hash value is calculated for each data block in the data block set, the consistent hashing algorithm is executed to map the weighted sum of the data block hash value and the access frequency to the hash space corresponding to the virtual node distribution sequence, and a data block storage location table is obtained;

[0119] The main index table recording the mapping relationship between the data block and the physical node is established based on the data block storage location table, the access frequency threshold is set based on the access frequency, and the secondary index table of hot data access is established, the physical distance and the logical distance are calculated, the routing weight is constructed, and the routing table is generated, and the main index table, the secondary index table and the routing table are combined to form a data access index.

[0120] The data set to be stored is initially blocked according to a preset size, the preset size is usually a fixed value, such as 4MB or 8MB, and during the initial blocking process, the continuous data is cut according to the preset size. For text data, cutting can be performed at natural paragraphs; for image data, cutting can be performed according to image boundaries; and for audio data, cutting can be performed at silent segments. After initial blocking, a data block set with similar sizes is formed. For example, a 100MB data set is initially blocked according to a preset size of 4MB, and about 25 initial data blocks are formed.

[0121] The optimal blocking factor containing the total size of the data set and the variance of the data density is calculated, and the optimal blocking factor considers the total size of the data set, the node storage capacity and the data density distribution. The data density can be measured by information entropy, and the data block containing more effective information has higher density. The calculation method of the optimal blocking factor is to multiply the total size of the data set by the standardization coefficient, and then divide by the sum of the data density variance and the adjustment parameter. The standardization coefficient is related to the size of the blockchain network, and the adjustment parameter is used to prevent the denominator from approaching zero. For example, for a data set with a total size of 100MB and a data density variance of 0.05, the optimal blocking factor calculation result is about 166.7 under the condition that the standardization coefficient is 0.1 and the adjustment parameter is 0.01.

[0122] According to the optimal block factor, the data set to be stored is adaptively blocked to obtain a data block set. In the adaptive blocking process, the result of the preliminary blocking is adjusted so that the area with high information density is blocked more finely and the area with low information density is blocked more coarsely. In specific implementation, the information entropy of each initial data block is calculated, and the information entropy is multiplied by the optimal block factor to obtain the number of sub-blocks that the data block should be divided into. For example, if the information entropy of a certain initial data block is 0.8 and the optimal block factor is 166.7, the data block should be divided into about 133 sub-blocks. Considering the actual storage efficiency, a minimum sub-block size limit, such as 10 KB, can be set. After adaptive blocking is completed, the size of each data block in the data block set formed is inversely proportional to its information density, and the optimal allocation of storage resources is achieved.

[0123] For each data block in the data block set, the access frequency of each data block is calculated based on the historical access times in a preset time window and a pre-set time decay factor. The preset time window can be one day, one week or one month, and the time decay factor is used to reduce the weight of earlier access records. The calculation method of the access frequency is to multiply the access times at each time point in the time window by the corresponding time decay value and then sum. The time decay value decreases as the interval between the time point and the current time increases, and an exponential decay function can be used. For example, if the access times of a certain data block in the past 7 days are [10, 5, 8, 12, 3, 0, 2], and the time decay factor is 0.8, then when calculating the access frequency, the access times of each day are multiplied by the corresponding decay value (from the nearest day to the farthest day, 1, 0.8, 0.64, 0.512, 0.41, 0.328, 0.262, respectively), to obtain the weighted access times [10, 4, 5.12, 6.14, 1.23, 0, 0.52], and the sum is about 27.01.

[0124] The storage node weight value is calculated based on the access frequency and the basic weight value corresponding to each storage node. The basic weight value is pre-set by the system administrator according to the actual situation of the node. The calculation method of the storage node weight value is to multiply the node basic weight value by the weighted average value of the access frequencies of all data blocks stored by the node. For example, if the basic weight value of a certain storage node is 1.2, and the node stores 3 data blocks with access frequencies of 27.01, 15.6 and 8.4, respectively, then the weight value of the node is 1.2*(27.01+15.6+8.4) / 3=20.4.

[0125] The storage nodes are virtually extended based on the storage node weight values to obtain a virtual node distribution sequence. The virtual node extension creates multiple virtual nodes for each physical node, making the data distribution more uniform. The number of virtual nodes is proportional to the node weight value and can be calculated by multiplying the node weight value by a system preset virtual node base. For example, if the system virtual node base is 100 and the node weight value is 20.4, the node will be expanded to about 2040 virtual nodes. The virtual nodes are uniformly distributed on the hash ring, and each virtual node corresponds to a hash value.

[0126] The data block hash value is calculated for each data block in the data block set, and a hash algorithm such as SHA-256 or MD5 can be used. The consistent hashing algorithm maps the weighted sum of the data block hash value and the access frequency to the hash space corresponding to the virtual node distribution sequence to obtain a data block storage location table. The calculation method of the weighted sum is to multiply the data block hash value by the weight coefficient a and multiply the access frequency by the weight coefficient (1-a), and then add them. The weight coefficient a is usually set between 0.3 and 0.7 to balance the randomness of the hash value and the preference of the access frequency. The mapping process is to find the virtual node closest to the weighted sum clockwise on the hash ring, and the physical node corresponding to the virtual node is the storage location of the data block. The data block storage location table contains the mapping relationship between the data block ID and the storage node ID, which is used for data storage and retrieval.

[0127] A primary index table recording the mapping relationship between the data block and the physical node is established based on the data block storage location table. The primary index table is a key-value pair structure, with the key being the data block ID and the value being the physical node ID storing the data block. The primary index table supports fast query with a query complexity of O(1) and is used to process all data block storage location query requests. For example, the primary index table may contain the following mappings: {data block 001: node A, data block 002: node B, data block 003: node A}, indicating that data blocks 001 and 003 are stored in node A and data block 002 is stored in node B.

[0128] An access frequency threshold is set based on the access frequency, and a secondary index table for hot data access is established. The access frequency threshold can be set as the average value of the access frequency of all data blocks multiplied by a coefficient β, which is usually greater than 1, for example, 1.5 or 2. Data blocks with access frequency exceeding the threshold are marked as hot data and included in the secondary index table. The secondary index table contains the mapping of hot data block ID and storage node ID, as well as additional access frequency information. The secondary index table is used to optimize the access performance of hot data and can achieve cache warming and load balancing.

[0129] The physical distance and the logical distance are calculated, the routing weight is constructed, and the routing table is generated. The physical distance refers to the network delay or geographical distance between nodes, and the logical distance refers to the distance of nodes on the hash ring. The routing weight is calculated by weighted average of the normalized value of the physical distance and the normalized value of the logical distance, and the weight can be adjusted according to system requirements. The routing table contains the optimal path information from the source node to the target node, which is used for routing decision of data transmission. The main index table, the secondary index table and the routing table are combined to form a data access index, which provides comprehensive support for data storage and retrieval.

[0130] In the embodiment, the data set to be stored is preliminarily blocked according to the preset size, and the optimal blocking factor is calculated based on the total size of the data set and the data density variance. Then, adaptive blocking is performed based on the optimal blocking factor, which can make the data block division more balanced and reasonable, thereby avoiding waste of storage resources or reduction of access efficiency caused by uneven data distribution. For the blocked data block set, the access frequency is calculated based on the historical access times within the preset time window and the time decay factor, which can dynamically reflect the actual use of the data block. Thus, accurate basis is provided for subsequent storage optimization and access optimization. The storage node weight is calculated based on the access frequency and the basic weight value of the storage node, and the virtual node is expanded, which can improve the balance of the consistent hash distribution, make the data block distribution more reasonable among the storage nodes, avoid data skew problem, improve the load balancing capability of the storage system, realize data storage balancing, access efficiency improvement and system overall performance optimization, and provide reliable guarantee for efficient storage and fast access of large-scale data.

[0131] In an alternative embodiment,

[0132] The data block is segmented and encoded to obtain a data block feature table, and the data access verification sequence is generated based on the data block feature table and the data access index, which includes:

[0133] The historical access record of the data block is obtained, and the access mode time sequence is generated based on the historical access record;

[0134] The content entropy, boundary sensitivity and data density of the content in the data block are calculated to construct a data block content feature vector, the optimal segmentation size is calculated based on the information entropy of the data block content feature vector and the variance of the access mode time sequence, and the data block is segmented based on the optimal segmentation size to obtain a segmentation sequence;

[0135] The historical error record of the data block is acquired to calculate the data block error rate, the number of information symbols is calculated based on the access frequency of the data block in the segmented sequence, the number of check symbols is calculated according to the data block error rate and the number of information symbols, the segmented sequence is encoded to obtain a coded block set, the preset minimum segmentation size of the data block is acquired, the tree structure depth is calculated according to the data block size and the tree structure is constructed, the node number of the coded block in the coded block set is allocated to generate a parent node pointer, and the hierarchical hash value of the coded block set is calculated, the hierarchical hash value, the tree structure depth information and the parent node pointer are combined to form a data block feature table;

[0136] The access frequency of the data block in the data access index is acquired to calculate a cache threshold, the data nodes with the access frequency greater than the cache threshold are cached and pre-fetched to establish a verification path, the current time of the system is acquired as a time stamp, the read-write state, the integrity state and the access permission state of the data block are acquired to form node state information, the hierarchical hash value, the time stamp and the node state information are signed to obtain a verification mark, and the data access verification sequence is generated based on the verification mark, the verification path and the data block feature table.

[0137] The historical access record of the data block is acquired, the historical access record includes access time stamp, access type (read / write), access user identification and other information, and is usually stored in a system log. An access pattern time sequence is generated based on the historical access record, and the number of accesses in each time unit needs to be counted to form time series data. The time unit can be set according to the application scenario, such as minutes, hours or days. For example, the access number sequence of a data block in 24 consecutive hours may be [5, 3, 0, 1, 7, 12, 15, 20, 18, 10, 8, 6, 4, 2, 1, 0, 2, 5, 8, 10, 15, 12, 8, 6], which reflects the time distribution characteristics of data access, such as the peak access during working hours and the low access at night.

[0138] The content entropy, boundary sensitivity and data density of the content in the data block are calculated to construct a data block content feature vector. The content entropy reflects the randomness and information amount of the data, and the calculation method is to calculate the product of the probability of each symbol in the data and the logarithm of the probability, and then take the negative value. The boundary sensitivity measures the continuity of the data at the segmentation boundary, and the calculation method is to analyze the difference between adjacent data units. The data density represents the amount of effective information contained in a unit of storage space, and the calculation method is to divide the content entropy by the data block size. For example, for a data block containing mixed text and image content, the content feature vector may be [0.85, 0.32, 0.67], indicating that the content entropy is 0.85, the boundary sensitivity is 0.32, and the data density is 0.67.

[0139] The optimal segment size is calculated based on the information entropy of the data block content feature vector and the variance of the access pattern time series. Data with high information entropy contains more information and requires more fine-grained segmentation. Data with large access pattern variance has large frequency fluctuations and is suitable for variable-length segmentation strategy. The optimal segment size is calculated by multiplying the base segment size (e.g., 4KB) by a weighted function of information entropy and access variance. The weighted function can be designed as the weighted sum of the logarithm of information entropy and the square root of access variance, with weights adjusted according to system performance requirements. For example, if the information entropy of a data block is 0.85, the access variance is 25, the base segment size is 4KB, and the weights are 0.6 and 0.4 respectively, then the optimal segment size is approximately 4KB x (0.6 x (-0.16) + 0.4 x 5) = 7.78KB.

[0140] The data block is segmented according to the optimal segment size to obtain a segmented sequence. The semantic boundaries of the data content need to be considered during segmentation. For text data, segmentation can be performed at paragraph or sentence boundaries. For image data, segmentation can be performed at image boundaries. For structured data, segmentation can be performed at record boundaries. The segmented sequence contains multiple data segments with sizes close to the optimal segment size, and each segment is attached with position index information. For example, a 10MB data block with an optimal segment size of 8KB may result in approximately 1280 segments, each with a size between 7.5KB and 8.5KB.

[0141] The data block error rate is calculated based on the historical error records of the data block. Historical error records include data read / write errors, check failures, node failures, etc. The data block error rate is calculated by dividing the number of errors by the total number of accesses. For example, if a data block has a total access count of 10000 and records 15 errors, the error rate is 0.0015. The number of information symbols is calculated based on the access frequency of the data block in the segmented sequence. The number of information symbols represents the actual number of information symbols in the data, which is related to the data size and encoding method. For binary data, the number of information symbols can be simplified to the number of bits; for text data, it can be the number of characters.

[0142] The number of check symbols is calculated based on the data block error rate and the number of information symbols. Check symbols are used for error detection and correction, and the number of check symbols is related to the error rate and the number of information symbols. The calculation method is to multiply the number of information symbols by a function of the error rate, which can be designed as the product of the logarithm of the error rate and the adjustment factor. For example, if the number of information symbols is 65536 (8KB data), the error rate is 0.0015, and the adjustment factor is -2, then the number of check symbols is approximately 65536 x (-2) x (-6.5) ≈ 852 symbols, accounting for about 1.3% of the total data. The encoded block set is obtained by encoding the segmented sequence. The encoding process uses forward error correction encoding technology to combine information symbols and check symbols into encoded blocks.

[0143] A preset minimum segment size of the data block is obtained, a tree structure depth is calculated according to the data block size, and the tree structure is constructed. The calculation method of the tree structure depth is to take the logarithm of the data block size divided by the minimum segment size, and then round up. For example, if the data block size is 10 MB and the minimum segment size is 1 KB, the tree structure depth is ceil(log(10*1024 / 1))=14. The constructed tree structure is a balanced tree, each leaf node corresponds to an encoding block, and a non-leaf node contains the hash value of the child node. The encoding blocks in the encoding block set are assigned node numbers to generate parent node pointers. The node numbers can be assigned in the order from left to right and from bottom to top, and the parent node pointers point to the parent node number of the current node.

[0144] The hierarchical hash value of the encoding block set is calculated. The hierarchical hash starts from the leaf node and calculates layer by layer upwards. The hash value of the leaf node is the hash value of the corresponding encoding block content; the hash value of the non-leaf node is the hash value of the connection of all child node hash values. The hash algorithm uses a secure hash function. The hierarchical hash structure makes any change to an encoding block result in a change in the root node hash value, facilitating data integrity verification. The hierarchical hash value, tree structure depth information and parent node pointer form a data block feature table, which is the unique identifier of the data block on the blockchain and is used for data integrity verification and efficient retrieval.

[0145] The access frequency of the data block in the data access index is obtained to calculate the cache threshold. The cache threshold is used to determine which data should be cached. The calculation method is to multiply the average value of the access frequency of all data blocks by a coefficient factor. The coefficient factor can be dynamically adjusted according to the system resource status, and is usually between 1.2 and 2.0. For example, if the average access frequency of all data blocks in the system is 10 times / hour, and the coefficient factor is 1.5, then the cache threshold is 15 times / hour. Data nodes with access frequency greater than the cache threshold are cached and pre-fetched to establish a verification path. The verification path is a sequence of hash values from the leaf node to the root node.

[0146] The current time of the system is obtained as a timestamp. The timestamp format can be Unix timestamp, accurate to milliseconds. The read-write state, integrity state and access permission state of the data block are obtained to form node state information. The read-write state indicates whether the data block is readable and writable; the integrity state indicates the data integrity check result; and the access permission state indicates the list of users or roles allowed to access. The hierarchical hash value, timestamp and node state information are signed to obtain a verification mark. The signature process uses asymmetric encryption algorithms such as RSA or ECDSA to ensure the non-falsifiability of the verification mark. Based on the verification mark, the verification path and the data block feature table, a data access verification sequence is generated. The verification sequence contains all the necessary information, enabling the client to verify the integrity, timeliness and permission legality of the data.

[0147] In this embodiment, by obtaining the historical access records of the data block and generating the access pattern time sequence, the access rules of the data block in different time periods can be extracted, thereby providing dynamic reference for subsequent segmentation and cache optimization. The content entropy, boundary sensitivity and data density of the data block content are calculated, the data block content feature vector is constructed, and the optimal segmentation size is calculated based on the information entropy of the content feature vector and the variance of the access pattern time sequence. The data can be adaptively segmented according to the complexity and access volatility of the data itself, so that the data division is more reasonable, which helps to improve the data access efficiency and subsequent coding performance. The error rate is calculated by obtaining the historical error records of the data block, and the information symbol number is calculated in combination with the access frequency of the data block in the segmentation sequence. Then, the number of check symbols is determined according to the error rate and the information symbol number, and the segmentation sequence is encoded. Elastic check and coding can be realized under different access intensity and error rate conditions, which improves the reliability and error correction capability of data storage. The cache threshold is calculated based on the data access index, and the data nodes with access frequency exceeding the threshold are pre-fetched for caching. The access efficiency of high-frequency data can be effectively improved, and the access delay can be reduced. The data division rationality, storage reliability, access efficiency and security and reliability are synergistically improved, which provides efficient and reliable technical support for large-scale data storage and access.

[0148] Figure 2 The data block verification and processing flowchart of the unstructured data storage optimization method based on information entropy and block chain of the embodiment of the application.

[0149] In an optional implementation,

[0150] According to the data access verification sequence, data integrity verification is performed, a proof containing data identification and signature is generated, and the verification result is recorded to the block chain, which includes:

[0151] Obtain the data access verification sequence, recursively verify the hierarchical hash value based on the verification path in the data access verification sequence, and generate a verification result;

[0152] According to the verification result, a data identification is generated, and an integrity signature is obtained by signing the data identification, and the data identification and the integrity signature are combined to form a verification proof;

[0153] The verification proof is submitted to the block chain network and recorded.

[0154] The data access verification sequence includes key information such as a verification mark, a verification path, and a data block feature table. The verification mark is generated by signing the hierarchical hash value, timestamp, and node state information, and is used to prove the authenticity and integrity of the data. The verification path is a sequence of hash values from the leaf node to the root node, and is used to verify whether a specific data block belongs to the original data set. The data block feature table includes the hierarchical hash value, tree structure depth information, and parent node pointer, and is used to locate and identify the data block. In actual application, the verification sequence can be obtained from the data provider through a secure channel, or obtained by querying the blockchain. For example, a certain data access verification sequence may include a verification mark "0x7a3bc4d...", a verification path ["0x1a2b3c...", "0x4d5e6f...", "0x7g8h9i..."], and a data block feature table {hash value: "0x9j8k7l...", depth: 5, parent node: [123, 456, 789]}.

[0155] The hierarchical hash value is recursively verified based on the verification path in the data access verification sequence to generate a verification result. The recursive verification process starts from the leaf node and verifies layer by layer upwards to the root node. The specific verification steps are as follows: take the first hash value in the verification path and compare it with the hash value of the current data block to confirm whether they are consistent; if they are consistent, combine the hash value with the next hash value in the verification path to calculate a new hash value; repeat the above process until the root node hash value is verified. If the final calculated root node hash value is consistent with the root hash value provided in the verification sequence, the verification is passed; otherwise, the verification fails. The hash algorithm used in the verification process should be consistent with the algorithm used to generate the hierarchical hash value, such as SHA-256. For example, assuming that the leaf node hash value is "0x1a2b3c...", and the verification path is ["0x4d5e6f...", "0x7g8h9i..."], the verification process is: calculate Hash("0x1a2b3c..."+"0x4d5e6f...") to get the intermediate hash value "0xabcd12..."; then calculate Hash("0xabcd12..."+"0x7g8h9i...") to get the root hash value "0x9876fe..."; finally, compare "0x9876fe..." with the expected root hash value to determine whether the verification is passed.

[0156] The verification result contains information such as verification status (pass / fail), verification time, and verification node identification. The verification status directly reflects the integrity and authenticity of the data, the verification time records the execution time of the verification operation, and the verification node identification indicates the blockchain node that performs the verification. The verification result can be represented in JSON format for easy storage and transmission, such as {"status": "verified", "timestamp": 1629456789, "node_id": "node012345", "details": {"hash_match": true, "time_valid": true}}.

[0157] According to the verification result, a data identifier is generated, which is a unique identifier of the data on the blockchain, used for subsequent reference and retrieval. The data identifier generation method is to connect the key fields of the verification result (such as verification status, timestamp, and data hash value) and calculate the hash value. The hash algorithm can use a secure hash function such as SHA-256 to ensure the uniqueness and unforgeability of the identifier. For example, if the verification result contains the verification status "verified", the timestamp 1629456789, and the data hash value "0x9876fe...", these values can be connected to calculate the hash value, and the data identifier "0xa1b2c3d4..." is obtained.

[0158] The data identifier is signed to obtain an integrity signature, and the signature process uses a private key to encrypt the data identifier to generate an unforgeable digital signature. The signature algorithm can use asymmetric encryption algorithms such as RSA and ECDSA to ensure the security and verifiability of the signature. The signature result is a fixed-length string, such as "0xe5f6g7h8...". The signature contains information such as signature algorithm identification, signature value, and signature time to ensure the integrity and traceability of the signature. The integrity signature can be represented as {"algorithm": "ECDSA", "value": "0xe5f6g7h8...", "timestamp": 1629456790}.

[0159] The data identifier and the integrity signature are combined into a verification proof for proving the credibility of the data to a third party, and the verification proof structure includes fields such as the data identifier, the integrity signature, the proof generation time, and the proof version number, to form a complete proof document. The verification proof can be serialized in JSON or XML format for easy storage and transmission. For example, the verification proof can be represented as {"data_id": "0xa1b2c3d4...", "signature": {"algorithm": "ECDSA", "value": "0xe5f6g7h8...", "timestamp": 1629456790}, "proof_time": 1629456795, "version": "1.0"}.

[0160] The verification proof is submitted to the blockchain network and recorded, and the submission process includes constructing a blockchain transaction, using the sender's private key to sign the transaction, broadcasting the signed transaction to the blockchain network, and waiting for the transaction to be packaged into a block and confirmed. After the blockchain network receives the transaction, it verifies the validity of the transaction signature and adds the valid transaction to the candidate block. Through a consensus mechanism (such as proof of work, proof of stake, etc.), the candidate block is confirmed as a new block and added to the blockchain. Once the transaction is confirmed, the verification proof is permanently recorded on the blockchain, and anyone can query and verify it.

[0161] In this embodiment, by obtaining the data access verification sequence and recursively verifying the hierarchical hash value based on the verification path, the integrity of the data in the storage and access process can be checked level by level, thereby effectively discovering and preventing data tampering or loss. According to the verification result, a data identifier is generated and signed to obtain an integrity signature, which can ensure the source credibility and content authenticity of the data in the transmission and storage process, avoid illegal forgery, combine the data identifier and the integrity signature into a verification proof, and submit it to the blockchain network for recording. The non-tamperability and traceability of the blockchain can be used to realize decentralized storage of data access and integrity verification, thereby greatly improving the data security and credibility of the system, realizing the automation and tamper-proofing of data integrity verification, and ensuring the reliability and traceability of the data in a distributed environment.

[0162] In a second aspect, the embodiment of the present application provides an unstructured data storage optimization system based on information entropy and a blockchain, which comprises:

[0163] The first unit is configured to establish a blockchain storage network and obtain unstructured data, construct a covariance matrix corresponding to the unstructured data, and perform eigenvalue decomposition to obtain reduced dimension data.

[0164] The second unit is configured to map the dimension-reduced data to a high-dimensional feature space by a preset kernel function, construct an optimal classification hyperplane, calculate a distance vector of each dimension-reduced data to the optimal classification hyperplane to obtain an initial redundant data set, generate a binary coding sequence based on data features in the initial redundant data set, perform selection, crossover and mutation operations on the binary coding sequence to obtain an optimized feature sequence, label redundant data according to the optimized feature sequence, calculate fuzzy similarity of different redundant data, and remove data with fuzzy similarity greater than a preset fuzzy similarity threshold to obtain a data set to be stored.

[0165] The third unit is configured to divide the data set to be stored into data block sets according to a preset size, calculate access frequencies of data blocks in the data block sets, perform a consistent hash algorithm on the data block sets according to the access frequencies to obtain a data block storage location table, establish a data access index based on the data block storage location table, segmentally encode each data block and calculate a hash value to obtain a data block feature table, and generate a data access verification sequence according to the data block feature table and the data access index.

[0166] The fourth unit is configured to perform data integrity verification according to the data access verification sequence, generate a proof containing data identification and a signature, and record a verification result to a block chain.

[0167] In a third aspect, an electronic device is provided, including:

[0168] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0169] In a fourth aspect, a computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0170] The present application can be a method, device, system and / or computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions stored therein, which are used to perform various aspects of the present application.

[0171] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for unstructured data storage optimization based on information entropy and blockchain, characterized in that, The method comprises the following steps: building a blockchain storage network and obtaining unstructured data, constructing a covariance matrix corresponding to the unstructured data and performing eigenvalue decomposition to obtain reduced dimension data; mapping the reduced dimension data to a high-dimensional feature space through a preset kernel function, constructing an optimal classification hyperplane, calculating the distance vector of each reduced dimension data to the optimal classification hyperplane to obtain an initial redundant data set, generating a binary code sequence based on the data features in the initial redundant data set, performing selection, crossover and mutation operations on the binary code sequence to obtain an optimized feature sequence, marking the redundant data according to the optimized feature sequence, calculating the fuzzy similarity of different redundant data, and removing data with fuzzy similarity greater than a preset fuzzy similarity threshold to obtain a data set to be stored; dividing the data set to be stored into blocks according to a preset size to obtain a data block set, calculating the access frequency of each data block in the data block set, and performing a consistent hash algorithm on the data block set based on the access frequency to obtain a data block storage location table, establishing a data access index based on the data block storage location table, segmenting and encoding each data block to obtain a data block feature table, and generating a data access verification sequence according to the data block feature table and the data access index; performing data integrity verification according to the data access verification sequence, generating a proof containing data identification and signature, and recording the verification result to the blockchain.

2. The method of claim 1, wherein, The method comprises the following steps: building a blockchain storage network and obtaining unstructured data, constructing a covariance matrix corresponding to the unstructured data and performing eigenvalue decomposition to obtain reduced dimension data includes: obtaining unstructured data, and building a blockchain storage network composed of multiple nodes; extracting frequency features and time sequence features from the unstructured data, calculating a feature vector based on the frequency features and the time sequence features, and constructing a covariance matrix based on the feature vector; 3. The method of claim 1, wherein, performing eigenvalue decomposition on the covariance matrix to obtain initial eigenvalues and reconstructing the unstructured data based on the initial eigenvalues to obtain reduced dimension data, and storing the reduced dimension data to the blockchain storage network. mapping the reduced dimension data to a high-dimensional feature space through a preset kernel function, constructing an optimal classification hyperplane, calculating the distance vector of each reduced dimension data to the optimal classification hyperplane to obtain an initial redundant data set, generating a binary code sequence based on the data features in the initial redundant data set, performing selection, crossover and mutation operations on the binary code sequence to obtain an optimized feature sequence includes: obtaining reduced dimension data, calculating the Euclidean distance between different data points in the reduced dimension data, constructing a Gaussian kernel function based on the Euclidean distance and a pre-set kernel function bandwidth parameter, and mapping the reduced dimension data to a high-dimensional feature space using the Gaussian kernel function to obtain high-dimensional feature data; constructing an optimization objective function based on the high-dimensional feature data, wherein the optimization objective function includes a normal vector length minimization term and a constraint relaxation term weighted based on a pre-set penalty factor, and solving the optimization objective function to obtain the normal vector and bias term of the optimal classification hyperplane; Based on the normal vector and the bias term, a distance value of each data point in the reduced dimension data to the optimal classification hyperplane is calculated, the distance values are sorted from small to large, data points located in the top 10% are selected to construct an initial redundant data set, a feature vector corresponding to each data point is extracted from the initial redundant data set, the mean value of each dimension feature component of the feature vector is calculated, the feature component is compared with the corresponding dimension mean value, the feature component greater than the mean value is assigned a binary value of 1, and the feature component less than the mean value is assigned a binary value of 0, and a binary encoding sequence is generated; The fitness value of each sequence in the binary encoding sequence is calculated, and a roulette wheel selection operation is performed based on the fitness value to construct a high-quality sequence, a single-point crossover operation is performed on the high-quality sequence to generate a crossover sequence, and a random inversion operation is performed on the bit position of the crossover sequence to obtain an optimized feature sequence.

4. The method of claim 1, wherein, According to the optimized feature sequence, the redundant data is labeled, the fuzzy similarity of different redundant data is calculated, the data with a fuzzy similarity greater than a preset fuzzy similarity threshold is removed, and a to-be-stored data set is obtained, including: An optimized feature sequence is obtained, and each feature dimension in the optimized feature sequence is mapped to a fuzzy membership value. A triangular membership function is constructed based on the fuzzy membership value. The parameters of the triangular membership function are dynamically adjusted based on the data distribution characteristics corresponding to the unstructured data, and an adaptive fuzzy feature is obtained. A multi-dimensional fuzzy rule base is constructed based on the adaptive fuzzy feature, an initial weight coefficient is assigned to each rule in the multi-dimensional fuzzy rule base, the fuzzy similarity between different data is calculated according to the multi-dimensional fuzzy rule base, and the data with a fuzzy similarity greater than a preset fuzzy similarity threshold is labeled as preliminary screening redundant data. Based on the distance between data points and the cutoff distance parameter, the local density value corresponding to the preliminary screening redundant data is calculated, and the preliminary screening redundant data is finely screened according to the local density value. The data with a local density value greater than a preset density threshold is labeled as candidate redundant data. The integrity score of the candidate redundant data is calculated, the candidate redundant data with an integrity score lower than a pre-set integrity threshold is removed, and a to-be-stored data set is constructed based on the retained candidate redundant data.

5. The method of claim 1, wherein, The to-be-stored data set is divided into data block sets according to a preset size, the access frequency of each data block in the data block set is calculated, a consistent hash algorithm is executed to allocate the data block set to obtain a data block storage location table, and a data access index is established based on the data block storage location table, including: The to-be-stored data set is preliminarily divided into data block sets according to a preset size, the optimal block factor including the total size of the data set and the data density variance is calculated, and the to-be-stored data set is adaptively divided into data block sets according to the optimal block factor. For each data block in the data block set, the access frequency of each data block in the data block set is calculated based on the historical access frequency in a preset time window and a pre-set time decay factor. The storage node weight value is calculated based on the access frequency and the base weight value corresponding to each storage node, the storage node weight value is used for virtual node expansion of the storage node, a virtual node distribution sequence is obtained, a data block hash value is calculated for each data block in the data block set, a consistent hash algorithm is executed to map the weighted sum of the data block hash value and the access frequency to a hash space corresponding to the virtual node distribution sequence, and a data block storage location table is obtained; A primary index table recording the mapping relationship between the data block and the physical node is established based on the data block storage location table, an access frequency threshold is set based on the access frequency, a secondary index table of hot data access is established, a physical distance and a logical distance are calculated, a routing weight is constructed, and a routing table is generated, and the primary index table, the secondary index table and the routing table are combined to form a data access index.

6. The method of claim 1, wherein, Segment coding is performed on each data block, and a hash value is calculated to obtain a data block feature table, and a data access verification sequence is generated according to the data block feature table and the data access index, including: Obtaining the historical access record of the data block, generating an access mode time sequence based on the historical access record; The content entropy, boundary sensitivity and data density of the content in the data block are calculated to construct a data block content feature vector, the information entropy of the data block content feature vector and the variance of the access mode time sequence are calculated to obtain an optimal segmentation size, and the data block is segmented according to the optimal segmentation size to obtain a segmentation sequence; The historical error record of the data block is obtained to calculate the error rate of the data block, the number of information symbols is calculated based on the access frequency of the data block in the segmentation sequence, the number of check symbols is calculated according to the error rate of the data block and the number of information symbols, the segmentation sequence is encoded to obtain an encoded block set, the preset minimum segmentation size of the data block is obtained, the tree structure depth is calculated according to the data block size, and the tree structure is constructed, the encoded blocks in the encoded block set are assigned node numbers to generate parent node pointers, and the hierarchical hash value, the tree structure depth information and the parent node pointer are combined to form the data block feature table; The access frequency of the data block in the data access index is obtained to calculate the cache threshold, the data node with the access frequency greater than the cache threshold is pre-fetched to establish a verification path, the current time of the system is obtained as a time stamp, the read-write state, the integrity state and the access permission state of the data block are obtained to form node state information, the hierarchical hash value, the time stamp and the node state information are signed to obtain a verification mark, and the data access verification sequence is generated based on the verification mark, the verification path and the data block feature table.

7. The method of claim 1, wherein, Data integrity verification is performed according to the data access verification sequence, a proof containing a data identifier and a signature is generated, and the verification result is recorded to a block chain, including: Obtaining the data access verification sequence, recursively verifying the hierarchical hash value based on the verification path in the data access verification sequence, and generating a verification result; Generating a data identifier according to the verification result, signing the data identifier to obtain an integrity signature, and combining the data identifier and the integrity signature to form a verification proof; The verification proof is submitted to a block chain network and recorded.

8. An information entropy and blockchain based unstructured data storage optimization system for implementing the method of any of the preceding claims 1-7, characterized in that, including: The first unit is configured to establish a blockchain storage network and acquire unstructured data, construct a covariance matrix corresponding to the unstructured data, perform eigenvalue decomposition, and obtain reduced dimension data. The second unit is configured to map the reduced dimension data to a high-dimensional feature space by using a preset kernel function, construct an optimal classification hyperplane, calculate a distance vector of each reduced dimension data to the optimal classification hyperplane to obtain an initial redundant data set, generate a binary coding sequence based on data features in the initial redundant data set, perform selection, crossover, and mutation operations on the binary coding sequence to obtain an optimized feature sequence, mark redundant data according to the optimized feature sequence, calculate fuzzy similarity of different redundant data, remove data with fuzzy similarity greater than a preset fuzzy similarity threshold, and obtain a to-be-stored data set. The third unit is configured to block the to-be-stored data set according to a preset size to obtain a data block set, calculate access frequencies of data blocks in the data block set, perform a consistent hash algorithm on the data block set based on the access frequencies to obtain a data block storage location table, establish a data access index based on the data block storage location table, segmentally encode each data block and calculate a hash value to obtain a data block feature table, and generate a data access verification sequence based on the data block feature table and the data access index. The fourth unit is configured to perform data integrity verification based on the data access verification sequence, generate a proof containing data identification and a signature, and record a verification result to a blockchain.

9. An electronic device, comprising: It comprises: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the method of any one of claims 1 to 7.

Images